Pan-tumor analysis to investigate the obesity paradox in immune checkpoint blockade

Abstract

Background

Obesity is a risk factor for developing cancer but is also associated with improved outcomes after treatment with immune checkpoint inhibitors (ICIs), a phenomenon called the obesity paradox. To interrogate mechanisms of divergent immune responses in obese and non-obese patients, we examined the relationship among obesity status, clinical responses, and immune profiles from a diverse, pan-tumor cohort of patients treated with ICI-based therapy.

Methods

From June 2021 to March 2023, we prospectively collected serial peripheral blood samples from patients with advanced or metastatic solid tumors who received ICI as standard of care at Johns Hopkins. Patients were stratified by obesity status at treatment initiation, with obesity defined as body mass index (BMI)≥30 at treatment initiation and BMI≥18.5 and <30 considered non-obese; underweight patients (BMI<18.5) were excluded. We evaluated the concentration of 37 cytokines and used cytometry by time of flight to characterize immune cell clusters and cell-surface expression markers at baseline and on-treatment.

Results

We enrolled 94 patients, of whom 30 (32%) were obese and 64 (68%) were non-obese. Compared with non-obese patients, obese patients had superior progression-free survival (HR: 0.44 (95% CI: 0.24 to 0.81), p=0.01) and overall survival (OS) (HR: 0.24 (95% CI: 0.07 to 0.80), p=0.02). Obese patients had lower serum IL-15 levels at treatment baseline and lower on-treatment levels of IL-6, IL-8, and IL-15. Low on-treatment IL-6 was associated with improved OS (HR: 0.27 (95% CI: 0.08 to 0.88), p=0.03), as was low on-treatment IL-8 (HR: 0.19 (95% CI: 0.05 to 0.70), p=0.01). Obese patients demonstrated lower levels of T effector cells with reduced expression of cytotoxicity markers and higher expression of exhaustion markers at baseline and on-treatment.

Conclusions

Obese and non-obese patients with cancer have divergent immunological responses to ICIs. Obesity is associated with reduced levels of certain inhibitory cytokines and higher expression of T-cell exhaustion markers. ICI-based therapy may more effectively reverse T-cell dysfunction in obese patients, potentially contributing to the paradoxically improved responses in this population.

Correspondence to Dr Mark Yarchoan; [email protected]

WHAT IS ALREADY KNOWN ON THIS TOPIC

• While obesity predisposes patients to developing cancer, it is also associated with improved outcomes after treatment with immune checkpoint blockade. Both obesity and malignancy create chronic inflammation, leading to an immunosuppressive environment characterized by T-cell exhaustion.

WHAT THIS STUDY ADDS

• This pan-tumor study describes the systemic changes in the immunologic states observed in obese as compared with non-obese patients at baseline and after treatment with immune checkpoint inhibitors.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The cytokines and cell-surface markers identified in this study should be further evaluated in the context of response to and potential targets of immune checkpoint blockade in both obese and non-obese patients.

Background

The incidence of obesity continues to rise, with the WHO estimating that approximately 16% of adults are obese, two times the prevalence in 1990.¹ In the USA, 42% of adults are obese.² Obesity is associated with an increased risk of developing multiple types of cancer, yet it has also been associated with improved outcomes after cancer therapy, a phenomenon known as the obesity paradox.^3–9

For example, obese and overweight patients with non-small cell lung cancer (NSCLC) experienced significant improvements in overall survival (OS) compared with patients with normal or overweight body mass index (BMI) when treated with the immune checkpoint inhibitor (ICI) atezolizumab, with obese BMI most favorable.¹⁰ This benefit was not observed in patients receiving chemotherapy alone, where there was no difference seen among the three BMI categories.¹⁰ Similarly, when comparing obese patients with those with normal BMI in a retrospective study of patients with metastatic melanoma receiving targeted therapy, ICIs, or chemotherapy, a survival benefit was only seen in patients treated with ICIs and targeted therapy, not in those treated with chemotherapy.¹¹ Larger studies involving patients with NSCLC, melanoma, and renal cell carcinoma (RCC) similarly showed improved outcomes after treatment with ICIs in overweight and obese patients, including progression-free survival (PFS), OS, and time to treatment failure.¹² These clinical data provide initial evidence that the obesity paradox may be related to immunological features of obesity.

Several studies have demonstrated that obesity induces chronic inflammation.^{4 9 13 14} Studies in human and preclinical models of obesity have demonstrated impaired CD8⁺ T-cell reactivity in the setting of obesity, paralleling increased expression of inhibitory cell-surface markers suggestive of T-cell dysfunction or exhaustion.^15–17 However, the association between obesity and chronic inflammation warrants confirmation in additional clinical cohorts, and the impact of this chronic immune state on immune responses to ICIs requires further investigation.

The effect of elevated BMI on immune cell populations has implications for treatment of the continually increasing population of obese patients throughout the world and in understanding drivers of ICI sensitivity and resistance across unselected patient populations. Here, we evaluate clinical outcomes, serum cytokine concentrations, immune cell clusters, and cell-surface markers in obesity using a diverse, pan-tumor cohort treated with ICIs. We demonstrate that obese patients have reduced levels of inhibitory cytokines after treatment with ICIs, as well as increased expression of cell-surface markers involved in inhibitory T-cell checkpoint expression that may be more prone to reversal with ICIs.

Methods

Patient selection and BMI categories

This single-institution study was conducted at the Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University. All patients were 18 years of age or older with pathologically confirmed or clinically suspected advanced or metastatic neoplasm. Patients receiving neoadjuvant or adjuvant therapy for potentially resectable cancers, who classically have a lower tumor burden, were excluded. All patients received at least one dose of ICI, either alone or in combination with another treatment modality; however, this did not necessarily represent the first line of treatment or first ICI exposure. While enrollment was focused on patients receiving ICIs as standard of care, patients receiving treatment as part of a clinical trial were also eligible.

BMI was calculated using height and weight measured at the time of therapy initiation and categorized based on the WHO classification.¹⁸ We considered BMI≥30 kg/m² obese and BMI≥18.5 kg/m² and<30 kg/m², including both normal (≥18.5 kg/m² and <25 kg/m²) and overweight (≥25 kg/m² and <30 kg/m²) BMI, non-obese, with patients categorized for further analysis based on their BMI at therapy initiation. In our database, there were 10 patients with BMI<18.5, considered underweight by the WHO, and, given the small number of patients and potential for confounding due to cachexia and/or poor performance status, these patients were excluded.

Sample collection and cytokine measurement

Blood samples were collected via venipuncture in heparinized syringes and stored immediately or processed, and then stored immediately within 2 hours of collection. The Bioplex 200 platform (Bio-Rad, Hercules, California, USA) was used to determine the concentration of 37 different target proteins in serum/plasma. Luminex bead-based immunoassays (Millipore, Billerica, Massachusetts, USA) were performed following Immune Monitoring Core standard operating procedures (SOPs), and concentrations were determined using five parameter log curve fits using Bioplex Manager V.6.0, with vendor-provided standards and quality controls. The HCYTA-60K panel was used to detect sCD40L, IL-1α, IL-1β, IL-1RA, IL-2, IL-3, IL-4, IL-5, IL-6, IL-8, IL-9, IL-10, IL-12(p40), IL-12(p70), IL-13, IL-15, IL-17A, IL-17E, IL-17F, IL-18, IL-22, G-CSF, GM-CSF, TNFα, IFNγ, MCP-1, MIP-1α, MIP-1β, RANTES, MIG, IP-10, VEGF-α, MIPF-1, IL-35, BAFF, ITAC, and MIP-3α. All concentrations outside of standard curve values were categorized as out of range, with values below the standard curve replaced with the lower limit of the standard curve of the assay and values above the standard curve replaced with the upper limit of the standard curve of the assay.

Samples were collected at baseline (up to 2 weeks before ICI initiation), 1 month (2–4 weeks after ICI initiation), 2 months (6–10 weeks after ICI initiation), 4 months (14–18 weeks after ICI initiation), 6 months (20–28 weeks after ICI initiation), and 1 year (50–70 weeks after ICI initiation). Patients were also asked to provide samples in the event of severe immune related adverse event (irAE). For this study, we evaluated the baseline measurement and the first on-treatment specimen available to assess pretreatment and on-treatment cytokine levels, immune cell clusters, and cell-surface markers. In evaluating fold change after treatment, only on-treatment samples that were run alongside their corresponding baseline sample were used. For all patients with baseline samples from multiple runs, the average concentration was used for analysis of baseline cytokine concentration.

Cytometry by time of flight (CyTOF)

Antibodies used for CyTOF staining and analysis are shown in online supplemental table S1. All custom antibodies were conjugated as described previously.¹⁹ Patient peripheral blood mononuclear cells underwent thawing in a 37°C water bath and were then slowly recovered using prewarmed Roswell Park Memorial Institute Medium (RPMI) supplemented with 10% fetal bovine serum (FBS). Following thaw, samples were counted, and 2×10⁶ cells from each sample were plated in 96-well plates. Plated cells rested in media for 30 minutes before undergoing a single wash in phosphate buffered saline (PBS) containing 2 mM EDTA. Afterward, cells were subjected to a 2.5 minute incubation at room temperature in a 20 µM Pt solution (Standard BioTools) in PBS to indicate viability. Following incubation, RPMI supplemented with 10% FBS was introduced to the cells to neutralize the platinum. Next, cells were washed two times using cell staining buffer (CSB) (Standard BioTools). Following washes, all samples were barcoded by incubating cells with unique combinations of metal conjugated anti-CD45 antibodies for 20 minutes. Samples were then washed two times with CSB, multiplexed, and transferred to v-bottom flow tubes using a 40 um filter. Each tube was incubated with anti-human Fc receptor (FcR) block for 10 min at room temperature (12 µL used for 15×10⁶ cells) to block non-specific FcR binding, followed by an incubation with a chemokine stain cocktail for 10 minutes in a 37°C water bath. Tubes were removed from the water bath and a surface stain cocktail was added for 30 minutes at room temperature. Samples were next washed two times with CSB followed by fixation and permeabilization using Cytofix/Cytoperm solution (BD Biosciences) for 30 minutes at room temperature. Fixed and permeabilized samples were washed with perm/wash solution (BD Biosciences) and stained using the intracellular cocktail for 30 minutes at room temperature. Samples were washed two times with perm/wash solution and stored in 1.6% paraformaldehyde in PBS at 4°C until the day of acquisition (within 1 week). On the day of acquisition, samples were stained with 1:500 ¹⁰³Rh in Maxpar Fix/Perm solution (Standard BioTools) for 30 minutes at room temperature for cell identification. Samples underwent two PBS washes following this incubation and were finally washed once more and resuspended in normalization beads (Standard BioTools).

A Helios mass cytometer (Standard BioTools) at the Johns Hopkins University Mass Cytometry Facility was used to acquire the data. All acquired data were randomized and normalized using CyTOF software (V.7.1.16389.0, Standard BioTools). Resulting flow cytometry standard (fcs) files were debarcoded by manual gating using FlowJo software (V.10.9.0, BD Biosciences). Cell events were gated using ¹⁰³Rh positivity. Live cells were gated using ¹⁹⁴Pt and ¹⁹⁵Pt viability staining. Next, sample debarcoding was performed based on positivity of unique combinations of CD45 barcodes. Each debarcoded sample was exported as an individual fcs file and normalized in R (V.4.0.2) using the CytoNorm algorithm using a repeated sample included in each staining batch to normalize samples based on goal quantiles of mean marker expression. Clustering analysis was performed in R using FlowSOM to generate 35 metaclusters, annotated using expression profile of markers included within the panel, resulting in 26 final clusters (online supplemental table S2). Of the 26 final clusters, DNT_IV was removed due to missing values in our data set, leaving 25 clusters.

Outcomes assessment

Patients underwent cross-sectional imaging at the discretion of the treating oncologist, but generally at intervals of approximately 8–12 weeks. Radiographic response was assessed by the primary investigator team using Response Evaluation Criteria in Solid Tumors V.1.1. The clinical outcomes evaluated in our study included objective response rate (ORR), defined as the percentage of patients who experienced complete response (CR) or partial response (PR), PFS, calculated as the time from the first dose of ICI to disease progression or death from any cause, and OS, calculated as the time from the first dose of ICI to death from any cause. In 11 patients, clinical progression was determined by the treating physician based on signs or symptoms consistent with progressive disease (PD) and/or a concomitant rise in tumor markers.

Data analysis

All analyses were performed using R (V.2023.12.1+402, Indianapolis, Indiana, USA). Descriptive statistics were reported using median (range) for continuous variables and number (percentage) for categorical variables. Cytokine concentrations were log-transformed, and cell counts were determined by multiplying cell proportion from CyTOF analysis with total cell counts collected from Epic and then log-transforming this value. Wilcoxon rank-sum tests, also known as the Mann-Whitney U test, were used to compare median cytokine levels, immune cell cluster numbers, and cell-surface markers between obese and non-obese patients at baseline and on-treatment, as well as the fold change in concentration, defined as the ratio of on-treatment to baseline levels, between these two groups. For cytokines that differed significantly at baseline or on-treatment when comparing obese with non-obese patients, the median cytokine concentration at each timepoint was used to dichotomize “high” and “low” cytokine levels, as described previously.^20–23 Spearman’s rank-order correlation was used to analyze the relationship between two continuous variables. Differences among categorical variables were assessed using Fisher’s exact test. Given that this analysis was exploratory in nature and all comparisons were of potential interest, multiple comparisons adjustments were not used.

Median PFS and OS were calculated using Kaplan-Meier estimates, with the log-rank test used to assess differences in PFS and OS across subgroups of interest using the survival and survminer packages in R. The univariate and multivariable Cox proportional hazards models were conducted, with adjustments of multivariable analysis selected based on clinical consideration by controlling baseline age (as a linearly continuous variable), sex (male vs female), cancer type (gastrointestinal vs genitourinary vs other), race (Caucasian vs African American vs other), and obesity (obese vs non-obese) unless otherwise specified. Continuous variables were described using median, IQR, number, and percentage. P values<0.05 were considered statistically significant.

Results

Patient demographics

From June 2021 to March 2023, 94 patients met eligibility criteria for inclusion and were enrolled. Patient demographics are shown in table 1. Overall, 30 patients were obese (32%) and 64 (68%) were non-obese, with median BMI at the start of treatment 24.5 (range: 18.5–29.9) for the non-obese cohort and median BMI at the start of treatment 33.8 (range: 30.0–52.0) for the obese cohort. Tumor types were categorized into three groups: gastrointestinal (n=35, 37%; including colorectal (n=1), biliary tract (n=2), hepatocellular carcinoma (HCC) (n=31), and pancreatic (n=1)), genitourinary (n=31, 33%; including RCC (n=23) and bladder cancer (n=8)), and other (n=28, 30%; including adrenal (n=1), breast (n=1), cervical (n=1), endometrial (n=3), head and neck squamous cell carcinoma (n=7), NSCLC (n=2), melanoma (n=2), Merkel cell carcinoma (n=1), neuroendocrine (n=2), sarcoma (n=5), and squamous cell skin cancer (n=3)). With respect to treatments received, 37 patients (39%) received PD-(L)1 blockade alone, 23 patients (24%) received combined PD-(L)1/CTLA-4 blockade, 15 patients (16%) received PD-(L)1 blockade with a multikinase inhibitor, 14 patients (15%) received PD-(L)1 blockade with a VEGF inhibitor, four patients (4%) received PD-(L)1 blockade with chemotherapy, and one patient (1%) received PD-(L)1 blockade with chemotherapy and a VEGF inhibitor.

Table 1

*Clinical characteristics for overall cohort, comparing obese with non-obese patients

Characteristic	Overall (n=94)	Obese (n=30)	Non-obese (n=64)	P value
Age, median (range)	65.5 (20.0–87.0)	67.0 (53.0–87.0)	65.0 (20.0–80.0)	0.20
BMI, median (range)	26.6 (18.5–52.0)	33.8 (30.0–52.0)	24.5 (18.5–29.9)	<0.01
Sex, female, n (%)	32 (34%)	15 (50%)	17 (27%)	0.04
Race, n (%)
Caucasian	60 (64%)	22 (73%)	38 (59%)	0.46
African American	27 (29%)	7 (23%)	20 (31%)
Other	7 (7%)	1 (3%)	6 (9%)
Cancer type, n (%)
GI	35 (37%)	9 (30%)	26 (41%)	0.51
GU	31 (33%)	12 (40%)	19 (30%)
Other	28 (30%)	9 (30%)	19 (30%)
Autoimmune disease, n (%)	11 (12%)	4 (13%)	7 (11%)	0.74
irAE, n (%)	41 (44%)	11 (37%)	30 (47%)	0.38

*Select percentages may not add up to 100% due to rounding.

BMI, body mass index; GI, gastrointestinal; GU, genitourinary; irAE, immune related adverse event.

The median age was 66 (range: 20–87), with 32 women (34%) and 62 (66%) men. Most patients were Caucasian (n=60, 64%), with the remainder African American (n=27, 29%) or other (n=7, 7%). Eleven patients (12%) had a history of autoimmune disease, and 41 patients (44%) experienced irAE(s) while on study. When comparing obese with non-obese patients, there was a significant difference in sex (50% vs 27% female, p=0.04) but not in age, cancer type, autoimmune disease, or irAE presence (table 1).

Clinical outcomes based on obesity

The median PFS and OS for the overall cohort were 4.3 months (95% CI: 3.4 to 8.9) and not reached (NR) (95% CI: NR to NR), respectively, with 12-month PFS 34.3% (95% CI: 25.4% to 46.3%) and 12-month OS 68.8% (95% CI: 58.6% to 80.7%). The obese cohort exhibited improved PFS (log-rank p<0.01) and OS (log-rank p=0.01), compared with non-obese patients (figure 1A,B). In comparing obese and non-obese patients, the probability of 12-month PFS was 51.7% (95% CI: 35.5% to 75.1%) and 26.5% (95% CI: 17.1% to 41.0%) and 12-month OS was 88.5% (95% CI: 77.0% to 100.0%) and 58.0% (95% CI: 44.5% to 75.5%), respectively. Obesity remained a significant predictor in univariate analysis of PFS (HR: 0.44 (95% CI: 0.24 to 0.81), p=0.01) and OS (HR: 0.24 (95% CI: 0.07 to 0.80), p=0.02), with a trend toward significance in multivariable analysis of PFS (HR: 0.52 (95% CI: 0.27 to 1.00), p=0.05) and OS (HR: 0.28 (95% CI: 0.08 to 0.99), p=0.05). In multivariable analysis of age, sex, race, and cancer type, none of these factors were found to be associated with PFS or OS, with the remaining univariate and multivariable analyses for PFS and OS shown in online supplemental table S3 and online supplemental table S4, respectively.

Figure 1. When comparing obese with non-obese patients, there was a significant difference in median (A) progression free survival (PFS) (log-rank p<0.01), with 12-month PFS 51.7% (95% CI: 35.5% to 75.1%) and 26.5% (95% CI: 17.1% to 41.0%), respectively, and (B) overall survival (OS) (log-rank p=0.01), with 12-month OS 88.5% (95% CI: 77.0% to 100.0%) and 58.0% (95% CI: 44.5% to 75.5%), respectively. (C) While there was a difference in PFS and OS, there was no difference in the best objective response (BOR) (p=0.21) comparing obese with non-obese patients. *Select percentages may not add up to 100% due to rounding. CR, complete response; PD, progressive disease; PR, partial response; SD, stable disease

In our cohort of 94 patients, 11 did not have sufficient data to determine best overall response (BOR). Of the 83 patients with evaluable BOR, Nine patients had a CR and 20 had a PR, for an ORR of 35% (n=29). 21 patients had stable disease, and 33 patients had PD as their BOR. There was not a significant difference in ORR comparing obese with non-obese patients (46% vs 30%, p=0.21) (figure 1C and online supplemental table S5).

Cytokine analysis at baseline based on BMI

Of 94 patients with any response data, 86 patients had cytokine data available at baseline, with 27 obese (31%) and 59 non-obese patients (69%). The time of baseline sample collection ranged from 8 days prior to ICI initiation to the day of initiation. At baseline, the only cytokine that differed significantly between obese and non-obese patients was IL-15 (2.08 vs 1.71, p=0.03, figure 2A and online supplemental table S6); however, there were no significant differences in PFS (log-rank p=0.23) or OS (log-rank p=0.80) when comparing low versus high IL-15 concentration at baseline in the overall cohort, with 12-month PFS 30.0% (95% CI: 18.7% to 48.1%) and 34.5% (95% CI: 21.4% to 55.7%) and 12-month OS 73.9% (95% CI: 61.0% to 89.5%) and 66.0% (95% CI: 49.5% to 87.9%), respectively.

Figure 2. When comparing obese with non-obese patients (A) at baseline, there was a significant difference in IL-15 concentration, while (B) on-treatment, there were significant differences in concentration of IL-6, IL-8, and IL-15. (C) When comparing the ratio of on-treatment to baseline concentration, IL-6, IL-17E, and VEGF-[alpha] differed significantly comparing obese with non-obese patients. BMI, body mass index.

Cytokine analysis on-treatment based on BMI

82 patients had on-treatment cytokine data available, with 28 obese (34%) and 54 non-obese (66%) patients. The median time of on-treatment sample collection was 42 days after treatment initiation (range: 21–349 days, IQR: 41–49 days). Six samples were collected at the time of irAE.

At the on-treatment timepoint, three cytokines, IL-6, IL-8, and IL-15, differed significantly when comparing obese with non-obese patients (figure 2B and online supplemental table S7). There was a trend toward a difference in median PFS comparing low versus high IL-6 concentration (log-rank p=0.05) (online supplemental figure S1A), with 12-month PFS 45.7% (95% CI: 32.0% to 65.2%) and 30.0% (95% CI: 18.1% to 49.5%), respectively. There was a significant difference in median OS when comparing low versus high IL-6 concentration (log-rank p<0.01) (online supplemental figure S1B), with 12-month OS 88.6% (95% CI: 78.6% to 100.0%) and 60.0% (95% CI: 44.5% to 80.9%), respectively. IL-6 concentration remained a significant predictor of OS on multivariable analysis adjusting for age, cancer type, gender, and race (HR: 0.27 (95% CI: 0.08 to 0.88), p=0.03) as well as when adding obesity to the analysis (HR: 0.29 (95% CI: 0.09 to 0.93), p=0.04). We observed a negative linear relationship between BMI and IL-6 concentration (Spearman’s rho=−0.24, p=0.03, online supplemental figure S2).

When comparing low versus high IL-8, there was a significant difference in median PFS (log-rank p=0.03) (online supplemental figure S3A), with 12-month PFS 47.8% (95% CI: 33.9% to 67.5%) and 27.2% (95% CI: 16.0% to 46.4%), respectively. There was also a significant difference in median OS when comparing low versus high IL-8 concentration (log-rank p<0.01) (online supplemental figure S3B), with 12-month OS 92.6% (95% CI: 84.9% to 100%) and 55.4% (95% CI: 39.7% to 77.3%), respectively. IL-8 remained a significant predictor of OS in multivariable analysis adjusting for age, cancer group, gender, and race for OS (HR: 0.19 (95% CI: 0.05 to 0.70), p=0.01), as well as when adding obesity to the analysis (HR: 0.25 (95% CI: 0.07 to 0.94), p=0.04), though it showed a trend toward significance in multivariable analysis for PFS (HR: 0.55 (0.30–1.00), p=0.05). Like IL-6, we observed a negative linear relationship between BMI and IL-8 concentration in our cohort (Spearman’s rho=−0.32, p<0.01, online supplemental figure S4). IL-15 concentration on-treatment was not significantly associated with PFS or OS using Kaplan-Meier or Cox proportional hazards analysis when comparing low versus high IL-15 concentration.

Change in cytokine levels on-treatment

We also evaluated fold change in cytokine concentration after treatment. Seventy-six patients had paired values on-treatment and at baseline, with 25 obese (33%) and 51 non-obese (67%) patients. Of the 37 cytokines evaluated, when comparing obese with non-obese patients, significant differences were observed in IL-6 (−0.68 vs 0.12, p=0.01), IL-17E (−0.44 vs 0, p=0.04), and VEGF-α (0 vs 0, p=0.04) (figure 2C, online supplemental table S8). Of note, we did not observe a significant difference in overall median BMI when comparing median baseline with on-treatment values in this cohort of patients (26.8 vs 26.7, p=0.80); however, we did observe a difference when evaluating paired samples using a paired Wilcoxon test (p=0.04) (online supplemental figure S5).

Differences in the immunologic states at a cellular level

Finally, we set out to determine if differences in treatment response could be explained by changes in immune cell phenotypes, using CyTOF to evaluate expression of 37 surface markers on circulating immune cells, followed by clustering analysis to generate 26 clusters using the expression marker profile (online supplemental table S2 and figure 3A,B). Within our cohort of 94 patients, there were 77 patients with CyTOF data available at baseline, with 27 obese (35%) and 50 non-obese (65%) patients included. At baseline, we found that there was a significant difference in the absolute number of four cell clusters: two T effector cell (TcEFF) clusters, one double-negative T-cell (DNT) cluster, and one T helper 2 (Th2) cluster (figure 3C and online supplemental table S9). In non-obese patients, significantly higher expression of T-bet, Ki67, and GZMB was observed in select TcEFF clusters at baseline, while, in obese patients, significantly higher expression of CD27 was seen in TcEFF clusters at baseline, with higher expression of TIM-3 in Th2 clusters as well (figure 4A,B).

Figure 3. (A) Using the entire cohort, samples were assayed with a 37-marker panel for CyTOF, with a FlowSOM algorithm used to make 35 metaclusters, which were finalized into 26 clusters, with the scaled marker profile shown. (B) UMAP plots with the annotated cell clusters. Boxplots comparing obese with non-obese patients (C) at baseline, where there were significant differences in the number of DNT_V, TcEFF_II, TcEFF_III, and Th2CM_II cell populations, and (D) on-treatment, where there were significant differences in the number of DNT_V, TcEFF_II, TcEFF_III, Th2CM_II, and Treg cell populations. CyTOF, cytometry by time of flight; DNT, double-negative T cells; MMI, mean metal intensity; TcEFF, T-effector cells; Th2, T-helper 2; Treg, T-regulatory;UMAP, uniform manifold approximation and projection.

Figure 4. (A) At baseline, obese patients had higher expression of activation and exhaustion markers, with non-obese patients demonstrating higher expression of markers of cytotoxicity, (B) with increased expression of CD27 and TIM-3 in obese as compared with non-obese patients. (C) On-treatment, obese patients continued to have higher expression of activation and exhaustion markers, with non-obese patients again showing higher expression of markers of cytotoxicity, (D) with increased expression of CD27, TIM-3, PD-1, and CTLA-4 observed in obese patients on-treatment. BMI, body mass index.

On-treatment, 76 patients had CyTOF data available, 27 obese (36%) and 49 (64%) non-obese patients. We again found significantly lower abundance of two TcEFF clusters and one DNT cluster in obese patients, whereas a Th2 cluster and T-regulatory cluster were more abundant in obese patients (figure 3D and online supplemental table S10). After treatment, we continued to see higher expression of T-bet, Ki67, and GZMB in TcEFF clusters in non-obese patients; however, we now observed a significant difference in PD-1 in addition to CD27 in TcEFF clusters and TIM-3 and CTLA-4 on Th2 clusters, with higher expression of all four markers in obese patients (figure 4C,D).

Discussion

In this pan-tumor cohort of patients treated with ICI-based therapy, we observed an association between baseline obesity and favorable clinical outcomes. Obesity is considered to cause an immunosuppressive state, and obese patients did demonstrate lower numbers of TcEFF at baseline with increased expression of T-cell checkpoint molecules, including CTLA-4, TIM-3, and PD-1. These results support the model that obesity causes chronic inflammation, leading to an acquired immunosuppressive state via upregulation of inhibitory cell-surface markers. This checkpoint-mediated T-cell dysfunction not only predisposes obese patients to develop malignancy by impairing immune surveillance, but might also paradoxically result in greater clinical benefit from ICIs, which may partially reverse this state and unleash a greater potential for tumor-reactive T cells.^24–28

A notable finding was that, upon treatment with ICIs, obese patients had lower circulating levels of the classically protumorigenic cytokines IL-6 and IL-8, consistent with a differential response to ICIs in the obese population. Low IL-6 and IL-8 concentrations have been associated with improved outcomes after ICI-based therapy, however, the relationship with obesity and ICIs has not been previously investigated to our knowledge.^20–23 IL-6 plays a critical role in the acute phase response, inhibiting apoptosis and cytotoxic T-lymphocyte (CTL) effector differentiation.^{22 29 30} Low IL-6 concentration before and after treatment has been associated with improved outcomes after ICI-based therapy, with IL-6 blockade improving CTL effector differentiation and infiltration.^{21 22 31 32} Similar to IL-6, IL-8 has protumorigenic effects, promoting the epithelial to mesenchymal transition, angiogenesis, and MDSC recruitment, and low IL-8 concentration has been correlated with improved outcomes in many tumor types.^{20 23 33–36} Our study did not observe a difference in baseline IL-6 or IL-8 concentration between obese and non-obese patients. These data suggest that the relationship between obesity and favorable clinical outcomes may be mediated through divergent immune responses, including differences in IL-6 and IL-8 signaling.

Our immune cluster and cell-surface marker analyses build on previous studies showing that obesity alters the immune environment, leading to T-cell exhaustion reversed by ICIs. Similar to adaptive immune resistance in malignancy, where T-cell exhaustion develops as a result of chronic exposure to tumor antigen, models of obesity demonstrate increased expression of inhibitory receptors on T cells to protect against chronic inflammation related to visceral adipose tissue inflammation.^37–45 First, focusing on adaptive immune resistance seen in malignancy, in patients with melanoma and NSCLC, CD8⁺ tumor infiltrating lymphocytes with signatures of T-cell exhaustion, including high PD-1 expression, have been shown to represent an immune cell population with high tumor reactivity, enhanced cytotoxicity, and increased replicative capacity that can be unleashed by PD-(L)1 blockade.^{28 40 46 47} A similar T-cell population has been observed in mouse models of diet-induced obesity, with phenotypic transition from PD-1⁻CD8⁺ T cells to PD-1⁺CD8+ T cells and increased expression of markers of T-cell exhaustion, including TIM-3, LAG-3, TIGIT, and Eomes, and decreased expression of proliferation markers, including Ki-67 and T-bet, in mice fed high-fat diets.^{15 24 48}

While TIM-3 was elevated at baseline in obese patients, other exhaustion markers were not significantly elevated at the time of treatment initiation. This observation stands in contrast to preclinical findings reported by Wang et al, where PD-1 expression was elevated on CD8⁺ T cells in the tumor microenvironment (TME).²⁴ It is possible that this variation results from differences in T-cell populations sampled, with our study focusing on circulating T cells and Wang et al focusing on the TME.²⁴ The observed increase in PD-1 and CTLA-4 expression after treatment in our cohort likely reflects increased TcEFF and Th2 tumor reactivity in obese patients after ICIs and has previously been associated with increased T-cell activation and infiltration, supporting a mechanism for improved response to ICI-based therapy.¹⁶

In addition, we found that CD27, a member of the tumor necrosis factor receptor superfamily with a role in T-cell activation, proliferation, and differentiation, was elevated at baseline and on-treatment in TcEFF in obese patients.^{49 50} CD27 has demonstrated synergy with PD-1 blockade in preclinical models, leading to increased T-cell proliferation and enhanced differentiation into TcEFF that results in strong antitumor immunity.⁵⁰ Increased expression of CD27 may be an important mechanism for reversal of T-cell exhaustion and represent a therapeutic target with particular efficacy in patients with obesity.^{49 50}

Limitations of the present study include a relatively small sample size and use of only peripheral blood, precluding direct analyses of the TME. We were not able to obtain historical weights for patients and thus cannot exclude the possibility that clinical outcomes within the non-obese cohort were negatively skewed by patients with a history of obesity who lost weight due to their cancer, leading them to be included in the non-obese cohort at the time of study enrollment. Although we excluded underweight patients (BMI<18.5 kg/m²) as a proxy for cachexia, a syndrome characterized by loss of muscle mass associated with poor outcomes, we note that there may be patients in both the obese and non-obese cohorts suffering from cancer cachexia.⁵¹ We felt that there was insufficient consensus regarding the definition of cachexia to further delineate patients with this condition in our cohort. Another limitation is that a large portion of our study cohort consisted of patients with HCC and RCC, with smaller numbers of patients with NSCLC and melanoma. Although we did not find any relationship between clinical outcomes and tumor type on multivariable adjustment, the results of this study may not be generalizable to the many tumor types that were underrepresented in our pan-tumor cohort. Furthermore, the use of a pan-tumor cohort may have led us to miss signatures unique to any specific tumor type. Additional studies with larger patient cohorts are needed to further investigate the relationship between obesity and response to ICI therapy.

Overall, we demonstrate that patients with obesity have divergent clinical and immune responses to ICI-based therapy as compared with patients without obesity. These data provide additional evidence supporting the obesity paradox in a pan-tumor cohort, as well as mechanistic insights to explain why obese patients respond favorably to ICI-based therapies.

We would like to acknowledge and thank the patients involved in this study for their participation.

Data availability statement

Data are available on reasonable request. The data generated in this study are not publicly available to preserve patient confidentiality and due to the Health Insurance Portability and Accountability Act; however, deidentified data are available on reasonable request from the corresponding author.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by Johns Hopkins Medicine Institutional Review Boards (IRBs), with IRB #00267960. Participants gave informed consent to participate in the study before taking part.

Footnote

X @sorencharmsaz, @AanikaBalaji, @MarkYarchoan

Contributors Concept and design: SLA, SC, SB, LT, GSC, RM, RG-H, EMJ, WJH, CK, and MY. Acquisition, analysis, or interpretation of data: SLA, SC, HLL, H-LT, LD, KM, MBrancati, AB, KH, EG, EJL, YG, MBaretti, JH-C, TS, SB, LT, GSC, RM, RG-H, EMJ, WJH, CK, and MY. Critical revision of the manuscript for important intellectual content: SLA, SC, H-LT, LD, EJL, YG, MBaretti, JH-C, TS, SB, LT, AG, GSC, RM, RG-H, EMJ, WJH, CK, and MY. Statistical analysis: SLA, SC, H-LT, LD, WJH, CK, and MY. Administrative, technical, or material support: KM, MBrancati, KH, EG, CT, JG, AH, NEG, EMC, EH, SSS, and JD. Supervision: SB, LT, GSC, RM, RG-H, EMJ, WJH, CK, and MY. MY is the guarantor.

Funding This study was funded by the JHU/imCORE Scientific Alliance, sponsored by Genentech.

Competing interests SSS reports travel honoraria from Standard BioTools. EJL reports funding from the Marilyn and Michael Glosserman Fund for Basal Cell Carcinoma and Melanoma Research, Bloomberg∼Kimmel Institute for Cancer Immunotherapy, Moving for Melanoma of Delaware, the Barney Family Foundation, the Laverna Hahn Charitable Trust, and the Julie Ann Robertson Cashour Memorial Fund outside the submitted work. RGH, GSC, and RM are employees of F. Hoffman-La Roche, and AG, SB, and LT are employees of Genentech. EMJ reports other support from AbMeta, other support from Adventris, personal fees from Achilles, personal fees from DragonFly, non-financial support from Parker Institute, personal fees from Surge, grants from Lustgarten, grants from Genentech, personal fees from Mestag, personal fees from Medical Home Group, non-financial support from BMS, grants from Break Through Cancer, personal fees from CPRIT, personal fees from Neuvogen, non-financial support from HDT Bio, and personal fees from NeoTX outside the submitted work. WJH reports support from NIH/NCI P30CA006973 and NIH/NCI R21CA264004, as well as patent royalties from Rodeo/Amgen, received grants from Sanofi and NeoTX (to Johns Hopkins), and speaking/travel honoraria from Exelixis and Standard BioTools. MY reports research funding (to Johns Hopkins) from Genentech, Bristol-Myers Squibb, and Incyte; consulting fees from Genentech, Exelixis, Eisai, AstraZeneca, Replimune, Hepion, and Lantheus; and equity in Adventris. The other authors declare no competing interests.

Provenance and peer review Not commissioned; externally peer reviewed.

Author note The Sidney Kimmel Comprehensive Cancer Center, Genentech Inc., and F. Hoffmann-La Roche Ltd. are all members of the imCORE Network.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ World Health Organization. Obesity and overweight. 2024. Available: https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight [Accessed 10 Mar 2024 ].

² Fryar CD, Carroll MD, Afful J. Prevalence of overweight, obesity, and severe obesity among adults aged 20 and over: united states, 1960–1962 through 2017–2018. Health E-Stats; 2021. Available: https://www.cdc.gov/nchs/data/hestat/obesity-adult-17-18/obesity-adult.htm [Accessed 10 Mar 2024 ].

³ Calle EE, Rodriguez C, Walker-Thurmond K, et al. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med 2003; 348: 1625–38. doi:10.1056/NEJMoa021423

⁴ Lennon H, Sperrin M, Badrick E, et al. The Obesity Paradox in Cancer: a Review. Curr Oncol Rep 2016; 18: 56. doi:10.1007/s11912-016-0539-4

⁵ Kroenke CH, Neugebauer R, Meyerhardt J, et al. Analysis of Body Mass Index and Mortality in Patients With Colorectal Cancer Using Causal Diagrams. JAMA Oncol 2016; 2: 1137–45. doi:10.1001/jamaoncol.2016.0732

⁶ Albiges L, Hakimi AA, Xie W, et al. Body Mass Index and Metastatic Renal Cell Carcinoma: Clinical and Biological Correlations. J Clin Oncol 2016; 34: 3655–63. doi:10.1200/JCO.2016.66.7311

⁷ De Pergola G, Silvestris F. Obesity as a major risk factor for cancer. J Obes 2013. doi:10.1155/2013/291546

⁸ Renehan AG, Tyson M, Egger M, et al. Body-mass index and incidence of cancer: a systematic review and meta-analysis of prospective observational studies. The Lancet 2008; 371: 569–78. doi:10.1016/S0140-6736(08)60269-X

⁹ Kyrgiou M, Kalliala I, Markozannes G, et al. Adiposity and cancer at major anatomical sites: umbrella review of the literature. BMJ 2017; 356: j477. doi:10.1136/bmj.j477

¹⁰ Kichenadasse G, Miners JO, Mangoni AA, et al. Association Between Body Mass Index and Overall Survival With Immune Checkpoint Inhibitor Therapy for Advanced Non-Small Cell Lung Cancer. JAMA Oncol 2020; 6: 512–8. doi:10.1001/jamaoncol.2019.5241

¹¹ McQuade JL, Daniel CR, Hess KR, et al. Association of body-mass index and outcomes in patients with metastatic melanoma treated with targeted therapy, immunotherapy, or chemotherapy: a retrospective, multicohort analysis. Lancet Oncol 2018; 19: 310–22. doi:10.1016/S1470-2045(18)30078-0

¹² Cortellini A, Bersanelli M, Buti S, et al. A multicenter study of body mass index in cancer patients treated with anti-PD-1/PD-L1 immune checkpoint inhibitors: when overweight becomes favorable. J Immunother Cancer 2019; 7: 57. doi:10.1186/s40425-019-0527-y

¹³ Park EJ, Lee JH, Yu G-Y, et al. Dietary and genetic obesity promote liver inflammation and tumorigenesis by enhancing IL-6 and TNF expression. Cell 2010; 140: 197–208. doi:10.1016/j.cell.2009.12.052

¹⁴ Kern L, Mittenbühler MJ, Vesting AJ, et al. Obesity-Induced TNFα and IL-6 Signaling: The Missing Link between Obesity and Inflammation-Driven Liver and Colorectal Cancers. Cancers (Basel) 2018; 11: 24. doi:10.3390/cancers11010024

¹⁵ Porsche CE, Delproposto JB, Geletka L, et al. Obesity results in adipose tissue T cell exhaustion. JCI Insight 2021; 6: e139793. doi:10.1172/jci.insight.139793

¹⁶ Dyck L, Prendeville H, Raverdeau M, et al. Suppressive effects of the obese tumor microenvironment on CD8 T cell infiltration and effector function. J Exp Med 2022; 219: e20210042. doi:10.1084/jem.20210042

¹⁷ Pingili AK, Chaib M, Sipe LM, et al. Immune checkpoint blockade reprograms systemic immune landscape and tumor microenvironment in obesity-associated breast cancer. Cell Rep 2021; 35: 109285. doi:10.1016/j.celrep.2021.109285

¹⁸ Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser 2000; 894: i–xii,.

¹⁹ Ho WJ, Zhu Q, Durham J, et al. Neoadjuvant Cabozantinib and Nivolumab Converts Locally Advanced HCC into Resectable Disease with Enhanced Antitumor Immunity. Nat Cancer 2021; 2: 891–903. doi:10.1038/s43018-021-00234-4

²⁰ Yuen KC, Liu L-F, Gupta V, et al. High systemic and tumor-associated IL-8 correlates with reduced clinical benefit of PD-L1 blockade. Nat Med 2020; 26: 693–8. doi:10.1038/s41591-020-0860-1

²¹ Laino AS, Woods D, Vassallo M, et al. Serum interleukin-6 and C-reactive protein are associated with survival in melanoma patients receiving immune checkpoint inhibition. J Immunother Cancer 2020; 8: e000842. doi:10.1136/jitc-2020-000842

²² Huseni MA, Wang L, Klementowicz JE, et al. CD8⁺ T cell-intrinsic IL-6 signaling promotes resistance to anti-PD-L1 immunotherapy. Cell Rep Med 2023; 4: 100878. doi:10.1016/j.xcrm.2022.100878

²³ Sanmamed MF, Carranza-Rua O, Alfaro C, et al. Serum interleukin-8 reflects tumor burden and treatment response across malignancies of multiple tissue origins. Clin Cancer Res 2014; 20: 5697–707. doi:10.1158/1078-0432.CCR-13-3203

²⁴ Wang Z, Aguilar EG, Luna JI, et al. Paradoxical effects of obesity on T cell function during tumor progression and PD-1 checkpoint blockade. Nat Med 2019; 25: 141–51. doi:10.1038/s41591-018-0221-5

²⁵ Wu B, Chiang H-C, Sun X, et al. Genetic ablation of adipocyte PD-L1 reduces tumor growth but accentuates obesity-associated inflammation. J Immunother Cancer 2020; 8: e000964. doi:10.1136/jitc-2020-000964

²⁶ Wu H, Ballantyne CM. Metabolic Inflammation and Insulin Resistance in Obesity. Circ Res 2020; 126: 1549–64. doi:10.1161/CIRCRESAHA.119.315896

²⁷ Seijkens T, Kusters P, Chatzigeorgiou A, et al. Immune cell crosstalk in obesity: a key role for costimulation? Diabetes 2014; 63: 3982–91. doi:10.2337/db14-0272

²⁸ Gros A, Robbins PF, Yao X, et al. PD-1 identifies the patient-specific CD8+ tumor-reactive repertoire infiltrating human tumors. J Clin Invest 2014; 124: 2246–59: 73639. doi:10.1172/JCI73639

²⁹ Weber R, Groth C, Lasser S, et al. IL-6 as a major regulator of MDSC activity and possible target for cancer immunotherapy. Cell Immunol 2021; 359: 104254. doi:10.1016/j.cellimm.2020.104254

³⁰ Liu C, Yang L, Xu H, et al. Systematic analysis of IL-6 as a predictive biomarker and desensitizer of immunotherapy responses in patients with non-small cell lung cancer. BMC Med 2022; 20: 187. doi:10.1186/s12916-022-02356-7

³¹ Keegan A, Ricciuti B, Garden P, et al. Plasma IL-6 changes correlate to PD-1 inhibitor responses in NSCLC. J Immunother Cancer 2020; 8: e000678. doi:10.1136/jitc-2020-000678

³² Hailemichael Y, Johnson DH, Abdel-Wahab N, et al. Interleukin-6 blockade abrogates immunotherapy toxicity and promotes tumor immunity. Cancer Cell 2022; 40: 509–23. doi:10.1016/j.ccell.2022.04.004

³³ Alfaro C, Teijeira A, Oñate C, et al. Tumor-Produced Interleukin-8 Attracts Human Myeloid-Derived Suppressor Cells and Elicits Extrusion of Neutrophil Extracellular Traps (NETs). Clin Cancer Res 2016; 22: 3924–36. doi:10.1158/1078-0432.CCR-15-2463

³⁴ Schalper KA, Carleton M, Zhou M, et al. Elevated serum interleukin-8 is associated with enhanced intratumor neutrophils and reduced clinical benefit of immune-checkpoint inhibitors. Nat Med 2020; 26: 688–92. doi:10.1038/s41591-020-0856-x

³⁵ Teijeira A, Garasa S, Ochoa MC, et al. IL8, Neutrophils, and NETs in a Collusion against Cancer Immunity and Immunotherapy. Clin Cancer Res 2021; 27: 2383–93. doi:10.1158/1078-0432.CCR-20-1319

³⁶ Sanmamed MF, Perez-Gracia JL, Schalper KA, et al. Changes in serum interleukin-8 (IL-8) levels reflect and predict response to anti-PD-1 treatment in melanoma and non-small-cell lung cancer patients. Ann Oncol 2017; 28: 1988–95. doi:10.1093/annonc/mdx190

³⁷ Ribas A. Adaptive Immune Resistance: How Cancer Protects from Immune Attack. Cancer Discov 2015; 5: 915–9. doi:10.1158/2159-8290.CD-15-0563

³⁸ Shirakawa K, Yan X, Shinmura K, et al. Obesity accelerates T cell senescence in murine visceral adipose tissue. J Clin Invest 2016; 126: 4626–39: 88606. doi:10.1172/JCI88606

³⁹ Le CT, Khuat LT, Caryotakis SE, et al. PD-1 Blockade Reverses Obesity-Mediated T Cell Priming Impairment. Front Immunol 2020; 11: 590568. doi:10.3389/fimmu.2020.590568

⁴⁰ Chow A, Perica K, Klebanoff CA, et al. Clinical implications of T cell exhaustion for cancer immunotherapy. Nat Rev Clin Oncol 2022; 19: 775–90. doi:10.1038/s41571-022-00689-z

⁴¹ Sade-Feldman M, Yizhak K, Bjorgaard SL, et al. Defining T Cell States Associated with Response to Checkpoint Immunotherapy in Melanoma. Cell 2018; 175: 998–1013. doi:10.1016/j.cell.2018.10.038

⁴² Jenkins E, Whitehead T, Fellermeyer M, et al. The current state and future of T-cell exhaustion research. Oxf Open Immunol 2023; 4: iqad006. doi:10.1093/oxfimm/iqad006

⁴³ Schietinger A, Philip M, Krisnawan VE, et al. Tumor-Specific T Cell Dysfunction Is a Dynamic Antigen-Driven Differentiation Program Initiated Early during Tumorigenesis. Immunity 2016; 45: 389–401. doi:10.1016/j.immuni.2016.07.011

⁴⁴ Budimir N, Thomas GD, Dolina JS, et al. Reversing T-cell Exhaustion in Cancer: Lessons Learned from PD-1/PD-L1 Immune Checkpoint Blockade. Cancer Immunol Res 2022; 10: 146–53. doi:10.1158/2326-6066.CIR-21-0515

⁴⁵ Crespo J, Sun H, Welling TH, et al. T cell anergy, exhaustion, senescence, and stemness in the tumor microenvironment. Curr Opin Immunol 2013; 25: 214–21. doi:10.1016/j.coi.2012.12.003

⁴⁶ Li H, van der Leun AM, Yofe I, et al. Dysfunctional CD8 T Cells Form a Proliferative, Dynamically Regulated Compartment within Human Melanoma. Cell 2019; 176: 775–89. doi:10.1016/j.cell.2018.11.043

⁴⁷ Thommen DS, Koelzer VH, Herzig P, et al. A transcriptionally and functionally distinct PD-1⁺ CD8⁺ T cell pool with predictive potential in non-small-cell lung cancer treated with PD-1 blockade. Nat Med 2018; 24: 994–1004. doi:10.1038/s41591-018-0057-z

⁴⁸ Kado T, Nawaz A, Takikawa A, et al. Linkage of CD8⁺ T cell exhaustion with high-fat diet-induced tumourigenesis. Sci Rep 2019; 9: 12284. doi:10.1038/s41598-019-48678-0

⁴⁹ Starzer AM, Berghoff AS. New emerging targets in cancer immunotherapy: CD27 (TNFRSF7). ESMO Open 2019; 4: e000629. doi:10.1136/esmoopen-2019-000629

⁵⁰ Buchan SL, Fallatah M, Thirdborough SM, et al. PD-1 Blockade and CD27 Stimulation Activate Distinct Transcriptional Programs That Synergize for CD8⁺ T-Cell-Driven Antitumor Immunity. Clin Cancer Res 2018; 24: 2383–94. doi:10.1158/1078-0432.CCR-17-3057

⁵¹ Bruggeman AR, Kamal AH, LeBlanc TW, et al. Cancer Cachexia: Beyond Weight Loss. J Oncol Pract 2016; 12: 1163–71. doi:10.1200/JOP.2016.016832