Abstract
Background/Aim: Immune checkpoint inhibitors (ICIs) play an important role in the treatment of esophageal cancer (EC). However, few patients achieve long-term survival, and some patients develop serious immune-related adverse events (irAEs). Reliable predictive biomarkers of efficacy and safety need to be established in order to improve efficacy. We retrospectively analyzed the outcomes of nivolumab monotherapy on EC at Showa University, Department of Medicine, to identify biomarkers and characteristics of patients who benefit from ICI monotherapy. Patients and Methods: Eighty-six patients with EC who received nivolumab monotherapy were included in the present study. Patient characteristics, efficacy, and safety were analyzed. A multivariable analysis evaluated the correlation among overall survival (OS), progression-free survival (PFS), best overall response (BOR), irAEs, and the following variables: sex, age, performance status (PS), neutrophil-to-lymphocyte ratio (NLR), C-reactive protein (CRP) level, albumin level, and body-mass index before treatment. Results: Median PFS was 3.1 months, and median OS was 9.0 months. In multivariable analysis, pretreatment PS, NLR, and sex were significantly correlated with OS and PFS. NLR <3.3 predicted longer survival (median OS 17.5 vs. 6.4 months for NLR ≥3.3; p<0.001). Median OS was 10.6 months for PS 0-1 and 1.3 months for PS 2-3 (p<0.001). NLR remained significantly predictive in the PS 0-1 group. The development of irAEs was significantly associated with increased OS and PFS. Conclusion: Patients with low NLR and good PS before treatment may maximize the benefits of ICIs. A low NLR may be an indicator of higher immunocompetence for anti-tumor immunity, suggesting that NLR may be a convenient predictive biomarker in daily practice.
- Immunotherapy
- neutrophil-to-lymphocyte ratio
- performance status
- immune checkpoint inhibitors
- nivolumab
- esophageal cancer
- predictive biomarker
Immune checkpoint inhibitors (ICIs) can result in long-term disease control and cancer cure in a minority of cancer patients, outcomes that were unachievable with traditional chemotherapy. ICIs have shown promise for esophageal cancer (EC). The combination of an anti-programmed death-1 (PD-1) antibody (nivolumab or pembrolizumab) with chemotherapy or an ICI doublet such as nivolumab and ipilimumab in first-line treatment, and nivolumab monotherapy in second-line treatment have been approved by the regulatory authorities for EC (1-3). However, only a limited number of patients achieve long-term survival. Serious immune-related adverse events (irAEs) can require treatment discontinuation (4).
Potential biomarkers of ICI efficacy include programmed death-ligand 1 (PD-L1) expression, microsatellite instability (MSI), and tumor mutation burden (TMB), which have been applied clinically (5-9). Although they are useful for identifying the population most likely to respond to ICIs, they are not sufficiently reliable as biomarkers because some patients with these characteristics do not respond to ICIs. Other potential biomarkers include gut microbiota, cytokine signatures, and tertiary lymphoid structures (TLS); however, they are not convenient and are still experimental.
In the present study, we performed a retrospective analysis of ICI efficacy on EC to identify convenient and cost-effective reliable biomarkers that can be used in daily practice. Focusing on mono-immunotherapy, we analyzed the efficacy and safety of nivolumab monotherapy for EC in clinical practice and identified the characteristics of patients who benefited from ICIs using predictive biomarker analysis.
Patients and Methods
Patients who received nivolumab monotherapy for pathologically-confirmed EC between February 2020 and June 2023 were included in the present study. Eighty-six patients with EC were administered nivolumab monotherapy. We retrospectively analyzed the efficacy, safety, and clinical-laboratory values of these patients. Data collection for this study was focused on routine medical care and included patient characteristics [age, sex, ECOG performance status (PS), height, weight, body-mass index (BMI), stage, and histology], laboratory findings, imaging studies, and evaluation of safety and efficacy. Biomarker analysis included age, sex, ECOG performance status, BMI, neutrophil-to-lymphocyte ratio (NLR), C-reactive protein (CRP), and albumin level prior to treatment initiation. Safety was evaluated according to the presence and severity of irAEs based on the Common Terminology Criteria for Adverse Events (CTCAE) ver. 5.0. Efficacy was evaluated using the Response Evaluation Criteria in Solid Tumors (RECIST) ver. 1.1, including the overall response rate (ORR), disease-control rate (DCR), progression-free survival (PFS), and overall survival (OS). The studies involving human participants were reviewed and approved by the Ethics Committee of Showa University School of Medicine (approval No. 22-168-A). Since this is a retrospective study using only data from medical records, informed consent was obtained in the form of opt-out on a website, ensuring appropriate opportunities for refusal regarding the use of medical records.
Statistical analysis for biomarkers. Efficacy was evaluated based on OS, PFS, and the best overall response (BOR). BOR was classified into two groups: progressive-disease (PD) group and non-PD [stable disease (SD)–complete response (CR)] group. Safety was evaluated based on irAEs and patients were classified into two groups according to the presence or absence of irAEs. Variables included in the multivariable analysis were sex (male vs. female), PS (0-1 vs. 2-3), age, BMI, CRP level, NLR, and albumin level, which were measured prior to treatment initiation. Multivariable analyses were performed using the Cox proportional hazards model for OS and PFS. Multivariable analysis was performed using nominal logistic regression for BOR and irAEs. In BOR analysis, PD was defined as an event. Survival-time analysis was performed by plotting survival curves using the Kaplan-Meier method, which were compared using the log-rank test. Two-sided p-values less than 0.05 were considered statistically significant. Statistical analyses were performed using JMP® Pro 15 (SAS Institute Inc., Cary, NC, USA).
Results
Baseline characteristics of the patients. The median patient age was 68 years (range=49-88 years), 80.2% were men, and 80.2% of the cases had ECOG PS 0-1, whereas 18.6% and 1.2% of the patients had PS 2 and 3, respectively. Stage IVB occurred in 96.4% of the cases. All patients were diagnosed with squamous cell carcinoma of the esophagus. With respect to treatment, 94.2% were second-line or later, while 5.8% were first-line treatment. This was primarily because of rejection by the patients of cytotoxic anticancer drugs. The study represented a diverse patient population, including extremely underweight patients and those with high neutrophil counts or CRP levels (Table I).
Baseline characteristics of esophageal cancer (EC) patients treated with nivolumab (n=86).
Response of EC patients to nivolumab treatment and safety. The median PFS was 3.1 months [95% confidence interval (CI)=2.2-4.6; Figure 1A], and the median OS was 9.0 months (95%CI=6.6-10.6; Figure 1B) for patients treated with nivolumab monotherapy. The best overall response rates (BOR) were 1.2% for CR, 12.8% for partial response (PR), 32.5% for SD, and 53.5% for progressive disease (PD), respectively. The ORR was 14.0% and the DCR was 46.5% (Table II). IrAEs were observed in 37.2% of cases of all grades and 8.1% in grades 3-4 including hepatitis in three cases, interstitial lung disease in one case, hypopituitarism in one case, hemophagocytic syndrome in one case, and skin rash in one case (Table III).
Survival analysis. Kaplan-Meier curves of progression-free survival (PFS) (A) and overall survival (OS) (B) for esophageal-cancer (EC) patients treated with nivolumab monotherapy.
Response of EC patients to nivolumab treatment.
Immune-related adverse events in EC patients treated with nivolumab based on CTCAE ver 5.0.
Multivariable analysis of candidate biomarkers of response of EC patients to nivolumab. Multivariable analysis was performed to determine predictive biomarkers from routine clinical data. NLR, PS, and sex were significantly correlated with OS and PFS (Table IV and Table V). The BOR in the non-PD (SD-CR) and PD groups was significantly correlated with PS and sex (Table VI). Sex was significantly correlated with irAEs (Table VII). Other variables, including age, BMI, CRP level, and albumin level did not correlate with any outcome.
Cox proportional hazards analysis of progression-free survival (PFS) of EC patients treated with nivolumab.
Cox proportional hazards analysis of overall survival (OS) of EC patients treated with nivolumab.
Nominal logistic regression analysis of BOR of EC patients treated with nivolumab.
Nominal logistic regression analysis of immune-related adverse events (irAEs) of EC patients treated with nivolumab.
We performed an in-depth analysis of the data with respect to NLR. Patients were classified into two groups based on 18.0 month survival. Because the median OS was 9.0 months, patients who survived for 18.0 months, which is twice the median OS, were classified as long-term survivors treated with immunotherapy. The cutoff value for NLR to predict long-term survival was 3.3, determined using the Youden index from the receiver operating characteristic (ROC) curve analysis. Survival-time analysis of the two groups based on the cut-off value of 3.3 for NLR was performed. Figure 2 shows the Kaplan-Meier curves of PFS and OS classified according to NLR. PFS was 5.9 months (95%CI=3.1-20.6) for the NLR <3.3 group vs. 2.1 months (95%CI=1.6-3.8; p<0.001) for the NLR ≥3.3 group; OS was 17.5 months (95%CI=9.3-NA) for the NLR <3.3 group vs. 6.4 months (95%CI=3.8-8.7; p<0.001) for the NLR ≥3.3 group, respectively.
Survival analysis stratified by neutrophil-lymphocyte ratio (NLR) of EC patients treated with nivolumab. Kaplan-Meier curves of progression-free survival (PFS) (A) and overall survival (OS) (B) stratified by NLR (NLR <3.3 vs. NLR ≥3.3). The group with NLR<3.3 had significantly better PFS and OS compared to the group with NLR ≥3.3 (p<0.001).
We found PS to also be a significant factor correlated with clinical outcomes. In Figure 3, the Kaplan-Meier curves of PFS and OS based on NLR or PS are shown. PFS and OS were determined in the two groups with a cutoff value of 3.3 for NLR and restricted to the PS 0-1 group. PFS was 7.0 months (95%CI=3.7-20.6) for the NLR <3.3 group vs. 3.8 months (95%CI=2.0-5.3; p<0.041) for the NLR ≥3.3 group; OS was 18.9 months (95%CI=10.2-NA) for the NLR <3.3 group vs. 9.0 months (95%CI=6.4-12.4; p<0.0328) for the NLR ≥3.3 group.
Survival analysis stratified by neutrophil-lymphocyte ratio (NLR) restricted to the performance status (PS) 0-1 group of EC patients treated with nivolumab. Kaplan-Meier curves of progression-free survival (PFS) (A) and overall survival (OS) (B) stratified by NLR (NLR <3.3 vs. NLR ≥3.3). The group with NLR <3.3 had a significantly better PFS (p<0.041) and OS (p<0.0328) than the group with NLR ≥3.3.
Patients in the good PS group, defined as PS 0 and 1, showed a significantly better clinical response than those in the impaired PS group, defined as PS 2 and 3 (Figure 4). PFS was 4.4 months (95%CI=3.1-6.5) for the good PS group vs. 0.8 months (95%CI=0.4-1.6; p<0.001) for the impaired PS group, respectively. OS was 10.6 months (95%CI=9.0-18.9) for the good PS group vs. 1.3 months (95%CI=0.7-3.2; p<0.001) for the impaired PS group, respectively.
Survival analysis stratified by performance status (PS) of EC patients treated with nivolumab. Kaplan-Meier curves of progression-free survival (PFS) (A) and overall survival (OS) (B) stratified by PS (PS 0-1 vs. PS 2-3). Patients with PS 0-1 had significantly better PFS and OS than the group with PS 2-3 (p<0.001).
Correlation between irAEs and clinical efficacy of nivolumab on EC patients. Clinical efficacy was also significantly correlated with the incidence of irAEs. The group with irAEs had significantly better clinical efficacy than the group without irAEs (Figure 5). PFS and OS were 5.8 months (95%CI=3.7-20.6) and 22.3 months (95%CI=10.2-NA), respectively, in the group with irAEs vs. 2.3 months (95%CI=1.7-3.1; p=0.005) and 6.4 months (95%CI=4.2-9.0; p<0.001), respectively in the group without irAEs.
Survival analysis stratified by immune-related adverse events (irAEs) of EC patients treated with nivolumab. Kaplan-Meier curves of progression-free survival (PFS) (A) and overall survival (OS) (B) stratified by irAEs (irAE+ vs. irAE−). IrAE (+) refers to patients who develop irAEs. The group with irAEs(+) had a significantly better PFS and OS than the group with irAEs(−).
Discussion
In the present study the clinical efficacy of nivolumab monotherapy for EC in clinical practice, including diverse patients with impaired PS, was comparable to that in a previous clinical trial in patients with good PS (3). In the present study ICI monotherapy achieved a disease-control rate of approximately 50%, indicating high efficacy for esophageal squamous-cell carcinoma. Predictive biomarker analysis showed that pretreatment NLR, PS, and sex were significantly correlated with the efficacy of nivolumab monotherapy. In particular, we focused on NLR because it has been reported to be related to immunological response. In addition, we analyzed the correlation between efficacy and safety to identify biomarkers from sources other than routine clinical data. Patients with irAEs showed significantly better clinical efficacy than those without irAEs in the present study.
With respect to the clinical efficacy and safety of nivolumab monotherapy in a previous clinical trial, the median OS was 10.9 months and median PFS was 1.7 months (3). The BOR results showed that the ORR was 19.9% and 37.0 %, respectively, in the clinical trial. The incidence of irAEs in the clinical trial was 65.0% for all grades and 18.0% for grades 3-4. Although the incidence of irAEs was low in in the present study, information may not have been fully collected due to the retrospective nature of the study. The present study included approximately 20% of patients with impaired PS, suggesting that the efficacy and safety are comparable to those of the previous clinical trial (3).
PD-L1 expression, MSI, and TMB have been proposed as predictive biomarkers of the clinical efficacy of ICIs (5-9). Although they are helpful in identifying the population that is most likely to respond to ICIs, their reliability as predictive biomarkers is insufficient because most patients do not respond to ICIs.
Other identified predictive biomarkers for response to ICIs include gut microbiota, cytokine signatures, and tertiary lymphoid structures (TLS), although they lack convenience and have many obstacles to overcome (10-16). The albumin to fibrinogen ratio (AFR) has also been identified as a significant biomarker in EC patients who received curative resection. However, the association between AFR and the efficacy of ICIs has not been thoroughly investigated (17).
In terms of NLR as a predictive biomarker, patients with a high NLR prior to treatment initiation showed low clinical efficacy in the present study. NLR is a convenient biomarker that can be used in daily practice. The cutoff value of NLR for predicting long-term survival was 3.3 in the present study. Since NLR was also valid as a predictive biomarker in survival analysis restricted to the good PS group, it was considered a significant biomarker. In contrast, high CRP levels before treatment initiation were not correlated with clinical efficacy in the present study. Both high NLR and CRP levels might indicate the presence of hypercytokinemia, high interleukin (IL)-6, and other immunosuppressive mediators derived from tumors in the background. The present results suggest that, in contrast to CRP, NLR is a better indicator of immune status that predicts the response to ICIs. NLR is generally indicative of chronic stress and inflammation. In the bigmouth buffalo (Ictiobus cyprinellus), a long-living fish, a negative correlation between age and NLR has been reported and NLR tends to increase with age in many vertebrates, including humans (18, 19). This suggests that a low NLR may indicate good health conditions with low stress and inflammation. Furthermore, a correlation has been reported between the therapeutic efficacy of ICIs and the presence TLS in tumor tissue, and a positive correlation has been demonstrated between low NLR and high TLS expression (20, 21). Thus, low NLR may be an indicator of high immunocompetence in anti-tumor immunity.
Previous reports have also shown that a high NLR prior to treatment initiation correlates with poor efficacy in patients with various solid tumors, such as pancreatic, lung, and colorectal tumors, treated with cytotoxic anticancer agents (22-24). Similar results have been shown in various cancer types, including non-small cell lung cancer, malignant melanoma, and urothelial carcinoma, treated with ICIs (25-28). However, some reports have found no correlation between NLR and clinical efficacy (29). Several meta-analyses, including those on cytotoxic anticancer agents and ICIs, have also shown that a high NLR prior to treatment initiation is correlated with poor therapeutic efficacy and prognosis. Caution is required in evaluation and utilization, of NLR as a biomarker of response of EC patients to nivolumab since the cutoff values vary in previous studies and its availability as a biomarker varies by cancer type and treatment (30, 31).
It has been reported that the elevation of NLR is mainly due to IL-6 derived from the tumor, which affects the bone marrow and induces an increase in neutrophils. In addition, IL-8 derived from the tumor activates neutrophils, which migrate toward tumor sites. Neutrophils secrete matrix metalloproteinase-8 (MMP-8), vascular endothelial growth factor (VEGF), and neutrophil elastase, which promote tumor growth and angiogenesis, and suppress T cells in the tumor microenvironment (32-34). IL-6 derived from tumors has been reported to be significantly correlated with tumor size, stage, and proliferative activity, as assessed by Ki-67; thus, it is assumed to increase when disease progression occurs (35, 36). In contrast, the lymphocyte count is a key factor in antitumor immunity. Therefore, a high NLR is not only a reflection of disease progression, but also of an unfavorable immune status for antitumor immunity.
CRP elevation is due to IL-6 derived from the tumor, which affects the liver, resulting in a relative decrease in albumin synthesis and elevated fibrinogen and hepcidin levels, causing anemia due to impaired iron utilization (37, 38). In contrast to NLR, high CRP levels have not been reported to have a detrimental effect on antitumor immunity. Both NLR and CRP reflect inflammation mainly due to IL-6, whereas high NLR is unfavorable for antitumor immunity. Patients with low NLR prior to treatment initiation may be the population that can maximize the efficacy of ICIs.
With regard to PS, whereas molecular-targeted agents can improve the PS of lung cancer patients with driver mutations with impaired PS, ICI monotherapy did not provide long-term survival for patients with impaired PS (39, 40). ICIs are sometimes chosen to treat patients who cannot receive cytotoxic agents due to impaired PS (41). However, patients with impaired PS may not receive sufficient benefit from ICIs compared to those treated with molecular-targeted agents. Previous reports have shown that impaired PS correlates with an insufficient response to ICI treatment, indicating that the immune status may also be impaired in those with impaired PS (42).
Regarding irAEs, several reports have shown that clinical efficacy of ICIs was higher in patients who developed irAEs than in those without, and we found similar results in the present study (43, 44). The development of irAEs may indicate that ICIs affect the immune system and irAEs may serve as biomarkers of clinical efficacy. It has been reported that patients who are forced to discontinue ICIs, due to severe irAEs, also have a better prognosis than those who do not develop irAEs (45-47). In contrast, patients with a high clinical efficacy of ICIs may have a high incidence of irAEs due to longer survival. irAEs should be further assessed as biomarkers of clinical efficacy of ICI treatment. In the present study, no correlation was observed between irAEs and NLR. The immunological mechanism by which the NLR affects clinical efficacy of ICIs may be different from the mechanism that underlies the correlation of the development of irAEs with clinical efficacy, although clinical efficacy and irAEs depend on the immune response.
A limitation of the present study is that it was a retrospective analysis conducted at a single institute, and it is necessary to confirm these findings in a prospective study.
Conclusion
Low NLR in patients with EC before treatment initiation may maximize the benefits of ICIs. A low NLR may be a predictive biomarker for ICI therapy for EC and an indicator of high immunocompetence in antitumor immunity. Patients with impaired PS demonstrate poorer clinical efficacy of ICIs. In addition, the development of irAEs can be used as a potential biomarker of increased ICI efficacy.
Acknowledgements
The Authors thank Professor Eisuke Inoue, Showa University Research Administration Center, Showa University, for advice on statistical analysis.
Footnotes
Authors’ Contributions
YH and TaT conceptualized the study, performed investigation, analyzed data, and wrote the article. EM, RS, ToT, NI, TI, RO, MS, HA, AH, SW, TA, SG performed data verification for data collection and analysis and provided technical support. YK (Yutaro Kubota), KO, MM, YK (Yuji Kiuchi), KY, RMH, and TaT organized the findings of this work and reviewed the manuscript. RMH revised the article. TaT administered the project. All Authors read and approved the final manuscript.
Funding
No funding was received for conducting this study.
Conflicts of Interest
The Authors declare no conflicts of interest in relation to this study.
- Received June 10, 2024.
- Revision received June 17, 2024.
- Accepted June 18, 2024.
- Copyright © 2024 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
