Nivolumab with definitive chemoradiotherapy for oesophageal squamous cell carcinoma (NOBEL): a multicentre, single-arm, phase 2 feasibility trial

Summary

Background

This single-arm, phase 2 trial investigated the safety, efficacy, and biomarkers associated with combining nivolumab, an immune checkpoint inhibitor, with definitive chemoradiotherapy in patients with operable or inoperable oesophageal squamous cell carcinoma (OSCC) because these have not been well established.

Methods

In this multicentre, single-arm, phase 2 feasibility trial, eligible patients (aged 20–75 years) with histologically confirmed OSCC, Eastern Cooperative Oncology Group performance status of 0–1, and adequate organ and bone marrow functions were enrolled from five Japanese institutions. The treatment involved concurrent chemoradiotherapy (cisplatin and 5-fluorouracil) with nivolumab, followed by maintenance nivolumab for up to 1 year. The primary endpoint was safety, defined as ≤10% incidence of grade ≥4 non-haematological toxicity and ≤15% incidence of grade ≥3 pneumonitis. Secondary endpoints included complete response, progression-free survival, and overall survival. Biomarker analyses of 51 immuno-related genes were performed on pretreatment biopsy specimens. This trial is registered in the Japan Registry of Clinical Trials, jRCT1091220408, and was terminated early due to slow patient enrolment.

Findings

Between January 2019 and September 2021, 42 patients were enrolled and included in the safety analysis set, whereas 41 patients who received at least one post-baseline tumour assessment comprised the efficacy analysis set. The trial met its primary safety endpoint with only 5% of patients (2 of 41) experiencing grade 3 pneumonitis. The most common adverse events of were oesophagitis, constipation, and lymphopenia, and no treatment-related deaths occurred. 1-year overall survival was 92.7% (95% CI 79.0–97.6), and 1-year progression-free survival was 65.4% (95% CI 48.6–77.9). The overall complete response rate was 73% (30 of 41; 95% CI 58–84%). Exploratory biomarker analyses were conducted to investigate immune-related gene expression, but these findings should be regarded as hypothesis-generating.

Interpretation

Nivolumab combined with definitive chemoradiotherapy is feasible and showed acceptable toxicity in patients with OSCC. Further validation of exploratory biomarker findings is warranted in larger controlled studies.

Funding

Ono Pharmaceutical.

Research in context.

Evidence before this study

We conducted a literature search in the electronic databases PubMed, EMBASE, and the Cochrane Library using the following keywords: “oesophageal cancer”, “chemoradiotherapy”, and “immune checkpoint inhibitors”, without language restrictions for article published up to December 31, 2018, before the start of this trial. No clinical trial in oesophageal cancer has evaluated the efficacy and safety of combining immune checkpoint inhibitors with definitive chemoradiotherapy, although such synergistic effects have been suggested in other cancer types, such as non-small cell lung cancer (PACIFIC trial).

Added value of this study

This multicentre, single-arm, phase 2 feasibility trial provides the first-of-its-kind data on the safety and efficacy of combining nivolumab with definitive chemoradiotherapy for both operable and inoperable oesophageal squamous cell carcinoma. The trial met its predefined safety endpoint, demonstrating that nivolumab addition did not increase the incidence of grade ≥3 pneumonitis (5%, 2 of 42). Moreover, the regimen achieved a high complete response rate (73%, 95% CI 58–84; 30 of 41 patients) and favourable 1-year overall survival (92.7%, 95% CI 79.0–97.6), surpassing historical outcomes for chemoradiotherapy alone. Exploratory biomarker analysis identified an immune high-active intrinsic subtype potentially associated with improved response to this combined treatment.

Implications of all the available evidence

These findings suggest that nivolumab and definitive chemoradiotherapy is feasible and safe for patients with oesophageal squamous cell carcinoma and suggest a possible efficacy signal that warrant confirmation. The results support further randomised trials to validate these observations and to refine patient selection through biomarker-guided strategies for curative-intent immune-chemoradiotherapy in oesophageal cancer.

Introduction

Oesophageal cancer is the seventh most common cancer worldwide.¹ Oesophageal squamous cell carcinoma (OSCC) is the predominant histological type globally, particularly in Asian countries, whereas the incidence of oesophageal adenocarcinoma has markedly increased in developed countries.² For operable oesophageal cancer, neoadjuvant chemotherapy followed by surgery is the standard treatment, regardless of histological type.^3,4 Alternatively, definitive chemoradiotherapy is a curative-intent option that allows oesophageal preservation.5, 6, 7

The standard definitive chemoradiotherapy regimen for oesophageal cancer, based on the Radiation Therapy Oncology Group 9405 trial, consists of four cycles of 5-fluorouracil (5-FU) plus cisplatin combined with 50.4 Gy radiation.⁵ In Japan, the JCOG0909 trial reported a complete response rate of 59% and a 3-year overall survival of 74.2% with chemoradiotherapy and salvage treatment.⁶ Despite these outcomes, further improvement in prognosis is needed.

Immune checkpoint inhibitors (ICIs), especially anti-PD-1 antibodies, have become a standard of care for recurrent or metastatic OSCC.8, 9, 10, 11 Clinical trials in other cancers have demonstrated that ICIs combined with chemoradiotherapy may have synergistic effects, but data on OSCC are limited.^12,13 ICIs function by blocking PD-1 and PD-L1 interactions, thereby enhancing T cell-mediated immune responses. Although PD-L1 expression is often used to predict the efficacy of PD-1 inhibitors, the variable and dynamic nature complicates its reliability as a biomarker. Consequently, no definitive predictive biomarkers for response to PD-1 inhibitors, including nivolumab, have been established.

We previously identified an immune-active subtype of OSCC through comprehensive gene expression profiling, which is associated with a high complete response rate following chemoradiotherapy.¹⁴ However, the predictive value of these subtypes in the setting of combined ICI and chemoradiotherapy in OSCC remains unclear.

This multi-centre phase 2 trial evaluated the safety, efficacy, and potential predictive biomarkers for the combined use of nivolumab and definitive chemoradiotherapy in both operable and inoperable OSCC.

Methods

Study design and participants

This was a multicentre, open-label feasibility study conducted across five hospitals in Japan. Patients were eligible for inclusion if they had histologically confirmed thoracic OSCC, adenosquamous carcinoma, or basaloid carcinoma and were suitanle for definitive chemoradiotherapy on the basis of one of the following conditions: operable stage defined as clinical stage T1N1-3M0 and T2-3 N0-3M0 according to the 8th edition of the International Union Against Cancer staging system, with patients refusing surgical resection as the initial therapy; or inoperable stage defined as clinical stage T4 or metastases to regional or supraclavicular lymph nodes involving other adjacent organs. Patients with bleeding or airway obstruction caused by oesophageal hiatus involvement, oesophago-respiratory fistula, oesophago-mediastinal fistula, or arterial invasion were ineligible. Additional eligibility criteria included: Eastern Cooperative Oncology Group performance status of 0–1; age 20–75 years; and adequate bone marrow, hepatic, and renal function (neutrophil count ≥1500/mm³, white blood cell count ≤12,000/mm³; haemoglobin ≥8.0 g/dL, platelet count ≥100,000/mm³, total bilirubin ≤1.5 times the upper limit of facility standard, aspartate aminotransferase ≤3.0 times the upper limit of facility standard, alanine transaminase ≤3.0 times the upper limit of facility standard, and creatinine clearance ≥60 mL/min).

Patients were excluded if they had a history of previous treatment for oesophageal cancer (except curative endoscopic resection), had previously received chemotherapy and/or chemoradiotherapy for other malignancies, or had any other active malignancies within 5 years prior to enrolment.

Ethics

Ethical approval was obtained from all local ethical committees of the participating hospitals as follows: Kyoto University Hospital Ethics Committee (approval No. K050), the Ethic Committee of Chiba Cancer Center (approval No. 3021), the Institutional Review Board of Kitasato University Hospital (approval No.2019014), the Institutional Review Board of National Cancer Center Hospital (approval No. T4739), and the Institutional Review Board of National Cancer Center Hospital East (approval No. K0834). Written informed consent was obtained from all participants. The study protocol was approved by the institutional review board of each participating centre. This trial was registered in the Japan Registry of Clinical Trials as jRCT1091220408 on January 29, 2019. The first participant was enrolled on February 25, 2019. The participant enrolment period spanned from February 25, 2019 to August 5, 2021, with the last follow-up for primary analysis occurring on April 30, 2023. The extended enrolment period was due to the health care system's recovery from COVID-19.

Procedures

Immunochemotherapy consisted of two courses of intravenous infusion of 5-FU (1000 mg/m²/d) on days 1–4, cisplatin (75 mg/m²) on day 1, and nivolumab (240 mg/body) on days 1 and 15, administered concurrently with radiotherapy. Each course was repeated every 28 days, provided the patient had adequately recovered from treatment-related toxicities.

Radiotherapy was delivered with megavoltage equipment (6–10 MV) using a multiple-field technique based on three-dimensional computed tomography (CT) simulation. Intensity-modulated radiotherapy was permitted when considered appropriate. A total dose of 50.4 Gy was delivered in 28 fractions. The gross tumour volume included the primary tumours and involved metastatic lymph nodes. The clinical target volume encompassed the primary tumour with a 2 cm craniocaudal margin, metastatic lymph nodes, and regional lymph nodes. Regional lymph nodes associated with oesophageal tumours were defined by the tumour's location. For tumours in the upper thoracic oesophagus, regional nodes included the bilateral supraclavicular fossae and superior mediastinal lymph nodes. When the tumour was located in the middle or lower thoracic oesophagus, regional nodes included the mediastinal and perigastric lymph nodes. For tumours in the lower thoracic oesophagus, regional lymph nodes included those near the celiac axis. The planning target volume was defined as the clinical target volume plus a 0.5–1 cm lateral margin and a 1–2 cm craniocaudal margin to account for respiratory motion and daily set-up variations. After delivering 41.4 Gy to the planning target volume, a boost dose of 9.0 Gy was administered to reduce the planning target volume encompassing the primary tumour and metastatic lymph nodes, bringing the total dose to 50.4 Gy. Dose constraints for organs at risk were as follows: the spinal cord volume receiving >48 Gy was limited to ≤1 mL, the percentage of lung receiving >20 Gy was limited to ≤30%, and the mean dose to the heart was limited to <40 Gy.

After the initial response assessment of immuno-chemoradiotherapy, patients received two additional cycles of immunochemotherapy consisting of 5-FU, cisplatin, and nivolumab in a similar manner. Maintenance therapy with nivolumab (240 mg every 2 weeks or 480 mg every 4 weeks) was continued for up to 1 year or until disease progression, unacceptable toxicity, investigator decision, or patient withdrawal of consent. Dose reduction for nivolumab was not permitted.

Toxicities were assessed based on the Common Terminology Criteria for Adverse Events (version 5.0).¹⁵ Distinguishing between radiation and immune-related pneumonitis is challenging, because both aetiologies can coexist and often present with overlapping radiographic findings. Therefore, pneumonitis events were classified as treatment-related pneumonitis. All grade ≥3 pneumonitis cases were independently reviewed and confirmed by the central efficacy and safety evaluation committee of the study.

Tumour response was evaluated by CT according to the Response Evaluation Criteria in Solid Tumours (version 1.1).¹⁶ Primary tumour response was assessed by endoscopy according to modified criteria from the 11th edition of the Japanese Society of Esophageal Diseases guidelines.¹⁷ Endoscopic biopsy for response evaluation was performed using standard biopsy forceps at the most representative site of the residual or scar lesion, avoiding necrotic or ulcerated areas, in accordance with the study protocol. A complete response required confirmation at least 4 weeks after the initial response evaluation. Response assessments were performed every 4 weeks after completing definitive chemoradiotherapy until complete response was confirmed. All CT and endoscopic images and findings used for response evaluation were documented and verified by the central review committee. After confirmation of complete response, surveillance CT and endoscopy were repeated every 3 months during the first 3 years, and every 6 months thereafter. Patients with recurrence during follow-up were referred for salvage treatment or palliative chemotherapy.

Two biopsy specimens (approximately 2 mm each) were collected using endoscopic biopsy forceps prior to treatment and between the first and second cycles of chemotherapy, avoiding necrotic lesions. All biopsy samples were immediately placed in RNAlater solution (Thermo Fisher Scientific, Rockford, IL, USA) and stored at −80 °C until analysis.

RNA was extracted from both biopsy specimens and cell lines and subjected to microarray analysis following quality assessment using the RNA 6000 Nano LabChip Kit (Agilent Technologies Ltd., Palo Alto, CA, USA). Total RNA was biotin-labelled and hybridised to high-density oligonucleotide microarrays (Human Genome U133PLUS2.0 Array; Affymetrix, Santa Clara, CA, USA). The arrays were scanned using the GeneChip Scanner 3000 (Affymetrix), and feature intensities were calculated using GeneChip Analysis Suite software (version 4.0; Affymetrix). Signal intensities for each probe set were normalised using the Microarray Suite 5.0 algorithm.

Biomarker analysis focused on 51 immune-related genes shown in Supplementary Table S1. These genes were selected as potentially related to immune responses for exploratory analysis. This selection was based on a list of 234 genes identified as cytotoxic T lymphocyte activation genes in previously study.¹⁴ This gene set includes, for example, genes encoding immune checkpoint molecules, IFNγ, and IFNγ receptors. The reason for this selection is provided in Supplementary Table S1. A heat map was generated to visualise the expression levels of these 51 immune-related genes. Clustering was performed to identify a distinctive immunological subtype. The log-transformed expression levels of 51 genes were reduced to two dimensions using Uniform Manifold Approximation and Projection (UMAP), followed by clustering with Density-Based Spatial Clustering of Applications with Noise.

Outcomes

The primary endpoint of this study was safety, defined as the incidence of grade ≥4 non-haematological toxicity ≤10% and the incidence of grade ≥3 pneumonitis ≤15%. The incidence of grade ≥3 radiation pneumonitis associated with definitive chemoradiotherapy for oesophageal cancer is reportedly approximately 5%.⁶ Although the cancer type differs, a phase 2 trial (NICOLAS trial) evaluating nivolumab in combination with concurrent chemoradiotherapy non-small cell lung cancer (NSCLC) reported a grade 3 pneumonitis rate of 10.3%. Given that the lung dose constraints in the present trial were consistent with those used in the NICOLAS trial, and that oesophageal cancer typically involves a broader radiation field across both lungs compared with NSCLC, the incidence of ≥3 pneumonitis was anticipated to possibly exceed 10%. Therefore, the threshold for grade ≥3 pneumonitis in this trial was set at 15%. The secondary endpoints included overall survival, progression-free survival, major progression-free survival, and complete response rate. Overall survival was calculated from the date of treatment initiation to the date of death from any cause, with censoring at the last follow-up date for surviving patients. Progression-free survival was defined as the period from the date of treatment initiation to the first recurrence, disease progression, or death from any cause, with censoring at the last assessment for patients without progression. Major progression-free survival was defined as the time from treatment initiation to the disease progression or death, excluding cases of successful curative resection through salvage endoscopic resection.

Statistics

The sample size was primarily determined based on feasibility. A target of 20 patients per group was set to allow for sufficient evaluation of safety. Assuming a sample size of 20 patients, the safety threshold of 15% of grade ≥3 pneumonitis corresponds to fewer than three cases of adverse events. A post-hoc evaluation using a simulation study showed that assuming the true incidence rate of grade ≥3 pneumonitis is less than 15%, the probability of fewer than 3 cases of pneumonia occurring is 77.5%, which was considered clinically acceptable for feasibility assessment. Furthermore, because the detection rate of the immune-related biomarker subtype was estimated to be approximately 20%, the total planned sample size of 60 patients was expected to allow an exploratory evaluation of treatment efficacy within this biomarker-defined subgroup.

Proportions and their 95% CIs were calculated using the Wilson's score method, and the continuity correction was applied when expected frequencies were small. Overall survival and progression-free survival were estimated using the Kaplan–Meier method, and 95% CIs were calculated with Greenwood's formula.

Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation; writing of the report, or decision to submit the paper for publication.

Results

Between January 2019 and September 2021, 42 patients were enrolled from five institutions. A patient flow diagram is shown in Fig. 1. All 42 patients underwent definitive immune-chemoradiotherapy. One patient discontinued the protocol treatment before any response evaluation due to a treatment-related adverse event and was therefore excluded from the full analysis set in accordance with the prespecified analysis plan. Among the 41 patients in the full analysis set, 25 had operable disease and 16 had inoperable disease. This approach ensured that efficacy analyses were based only on patients who completed at least one post-baseline tumour assessment, consistent with the standard full analysis set definitions in clinical trials. Enrolment for the operable group was completed, whereas enrolment for the inoperable group was terminated early due to slow accrual. Given that the overall target sample size (n = 60) was not achieved and the trial was prematurely terminated, the statistical power for subgroup and biomarker analyses was limited. At the data cutoff for this analysis (Oct 13 2023), the median follow-up duration was 28.7 months (IQR 21.7–36.5). All patients were followed up for at least 12 months after enrolment, allowing for a reliable estimation of 1-year survival outcomes. Patient characteristics are summarised in Table 1.

Fig. 1.

Patient disposition.

Table 1.

Patient characteristics.

	The efficacy analysis set n = 41	The safety analysis set N = 42
Age (y)
Median (interquartile range)	65 (58–70)	65 (59–70)
Sex
Male	36	37
Female	5	5
ECOG PS
0	35	36
1	6	6
Location of primary tumour
Upper thoracic	9	9
Middle thoracic	20	20
Lower thoracic	12	13
Histologic differentiation
Well-diff	6	6
Moderate-diff	13	13
Unknown	22	23
BMI (kg/m2)
Median (interquartile range)	23.3 (21.3–24.8)	23.3 (21.2–24.9)
Alb (g/dL)
Median (interquartile range)	4.0 (3.8–4.2)	4.0 (3.8–4.2)
Clinical T stage (UICC8th)
T1a	0	0
T1b	6	6
T2	6	6
T3	22	23
T4a	0	0
T4b	7	7
Clinical N stage (UICC8th)
N0	6	6
N1	20	21
N2	10	10
N3	5	5
Clinical M stage (UICC8th)
M0	32	33
M1	9	9
Clinical TNM (UICC8th)
I (T1N0M0)	0	0
I (T1N1M0)	6	6
II	6	6
III	13	14
IVA	7	7
IVB	9	9
Resectability
Operable	25	26
Inoperable	16	16

Abbreviations: ECOG PS, Eastern Cooperative Oncology Group performance status; BMI, body mass index; Alb, albumin; UICC, Union for International Cancer Control; TNM: tumor-node-metastasis.

Of the 42 eligible patients, 40 completed definitive immune-chemoradiotherapy, which consisted of two cycles of immunochemotherapy plus radiotherapy (total dose of 50.4 Gy); 35 patients completed two additional cycles of immunochemotherapy.

Two patients discontinued definitive immune-chemoradiotherapy due to adverse events: one developed erythema multiforme, and the other experienced gastrointestinal bleeding. Four patients skipped at least one cycle of immunochemotherapy due to adverse events, and one patient discontinued treatment due to disease progression. 36 patients received nivolumab maintenance therapy, of whom 20 (56%) completed therapy for up to 1 year. Reasons for discontinuation of nivolumab maintenance included disease progression (n = 9), adverse events (n = 4), and investigator decision (n = 3). The median relative dose intensity was 92% (IQR 83–96) for nivolumab, 92% (IQR 84–95) for cisplatin, and 92% (IQR, 84–97) for 5-FU.

The most common adverse events of any grade were oesophagitis (n = 39 [93%]), constipation (n = 37 [88%]), lymphopenia (n = 37 [88%]), decreased white blood count (n = 36 [86%]), and pneumonitis (n = 33 [79%]; Table 2). The most common grade ≥3 adverse events were lymphopenia (n = 36 [86%]), decreased white blood count (n = 18 [43%]), and neutropenia (n = 16 [38%]). Among the non-haematological grade ≥3 adverse events, oesophagitis was the most common (n = 8 [19%]). No treatment-related deaths were observed. Grade 3 pneumonitis occurred in two patients (5%), which met the predefined safety threshold.

Table 2.

Adverse events.

	Grade 1	Grade 2	Grade 3	Grade 4	Total	≥Grade 3
White blood cell decreased	0	18	18	0	86% (36/42)	43% (18/42)
Anaemia	11	13	7	0	74% (31/42)	17% (7/42)
Lymphocyte decreased	0	1	7	29	88% (37/42)	86% (36/42)
Neutrophile decreased	0	13	14	2	69% (29/42)	38% (16/42)
Platelet cell decreased	8	9	4	2	55% (23/42)	14% (6/42)
Febrile neutropenia	0	0	3	0	7% (3/42)	7% (3/42)
ALT increased	6	3	0	0	21% (9/42)
AST increased	9	1	0	0	24% (10/42)
Cholesterol increased	5	0	0	0	12% (5/42)
Creatinine increased	12	1	0	0	31% (13/42)
LDH increased	5	0	0	0	12% (5/42)
Hypothyroidism	1	5	0	0	14% (6/42)
TSH increased	5	0	0	0	12% (5/42)
γ-GTP increased	14	2	1	0	41% (17/42)	2% (1/42)
Hypertriglyceridemia	4	1	0	0	12% (5/42)
Hyperuricaemia	5	0	0	0	12% (5/42)
Hypoalbuminaemia	15	5	0	0	48% (20/42)
Hypokalaemia	7	0	2	0	21% (9/42)	5% (2/42)
Hyponatraemia	13	2	1	0	38% (16/42)	2% (1/42)
Constipation	23	13	1	0	88% (37/42)	2% (1/42)
Diarrhoea	5	3	0	0	19% (8/42)
Nausea	18	8	3	0	69% (29/42)	7% (3/42)
Oesophagitis	2	29	8	0	93% (39/42)	19% (8/42)
Stomatitis oral	7	8	0	0	36% (15/42)
Vomiting	5	1	1	0	17% (7/42)	2% (1/42)
General fatigue	8	1	1	0	24% (10/42)	2% (1/42)
Fever	4	1	0	0	12% (5/42)
Injection site extravasation	8	3	0	0	26% (11/42)
Radiation dermatitis	15	1	0	0	38% (16/42)
Weight loss	6	10	1	0	41% (17/42)	2% (1/42)
Anorexia	14	7	5	0	62% (26/42)	12% (5/42)
Dysgeusia	7	1	0	0	19% (8/42)
Peripheral sensory neuropathy	4	3	0	0	17% (7/42)
Insomnia	9	0	0	0	24% (9/42)
Protein urea	5	2	0	0	17% (7/42)
Cough	2	4	0	0	14% (6/42)
Hiccups	7	17	1	0	60% (25/42)	2% (1/42)
Pneumonitis	22	9	2	0	79% (33/42)	5% (2/42)
Alopecia	8	0	0	0	19% (8/42)
Pruritus	5	0	0	0	12% (5/42)
Maculopapular	7	4	0	0	26% (11/42)
Phlebitis	0	8	0	0	19% (8/42)

Abbreviations: ALT, alanine transaminase; AST, aspartate transaminase; LDH, lactate dehydrogenase; TSH, thyroid-stimulating hormone; GTP, glutamyl transferase.

Pneumonitis occurred during immune-chemoradiotherapy in two patients, during additional immune-chemotherapy in six of 38 patients, and during nivolumab maintenance in 19 of 36 patients. The median duration of pneumonitis was 4.1 months (IQR 3.6–5.3) from the start of treatment.

24 (58%) of 41 patients had a complete response after definitive immune-chemoradiotherapy plus two cycles of additional immunochemotherapy, and an additional six patients subsequently had a complete response during the post-immune-chemoradiotherapy phase, resulting in a best overall complete response rate of 73% (30 of 41 patients; 95% CI 58–84), with a median time to confirmed complete response of 130 days (IQR 113–189 days; Fig. 2). The complete response rate was 84% (21 of 25 patients; 95% CI 65–94) in the operable group and 56% (9 of 16 patients; 95% CI 33–77) in the inoperable group. The median duration of complete response was 23.6 months (IQR 15.3–32.0).

Fig. 2.

Best overall response following immune-chemoradiotherapy. The complete response (CR) is shown for all patients, stratified by resectability status.

Overall survival data for all 41 patients are shown in Fig. 3B and C. 1-year overall survival was 92.7% (95% CI 79.0–97.6; Fig. 3B). By resectability, 1-year overall survival was 100.0% in the operable group and 81.3% (95% CI 52.5–93.5) in the inoperable group (Fig. 3C).

Fig. 3.

Summary of clinical outcomes and biomarker analysis. (A) Swimmer plots showing duration of tumour control in individual patients. Each bar represents the duration of tumour control for a single patient. Arrows indicate ongoing tumour control at the time of analysis. (B) Kaplan–Meier curve showing overall survival (OS) for all 41 patients. (C) Kaplan–Meier curves showing OS stratified by resectability (operable vs. inoperable cases). (D) Kaplan–Meier curve showing progression-free survival (PFS) for all 41 patients. (E) Kaplan–Meier curves illustrating PFS stratified by resectability (operable vs. inoperable cases). (F) Heat map of the immune-related gene expression in 42 patients with oesophageal squamous cell carcinoma (OSCC). Each row represents a gene, and each column represents a patient. (G) Clustering results after dimensionality reduction of the 51 immuno-related genes using Uniform Manifold Approximation and Projection (UMAP). The red cluster corresponds to the five leftmost subjects highlighted in black in panel A. Based on the expression levels, patients were categorized into immune “high-active” or immune “moderate-active” intrinsic subtypes.

1-year progression-free survival was 65.4% (95% CI 48.6–77.9; Fig. 3D and E). By resectability, 1-year progression-free survival was 79.6% (95% CI 57.7–91.0) in the operable group and 43.8% (95% CI 19.8–65.6) in the inoperable group.

1-year major progression-free survival was 67.9% (95% CI 51.2–80.0). By resectability, 1-year major progression-free survival was 83.8% (95% CI 62.4–93.6) in the operable group and 43.8% (95% CI 19.8–65.6) in the inoperable group.

Nine patients received salvage therapy. Five of the 11 patients who did not have a complete response received salvage surgery. Of the 30 patients who had a complete response, four developed recurrence and received salvage treatment, two patients received salvage surgery, and two patients received salvage endoscopic submucosal dissection. No treatment-related deaths occurred after salvage surgery. Ten patients were treated with chemotherapy after recurrence or progression: eight with paclitaxel, one with nivolumab plus ipilimumab, and one with nivolumab monotherapy.

Fig. 3F and G shows a heat map (Fig. 3F) of immune-related gene expression in 42 patients with OSCC and the clustering results (Fig. 3G) after dimensionality reduction of 51 genes to two dimensions using UMAP. The clustering results identified two distinctive clusters (Fig. 3G). These clusters corresponded to a group with high gene expression levels on the heat map and the remaining population (Fig. 3F). Based on these findings, we defined the two groups as immune high-active (n = 5) or immune moderate-active intrinsic subtype (n = 37). Because many genes were highly expressed, there was no immune low-active subtype. Consistent with these clusters, all four patients classified as immune “high-active” before treatment and for whom efficacy assessment was possible had a complete response (95% CI 40–98; one patient was non-evaluable), whereas those classified as “moderate-active” had a complete response rate of 70% (26 of 37 patients; 95% CI 54–83).

Discussion

In this phase 2 trial, nivolumab combined with definitive chemoradiotherapy had high complete response rates and favourable survival outcomes, with a low incidence of severe pneumonitis in patients with both operable and inoperable OSCC. These results suggest that adding ICI to definitive-intent chemoradiotherapy may enhance efficacy without compromising safety.

In the operable cohort, the complete response rate was 84% and 1-year overall survival was 100%. These outcomes exceed historical benchmarks. For example, the JCOG0909 trial reported a complete response rate of 58.5% and 1-year overall survival of 92.5%.⁶ Given that the treatment protocol was identical to JCOG0909 in radiation dose and target delineation, the enhanced outcomes are likely attributable to the immunotherapeutic component rather than radiation intensification. In the inoperable setting, the complete response rate was 56% and 1-year overall survival was 81.3%. These results were markedly higher than those reported in the JCOG0303 trial (complete response rate 0–1.4%; 1-year overall survival 55.9–56.3%), despite using a low radiation dose (50.4 Gy vs. 60 Gy).⁷ These results indicate a potential benefit of adding nivolumab to chemoradiotherapy even in patients with operable OSCC. However, this interpretation is limited by the relatively short follow-up period and small sample size. To confirm this efficacy, it must be validated in larger, randomized controlled trial.

The incidence of grade ≥3 pneumonitis was only 5% overall and 8% in operable cases, compared with 2.2% in JCOG0909 under the same radiation conditions. Given previous reports of treatment-related deaths from pulmonary toxicity with nivolumab in the ATTRACTION-3 and CheckMate-648 trials,^9,10 severe pneumonitis was established as a primary safety end-point. Importantly, the addition of nivolumab did not increase the risk of pneumonitis, and other non-haematological toxicities were comparable with previous chemoradiotherapy studies. However, haematologic toxicities, particularly lymphopenia (grade ≥3 in 86% of patients), were frequently observed. These events were consistent with prior chemoradiotherapy regimens using cisplatin and 5-FU but may have been influenced by concurrent immunotherapy. Importantly, most haematologic toxicities were reversible and manageable with standard supportive care, and no treatment-related mortality was observed. The safety of salvage oesophagectomy after immune-chemoradiotherapy was not prospectively defined or centrally adjudicated in this trial, and perioperative morbidity/mortality endpoints were not pre-specified. Although no treatment-related deaths occurred, the effect of prior immune-chemoradiotherapy on surgical risks remains uncertain and warrants a dedicated evaluation. In addition, a subset of patients in both the operable and inoperable cohorts underwent salvage and conversion surgery, which may have contributed to the prolonged overall survival. However, given the limited number of such cases and the absence of prespecified analyses adjusting for these post-protocol interventions, the potential impact of conversion surgery on survival outcomes remains uncertain. Therefore, the potential confounding effect of salvage and conversion surgery on overall survival should be acknowledged as a limitation of this study.

Some previous studies of definitive chemoradiotherapy enrolled both operable and inoperable patients without stratification. However, our trial prospectively categorised patients by resectability, recognising its impact on prognosis and treatment efficacy. Previous studies have consistently shown that patients reaching clinical complete response after definitive chemoradiotherapy have significantly better survival outcomes compared with those with residual disease.18, 19, 20 In our study, confirmation of complete response at the primary site required two consecutive assessments by both imaging and endoscopic biopsy separated by at least 4 weeks. By contrast, most previous studies did not apply such stringent criteria, which may have led to overestimation of complete response rates in those studies.

The JCOG0303 trial used the same concurrent chemotherapy regimen (cisplatin plus 5-FU) as our trial, but administered a higher total radiation dose of 60 Gy compared with the dose in our trial of 50.4 Gy. Previous studies have shown no overall survival benefit from increasing the dose beyond 50 Gy for oesophageal cancer.^5,21, 22, 23 Despite this lower dose, our trial had favourable outcomes, suggesting a potential additive effect of nivolumab. Even though some research indicates that higher radiation doses may be beneficial for inoperable OSCC,²⁴ our findings suggest that nivolumab can enhance the therapeutic effects of chemoradiotherapy without dose escalation in patients with inoperable disease.

Compared with the EC-CRT-001 trial,²⁵ which is the only other prospective trial investigating definitive chemoradiotherapy combined with a PD-1 inhibitor for OSCC, our study had a higher complete response rate (73% vs. 62%), 1-year progression-free survival (65.4% vs. 54.5%), and 1-year overall survival (92.7% vs. 78.4%). Although the clinical tumour-node-metastasis stage distribution and cT4 frequency were similar between the EC-CRT-001 trial and our trial, differences in treatment protocols, such as the use of elective nodal irradiation and administration of post-chemoradiotherapy chemotherapy after chemoradiotherapy, might have influenced the outcomes. While these factors have not shown a significant impact on chemoradiotherapy outcomes,26, 27, 28, 29 they may influence efficacy in combination with PD-1 blockade.

Biomarker analysis identified an immune high-active intrinsic subtype, defined by the high expression of 51 immune-related genes, with a 100% (4 of 4 patients; 95% CI 40–98) complete response rate compared with 70% (26 of 37 patients; 95% CI 54–83) in the moderate-active subtype. However, given that only four patients were classified as the immune high-active subtype, these findings should be viewed as preliminary. Moreover, the observed association between gene expression patterns and clinical outcomes requires further validation due to the small sample size. The results nevertheless suggest that the genes identified in a previous study¹⁴ may be compatible with tumour immune activity, but their clinical relevance requires confirmation in larger, prospective cohorts.

PD-L1 expression was not evaluated in this study because our biomarker program primarily focused on gene expression-based immune signatures rather than protein-level markers. Moreover, the biomarker findings have not yet been validated in an independent cohort, further limiting their generalisability. Future studies incorporating PD-L1 and other immune markers are warranted to confirm and expand these findings. In addition, more detailed immunological analyses are ongoing and will be reported together with the 3-year follow-up results.

As these biomarkers were designed to assess tumour immune activity associated with therapeutic efficacy, they were not intended to evaluate safety or toxicity. Therefore, biomarker analyses were not performed in relation to primary safety endpoints.

This trial had several limitations. The inoperable group was terminated early due to slow accrual, and the small sample size, single-arm design, and relatively short follow-up reduced statistical power and generalizability. Because no randomized control arm was included, the efficacy of nivolumab cannot be definitively determined, and the efficacy results should be regarded as exploratory. Furthermore, the inclusion of both operable and inoperable patients introduced clinical heterogeneity, making subgroup comparisons descriptive rather than confirmatory. PD-L1 expression was not assessed, which may have provided further predictive information. Randomised controlled trials are needed to determine the significance of adding ICI to definitive chemoradiotherapy.

In conclusion, nivolumab combined with definitive chemoradiotherapy was feasible and showed acceptable safety in patients with OSCC. The high complete response rate in both operable and inoperable groups suggested that this combination treatment was a promising treatment option. However, these results were preliminary and should be interpreted with caution due to the small sample size and premature termination of the trial. These results should be validated in larger, randomized trials to establish their clinical significance.

Contributors

Conceptualisation and design: MN and MM. MN, KS, CK, AW, YA, KM, KK, TK, and MM had a major contribution in patient accrual. Data collection: MN, KS, JS, SO, CK, AW, YA, KM, KK, TK, and MM. Data curation and verification: IK and HT. Data analysis: KN and AN. MN, KS, JS, SO, CK, AW, YA, KM, KK, TK, KN, IK, HT, AN, and MM had access to the data and were responsible for the interpretation of data and writing the manuscript, as well as reviewing and approving the manuscript for submission. MN, KS, JS, SO, CK, AW, YA, KM, KK, TK, KN, IK, HT, AN, and MM accept responsibility for the decision to submit the manuscript for publication.

Data sharing statement

Individual participant data are not publicly available, because this requirement was not anticipated in the study protocol and sharing was not included in the ethical approvals.

Declaration of interests

MN has received honoraria from Bristol Myers Squibb, Ono Pharmaceutical, and MSD. KS has received Speakers’ Bureau from MSD, Ono Pharmaceutical; and Travel, Accommodations, Expenses from MSD, Ono Pharmaceutical. KM has received research funding for institution from Amgen, MSD, Daiichi Sankyo, BeiGene, Inc., and Taiho Pharmaceutical Co., Ltd. KK has received consulting fees from Bristol Myers Squibb, Merk Sharp & Dohme, BeiGene, Roche, AstraZeneca, and Bayer; honoraria from Ono Pharmaceutical, Bristol Myers Squibb, and Taiho pharmaceutical; and research funding outside the submitted work from Ono Pharmaceuticals, Bristol Myers Squibb, Merk Sharp & Dohme, BeiGene, Chugai, Shionogi, AstraZeneca, and Bayer. TK has received grants from Beigene Ltd., Astra Zeneca, Chugai Pharmaceutical, Parexel International, Taiho Pharmaceutical, Astellas Amgen BioPharma, MSD, and Ono Pharmaceutical; honoraria from Ono Pharmaceutical, Covidien Japan, MSD, Nippon Boehringer Ingelheim Co., Ltd., Kyowa Kirin, EA Pharma, Bristol Myers Squibb, Taiho Pharmaceutical, Japanese Society of Pharmaceutical Health Care and Sciences, and The Japan Esophageal Society; and advisory roles for Beigene Ltd., Nippon Boehringer Ingelheim Co., Ltd. and MSD. KN has received consulting fees from Senju Pharmaceutical, Toray Industries, and Kowa Company. MM has received grants from Ono Pharmaceutical, Chugai Pharmaceutical, Taiho Pharma, AstraZeneca, Takeda Pharmaceutical and Meiji Seika Pharma and has received honoraria from Bristol Myers Squibb, MSD, Chugai Pharmaceutical, Meiji Seika Pharma, Daiichi Sankyo Pharmaceutical and Ono Pharmaceutical. JS, SO, CK, AW, YA, IK, HT, and AN have nothing to disclose.