Camrelizumab plus nimotuzumab as second-line therapy for advanced esophageal squamous cell carcinoma: a multicenter phase II study

Abstract

Background

Therapeutic options for advanced esophageal squamous cell carcinoma (ESCC) after first-line failure remain limited, particularly in patients previously exposed to immune checkpoint inhibitors (ICIs). We evaluated the efficacy, safety, and exploratory biomarker correlates of camrelizumab, an ICI, combined with nimotuzumab, an anti-epidermal growth factor receptor (EGFR) monoclonal antibody, as second-line therapy for ESCC.

Methods

In this multicenter, single-arm, phase II study, patients with advanced ESCC who progressed after first-line therapy received camrelizumab (200 mg every 2 weeks) plus nimotuzumab (400 mg weekly). The primary endpoint was the objective response rate (ORR).

Results

Between November 2021 and December 2024, 46 patients were enrolled. The confirmed ORR was 32.6% (15/46; 95% CI 19.5 to 48.0) and the disease control rate was 82.6% (38/46; 95% CI 68.6 to 92.2). Median progression-free survival (PFS) was 9.13 months (95% CI 5.95 to 9.76), and median overall survival (OS) was 12.62 months (95% CI 9.40 to 15.01). Clinical activity was observed across subgroups, including immunotherapy-naïve and previously treated patients (ORR 32.0% vs 33.3%). Patients with EGFR amplification demonstrated a higher ORR (47.1% vs 24.0%) and longer median OS (13.17 vs 9.99 months). Among patients with M1 disease (n=39), the ORR was 30.8% (95% CI 17.0 to 47.6), the median PFS was 8.48 months (95% CI 5.95 to 9.59), and the median OS was 12.55 months (95% CI 7.95 to 14.88). Treatment-related adverse events occurred in 80.4% of patients, with grade ≥3 events in 8.7%. Exploratory analyses suggested that MUC16 mutations were associated with lower ORR (8.3% vs 46.4%, p=0.030), NOTCH3 mutations with prolonged survival (median PFS not reached vs 8.21 months, HR 0.20, p=0.015; median OS not reached vs 10.58 months, HR 0.22, p=0.026), and MTAP deletions with shorter PFS (3.71 vs 9.49 months, HR 3.18, p=0.005).

Conclusions

Camrelizumab combined with nimotuzumab demonstrated encouraging antitumor activity and a manageable safety profile as second-line therapy for advanced ESCC.

Trial registration number

NCT03766178.

WHAT IS ALREADY KNOWN ON THIS TOPIC.

WHAT THIS STUDY ADDS

• This multicenter phase II study demonstrates that combining camrelizumab with the anti-EGFR antibody nimotuzumab yields encouraging antitumor activity and a manageable safety profile in second-line ESCC, including in patients with prior immunotherapy exposure.

• Exploratory analyses also identify potential biomarker signals, suggesting that EGFR amplification, NOTCH3, MUC16, and MTAP alterations may be associated with treatment outcomes.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings support further investigation of EGFR-targeted and immunotherapy combinations in ESCC and highlight the potential role of molecular stratification to guide treatment selection.

• Prospective randomized trials are warranted to validate efficacy and confirm the predictive value of the identified biomarkers.

Introduction

Esophageal cancer remains a major global health burden, ranking as the 11th most common malignancy and the seventh leading cause of cancer-related mortality worldwide. In China, the disease burden is particularly substantial, with an estimated 224 000 new cases and 187 500 deaths reported in 2022.^{1 2} Esophageal squamous cell carcinoma (ESCC) is the predominant histological subtype in this region and is frequently associated with non-specific early symptoms, resulting in delayed diagnosis. Consequently, the majority of patients present with unresectable or metastatic disease and are ineligible for curative treatment.^{3 4} Although immune checkpoint inhibitors (ICIs) have been incorporated into first-line therapy, treatment options after disease progression remain limited and clinically challenging. In the second-line setting, single-agent ICIs represent a commonly used strategy; however, their efficacy remains modest, with reported objective response rates (ORRs) of approximately 20%, median progression-free survival (PFS) of approximately 2 months, and median overall survival (OS) of around 7–11 months in pivotal trials.^5–9 Importantly, these outcomes were largely derived from immunotherapy-naïve populations and may not be generalizable to patients previously exposed to ICIs. Indeed, among patients who progress on prior immunotherapy, response rates decline substantially, with reported ORRs as low as 10%.¹⁰ Thus, the development of more effective therapeutic strategies for this population represents an urgent unmet clinical need.

The epidermal growth factor receptor (EGFR) signaling pathway plays a critical role in tumor proliferation, invasion, and survival, particularly in tumors harboring EGFR amplification.^{11 12} Monoclonal antibodies targeting EGFR inhibit ligand-dependent receptor activation, thereby suppressing downstream oncogenic signaling and promoting tumor cell apoptosis.^13–15 In parallel, ICIs restore antitumor immunity by reversing T-cell exhaustion. Given these complementary mechanisms, combining EGFR inhibition with programmed death-1 (PD-1) blockade represents a rational therapeutic strategy with the potential for synergistic antitumor activity. This concept has been supported by clinical evidence in squamous cell carcinomas of the head and neck, where pembrolizumab plus cetuximab demonstrated an ORR of 45%, exceeding that observed with either agent alone.^{16 17} Similarly, in metastatic oral squamous cell carcinoma, the combination of nimotuzumab and tislelizumab achieved an ORR of approximately 40%.¹⁸

Nimotuzumab is a humanized anti-EGFR monoclonal antibody characterized by intermediate affinity and favorable safety and has been approved for the treatment of ESCC in multiple countries. Camrelizumab, an anti-PD-1 antibody, is approved as second-line therapy for advanced ESCC. However, clinical evidence supporting the combination of EGFR-targeted therapy and ICIs in this setting remains limited. Therefore, we conducted this multicenter phase II study to evaluate the efficacy, safety, and potential biomarkers of camrelizumab in combination with nimotuzumab as second-line treatment for advanced ESCC.

Methods

Study design and patients

This multicenter, single-arm, phase II trial (ClinicalTrials.gov identifier: NCT03766178) was conducted across five centers in China (online supplemental table 1). Eligible patients were aged 18–75 years with histologically confirmed advanced or metastatic ESCC that had progressed following first-line systemic therapy. Additional inclusion criteria comprised an Eastern Cooperative Oncology Group (ECOG) performance status of 0–1, the presence of at least one measurable lesion according to the Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1, and adequate hematological, hepatic, and renal function. Key exclusion and eligibility criteria are detailed in the study protocol (online supplemental file 1).

Treatment

Patients received intravenous camrelizumab at a fixed dose of 200 mg every 2 weeks in combination with nimotuzumab 400 mg administered weekly. Treatment was continued until radiologically confirmed disease progression, unacceptable toxicity, or completion of a maximum duration of 24 months. Dose selection for both agents was based on prior clinical studies.^{5 18} Dose reductions were not permitted; however, temporary treatment interruption was allowed in the event of treatment-related adverse events (TRAEs), according to predefined protocol criteria.

Assessments and endpoints

Tumor response was assessed every 4 weeks during the first 3 months and every 6 weeks thereafter until disease progression or treatment discontinuation. Imaging evaluations were performed using contrast-enhanced CT or MRI and were reviewed according to the RECIST V.1.1. Objective responses required confirmation by repeat imaging at least 4 weeks after the initial documentation. After treatment discontinuation, survival status was assessed every 3 months until death or withdrawal of consent. Adverse events were graded in accordance with the National Cancer Institute Common Terminology Criteria for Adverse Events, V.5.0.

The primary endpoint was investigator-assessed ORR, defined as the proportion of patients achieving a confirmed complete response or partial response. Secondary endpoints included OS, defined as the time from treatment initiation to death from any cause; PFS, defined as the time from treatment initiation to disease progression or death, whichever occurred first; disease control rate (DCR), defined as the proportion of patients achieving complete response, partial response, or stable disease; duration of response (DOR), defined as the time from first documented response to disease progression or death; and time to response (TTR), defined as the interval from treatment initiation to first documented response; and safety.

Next-generation sequencing

Genomic DNA was extracted from formalin-fixed, paraffin-embedded (FFPE) tumor specimens and matched peripheral blood leukocytes using the QIAamp DNA FFPE Tissue Kit and DNeasy Blood and Tissue Kit (Qiagen), respectively. Whole-exome sequencing was performed using the Target Cap Core Exome Panel V.3.0 (Boko Biotechnology) according to the manufacturer’s protocol. Sequencing data were processed and analyzed as previously described.¹⁹ Tumor mutational burden (TMB) was defined as the number of non-synonymous somatic mutations per megabase of coding genome. Tumor purity was estimated using ABSOLUTE,²⁰ and purity-adjusted data were used to infer gene-level copy number variations (CNVs) with FACETS.²¹ EGFR amplification was determined based on copy number status using a predefined five-tier classification system incorporating total copy number and tumor ploidy.²² Chromosomal instability was quantified as the proportion of the genome exhibiting segmented copy number alterations. EGFR amplification was defined as a CNV level ≥1 (ie, 1 ≤total copy number—ploidy <3), whereas all other cases were classified as non-amplified. Intratumoral heterogeneity was assessed using the PyClone algorithm. Additional details regarding molecular profiling and biomarker analyses are provided in the online supplemental methods.

Gene set enrichment analysis and immune infiltrate analysis

Transcriptomic and genomic data from patients with ESCC (n=95) were obtained from The Cancer Genome Atlas (TCGA) Firehose Legacy cohort via the cBioPortal platform (https://www.cbioportal.org/study/summary?id=esca_tcga). Differential gene expression analysis was performed using the DESeq2 R package, based on a negative binomial distribution model. Gene set enrichment analysis (GSEA) was conducted using the clusterProfiler package (V.4.2.2), with the Hallmark gene set collection (‘h.all.v2024.1.Hs.symbols.gmt’) as the reference. Enriched pathways were considered statistically significant at an adjusted p value ≤0.05 and an absolute normalized enrichment score >1. The relative abundance of tumor-infiltrating immune cell populations was estimated using the QUANTISEQ algorithm.²³

Programmed death-ligand 1 expression

Programmed death-ligand 1 (PD-L1) expression was assessed in FFPE tumor specimens using immunohistochemistry. PD-L1 expression in tumor cells was quantified as the percentage of viable tumor cells exhibiting partial or complete membrane staining, evaluated in a minimum of 100 tumor cells. The combined positive score (CPS) was calculated as the number of PD-L1-staining tumor and immune cells divided by the total number of viable tumor cells, multiplied by 100. PD-L1 positivity was defined as a CPS ≥1.

Statistical analysis

Sample size was determined using Simon’s optimal two-stage design,²⁴ with a one-sided α of 0.05 and β of 0.10 (power, 90%). The null hypothesis assumed an ORR of 15%, based on historical controls,^{5 6 9} whereas the alternative hypothesis specified an ORR of 35% for the combination regimen. In the first stage, 23 patients were enrolled, and continuation to the second stage required at least three objective responses. An additional 15 patients were subsequently enrolled, yielding a planned sample size of 38 evaluable patients. Assuming a 10% dropout rate, a total enrollment of 42 patients was planned. Given the potential for increased attrition during the COVID-19 pandemic, four additional patients were enrolled to ensure an adequate number of evaluable patients, resulting in a final sample size of 46 patients. The regimen was considered clinically promising if the lower bound of the 95% CI for ORR exceeded 15%, as estimated using the Clopper-Pearson method.

Efficacy analyses were performed in the full analysis set (FAS), defined as all patients receiving at least one dose of study treatment. The per-protocol set (PPS) included patients evaluable for tumor response, and the safety set (SS) comprised all patients receiving at least one dose of study treatment with available safety data. Categorical variables were compared using the two-sided Fisher’s exact test, and continuous variables using the Wilcoxon rank-sum test. Survival outcomes, including PFS and OS, were estimated using the Kaplan-Meier method and compared with the log-rank test. HRs and corresponding 95% CIs were derived from univariable Cox proportional hazards models. Prespecified subgroup analyses were conducted for efficacy endpoints. Exploratory multivariable Cox proportional hazards regression analyses were performed to evaluate factors associated with PFS and OS. Clinical covariates included age, sex, ECOG performance status, clinical M stage, and prior immunotherapy exposure. Additional multivariable models were constructed to assess the associations between candidate genomic biomarkers and survival outcomes after adjustment for clinical M stage, ECOG performance status, and prior immunotherapy exposure. HRs and corresponding 95% CIs were reported. All statistical analyses were performed using R (V.4.1.2) and SAS (V.9.4). A two-sided p value<0.05 was considered statistically significant.

Results

Patient characteristics

Between November 2021 and December 2024, 47 patients were screened, of whom 46 received at least one dose of study treatment and were included in the FAS and SS. Forty-one patients were evaluable for tumor response and comprised the PPS (figure 1). At the data cut-off, the median follow-up duration was 20.47 months (95% CI 19.53 to 26.81). A total of 27 patients had died, six remained on treatment, and no patients were lost to follow-up.

Figure 1. Patient flow diagram.

Baseline characteristics are summarized in table 1. The median age was 66 years and 67.4% of patients were male. Most patients had an ECOG performance status of 1 (78.3%) and metastatic disease (M1, 84.8%). Twenty-five patients (54.3%) were immunotherapy-naïve, whereas 21 (45.7%) had received prior immunotherapy. PD-L1 positivity (CPS ≥1) was observed in 76.1% of patients. Tumor samples were available for next-generation sequencing (NGS) in 42 patients (91.3%), among whom EGFR amplification was identified in 37.0%.

Table 1. Patient characteristics.

Characteristic	Camrelizumab plus nimotuzumab (n=46)
Age, years
Median (range)	66 (42–77)
≤65	20 (43.5%)
>65	26 (56.5%)
Sex
Male	31 (67.4%)
Female	15 (32.6%)
ECOG performance status*
0	10 (21.7%)
1	36 (78.3%)
Clinical M stage†
0	7 (15.2%)
1	39 (84.8%)
Lymph node metastases
Yes	21 (45.7%)
No	25 (54.3%)
Location of metastases
Liver	10 (21.7%)
Lung	12 (26.1%)
Lymph node	37 (80.4%)
Bone	4 (8.7%)
Adrenal gland	2 (4.3%)
pleura	2 (4.3%)
kidney	2 (4.3%)
Previous therapies
Surgery	27 (58.7%)
Radiotherapy	19 (41.3%)
Chemotherapy	45 (97.8%)
Immunotherapy status
Immune-naïve	25 (54.3%)
Immune-treated	21 (45.7%)
PD-L1 expression
Missing	2 (4.3%)
CPS <1	9 (19.6%)
CPS ≥1	35 (76.1%)
NGS testing	42 (91.3%)
EGFR amplification	17 (37.0%)
Non-EGFR amplification	25 (54.3%)

Data are expressed as number (%) unless otherwise indicated.

^*ECOG performance status scores range from 0 to 5, with 0 indicating no symptoms and higher scores indicating greater disability.

^†American Joint Committee on Cancer stage.

CPS, combined positive score; ECOG, Eastern Cooperative Oncology Group; EGFR, epidermal growth factor receptor; NGS, next-generation sequencing; PD-L1, programmed death-ligand 1.

The median number of treatment cycles was 15 (IQR 11–39) for nimotuzumab and 13 (IQR 6–20) for camrelizumab, with adherence rates of 82% and 86%, respectively. Treatment discontinuation was primarily due to disease progression (n=19), followed by voluntary withdrawal (n=9; including five related to COVID-19 restrictions), adverse events (AEs) (n=2), use of prohibited concomitant medications (n=1), protocol completion (n=3), and poor compliance (n=1).

Efficacy

In the FAS, two patients achieved a complete response and 13 achieved a partial response (table 2; figure 2A and B), yielding a confirmed ORR of 32.6% (15/46; 95% CI 19.5 to 48.0). In the PPS, the ORR was 36.6% (15/41; 95% CI 22.1 to 53.1). Response rates were generally consistent across prespecified subgroups, including prior immunotherapy exposure (immune-naïve, 32.0% vs immune-treated, 33.3%), disease stage (M0, 42.9% vs M1, 30.8%), EGFR amplification status (47.1% vs 24.0%), and PD-L1 expression (CPS ≥1, 34.3% vs CPS <1, 33.3%). The median TTR and DOR were not reached at the time of analysis. Stable disease was observed in 23 patients, resulting in a DCR of 82.6% (38/46; 95% CI 68.6 to 92.2) (figure 2A). Additional subgroup analyses are presented in online supplemental table 2.

Table 2. Tumor response.

Response	Full analysis set (n=46)	Per-protocol set (n=41)
Complete response	2 (4.3%)	2 (4.9%)
Partial response	13 (28.3%)	13 (31.7%)
Stable disease	23 (50.0%)	23 (56.1%)
Progressive disease	3 (6.5%)	3 (7.3%)
Not evaluable	5 (10.9%)	0 (0%)
Objective response	15 (32.6% (19.5–48.0))	15 (36.6% (22.1–53.1))
Disease control	38 (82.6% (68.6–92.2))	38 (92.7% (80.1–98.5))

Data are n (%) or n (% (95% CI)).

Figure 2. Treatment response and survival outcomes. (A) Waterfall plot showing the best percentage change in the sum of target lesion diameters for individual patients with advanced or metastatic esophageal squamous cell carcinoma (ESCC), assessed by investigators according to Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1. Horizontal dashed lines indicate a 20% increase, 30% decrease, and 100% decrease in tumor burden. Each bar represents an individual patient in the full analysis set (FAS). (B) Swimmer plot depicting time to tumor response (months) for individual patients. Each lane represents one patient in the FAS. (C) Kaplan-Meier estimates of overall survival (OS) in the FAS population. (D) Kaplan-Meier estimates of progression-free survival (PFS) in the FAS population. CR, complete response; EGFR, epidermal growth factor receptor; PD, progressive disease; PR, partial response; SD, stable disease.

At the time of data cut-off, 32 OS events had occurred in the FAS. The median OS was 12.62 months (95% CI 9.40 to 15.01), with estimated 9-month and 12-month OS rates of 67.4% (95% CI 51.9 to 78.9) and 54.3% (95% CI 38.9 to 67.3), respectively (figure 2C). Survival outcomes were broadly comparable across subgroups. Median OS was 10.94 months (95% CI 7.10 to 18.92) in immunotherapy-naïve patients and 14.52 months (95% CI 7.95 to 15.38) in previously treated patients. Patients with M0 disease had not reached median OS, whereas those with M1 disease had a median OS of 12.55 months (95% CI 7.95 to 14.88). Median OS was 13.17 months (95% CI 10.22 to 18.92) in patients with EGFR amplification and 9.99 months (95% CI 7.06 to 15.38) in those without amplification. Median OS was 12.62 months (95% CI 9.89 to 15.38) for patients with PD-L1 CPS ≥1 and 14.26 months (95% CI 0.56 to 21.09) for those with CPS <1 (online supplemental table 3).

A total of 35 PFS events were observed. The median PFS was 9.13 months (95% CI 5.95 to 9.76), with 9-month and 12-month PFS rates of 52.2% (95% CI 36.6 to 65.7) and 28.0% (95% CI 15.5 to 42.0), respectively (figure 2D). Median PFS was similar across key subgroups, including immunotherapy-naïve versus immunotherapy-treated patients (9.40 vs 8.80 months), M0 vs M1 disease (14.52 vs 8.48 months), EGFR amplification versus non-amplification (9.43 vs 9.40 months), and PD-L1 CPS ≥1 vs <1 (9.13 vs 9.40 months). Detailed subgroup analyses are provided in online supplemental table 3.

A total of 35 PFS events were observed. The median PFS was 9.13 months (95% CI, 5.95 to 9.76), with 9-month and 12-month PFS rates of 52.2% (95% CI, 36.6 to 65.7) and 28.0% (95% CI, 15.5 to 42.0), respectively (figure 2D). Median PFS was similar across key subgroups, including immunotherapy-naïve vs treated patients (9.40 vs 8.80 months), M0 vs M1 disease (14.52 vs 8.48 months), EGFR amplification vs non-amplification (9.43 vs 9.40 months), and PD-L1 CPS≥1 versus <1 (9.13 vs 9.40 months). Detailed subgroup analyses are provided in online supplemental table 3.

Exploratory multivariable Cox regression analyses incorporating baseline clinical characteristics (age, sex, ECOG performance status, clinical M stage, and prior immunotherapy exposure) are presented in online supplemental table 4. No clinical variable was independently associated with PFS or OS.

Safety

In the SS, TRAEs of any grade occurred in 37 patients (80.4%). The most common grade 3 or higher TRAEs were anemia (n=2), leukopenia (n=2), hypokalemia (n=2), and dysphagia (n=2). No treatment-related deaths were reported. Treatment discontinuation due to AEs occurred in three patients, including grade 1 hypokalemia (n=1), grade 1 fever (n=1), and grade 3 COVID-19 infection (n=1). Immune-related adverse events (irAEs) were predominantly low grade, with the most common event as reactive cutaneous capillary endothelial proliferation (32.6%, table 3).

Table 3. Treatment-related adverse events occurring in at least 5% of patients and immune-related adverse events.

Events, n (%)	Any grade	Grade 1	Grade 2	Grade 3	Grade 4
Treatment-related adverse events
Anemia	23 (50.0%)	18 (39.1%)	12 (26.1%)	2 (4.3%)	0 (0%)
Reactive cutaneous capillary endothelial proliferation	17 (37.0%)	17 (37.0%)	0 (0%)	0 (0%)	0 (0%)
Leukopenia	15 (32.6%)	12 (26.1%)	9 (19.6%)	2 (4.3%)	0 (0%)
Neutropenia	13 (28.3%)	6 (13.0%)	9 (19.6%)	1 (2.2%)	0 (0%)
Cough	10 (21.7%)	7 (15.2%)	7 (15.2%)	0 (0%)	0 (0%)
Coagulopathy	8 (17.4%)	4 (9%)	5 (10.9%)	0 (0%)	0 (0%)
Hypoalbuminemia	7 (15.2%)	7 (15.2%)	1 (2.2%)	0 (0%)	0 (0%)
Hypokalemia	6 (13.0%)	6 (13.0%)	1 (2.2%)	2 (4.3%)	0 (0%)
Decreased appetite	6 (13.0%)	2 (4.3%)	5 (10.9%)	0 (0%)	0 (0%)
Constipation	5 (10.9%)	3 (6.5%)	3 (6.5%)	0 (0%)	0 (0%)
Thrombocytopenia	5 (10.9%)	5 (10.9%)	1 (2.2%)	1 (2.2%)	0 (0%)
Abnormal liver function	4 (8.7%)	3 (6.5%)	2 (4.3%)	0 (0%)	0 (0%)
Dysphagia	4 (8.7%)	1 (2.2%)	3 (6.5%)	2 (4.3%)	0 (0%)
Hyponatremia	3 (6.5%)	3 (6.5%)	0 (0%)	0 (0%)	0 (0%)
Fever	3 (6.5%)	2 (4.3%)	1 (2.2%)	0 (0%)	0 (0%)
Acid reflux	3 (6.5%)	1 (2.2%)	3 (6.5%)	0 (0%)	0 (0%)
Hypothyroidism	3 (6.5%)	1 (2.2%)	2 (4.3%)	0 (0%)	0 (0%)
Insomnia	3 (6.5%)	0 (0%)	3 (6.5%)	0 (0%)	0 (0%)
Dizziness	3 (6.5%)	2 (4.3%)	1 (2.2%)	0 (0%)	0 (0%)
Immune-related adverse events
Reactive cutaneous capillary endothelial proliferation	15 (32.6%)	15 (32.6%)	0 (0%)	0 (0%)	0 (0%)
Hypothyroidism	3 (6.5%)	1 (2.2%)	1 (2.2%)	1 (2.2%)	0 (0%)
Neutropenia	3 (6.5%)	2 (4.3%)	2 (4.3%)	0 (0%)	0 (0%)
Leukopenia	3 (6.5%)	2 (4.3%)	3 (6.5%)	0 (0%)	0 (0%)
Capillary hemangioma	2 (4.3%)	0 (0%)	2 (4.3%)	1 (2.2%)	0 (0%)
Blood thyroid-stimulating hormone increased	1 (2.2%)	1 (2.2%)	0 (0%)	0 (0%)	0 (0%)
Fecal occult blood positive	1 (2.2%)	1 (2.2%)	0 (0%)	0 (0%)	0 (0%)
Blood creatinine increased	1 (2.2%)	1 (2.2%)	1 (2.2%)	0 (0%)	0 (0%)
Asthenia	1 (2.2%)	0 (0%)	0 (0%)	1 (2.2%)	0 (0%)
Increased creatinine	1 (2.2%)	1 (2.2%)	0 (0%)	0 (0%)	0 (0%)
Rash	1 (2.2%)	1 (2.2%)	1 (2.2%)	1 (2.2%)	0 (0%)
Anemia	1 (2.2%)	0 (0%)	0 (0%)	1 (2.2%)	0 (0%)

Genomic exploratory analysis and clinical efficacy

Among the 42 patients who underwent whole-exome sequencing, all samples yielded evaluable genomic profiles (100% success rate). Genes mutated in at least three patients were included in subsequent analyses to explore associations with clinical outcomes. The most frequently mutated genes were TP53 (85.7%), TTN (45.2%), MUC16 (28.6%), FAT1 (19.0%), NOTCH3 (14.3%), and NOTCH1 (11.9%). Recurrent copy number alterations included amplifications of CCND1, FGF19, FGF3, FGF4, and deletions involving MTAP (online supplemental figure 1).

No associations were observed between TMB, intratumoral heterogeneity, or chromosomal instability and treatment response (all p>0.05). In contrast, specific genomic alterations demonstrated associations with clinical outcomes. Patients harboring MUC16 mutations had a lower ORR compared with those with wild-type tumors (8.3% vs 46.4%, p=0.030; figure 3A), whereas other frequently mutated genes showed no clear relationship with response (figure 3B and C).

Figure 3. Somatic genomic alterations and associations with clinical outcomes. (A) Association between MUC16 mutation status and objective response rate (ORR). (B–C) Association of NOTCH3 mutation and MTAP copy number deletion status with ORR. (D–I) Kaplan-Meier analyses of PFS and OS according to NOTCH3 mutation and MTAP deletion status. NOTCH3 mutations were associated with prolonged survival, whereas MTAP deletions were associated with shorter PFS. CR, complete response; mOS, median overall survival; mPFS, median progression-free survival; NA, not available; PD, progressive disease; PR, partial response; SD, stable disease.

Regarding survival outcomes, NOTCH3 mutations were associated with prolonged PFS (median not reached vs 8.21 months; HR 0.20, p=0.015; figure 3E) and OS (median not reached vs 10.58 months; HR 0.22, p=0.026; figure 3H). In contrast, MTAP copy number deletions were associated with shorter PFS (3.71 vs 9.49 months; HR 3.18, p=0.005; figure 3F) and a trend toward reduced OS (p=0.070; figure 3I). No clear differences in survival outcomes were observed according to MUC16 mutation status (figure 3D and G). Detailed results are provided in online supplemental table 5. After adjustment for clinical M stage, ECOG performance status, and prior immunotherapy exposure, the associations between NOTCH3 mutation and improved survival outcomes remained significant (PFS: HR 0.15, p=0.012; OS: HR 0.22, p=0.049), while MTAP deletion remained associated with shorter PFS (HR 3.75, p=0.009) (online supplemental table 6).

To explore potential biological mechanisms, external validation analyses were performed using transcriptomic and mutation data from TCGA ESCC cohort. MUC16-mutant tumors exhibited reduced infiltration of CD8+ T cells and regulatory T cells (online supplemental figure 2), with no evident differences in pathway enrichment. NOTCH3 mutations were associated with decreased infiltration of M2 macrophages (p=0.002; online supplemental figure 3A), without enrichment of specific signaling pathways. In contrast, tumors with MTAP deletions demonstrated increased infiltration of regulatory T cells, natural killer cells, B cells, and neutrophils (online supplemental figure 3B). GSEA revealed upregulation of epithelial-mesenchymal transition-related pathways and downregulation of DNA damage repair, G2M checkpoint, interferon-α/γ signaling, p53, mechanistic target of rapamycin complex 1, and tumor necrosis factor-α/nuclear factor κB pathways in MTAP-deleted tumors (online supplemental figure 3C).

Discussion

This study represents, to our knowledge, the first prospective evaluation of combined PD-1 blockade and EGFR-targeted therapy with camrelizumab plus nimotuzumab as second-line treatment for advanced ESCC. The regimen demonstrated encouraging antitumor activity, with an ORR of 32.6%, a median PFS of 9.13 months, and a median OS of 12.62 months. These outcomes appear favorable relative to those reported in previous studies of second-line ICI monotherapy, which yielded ORRs of 12.6%–20.3%, median PFS of 1.6–2.1 months, and median OS of 7.1–10.9 months.^{5 6 9} However, cross-trial comparisons should be interpreted with caution because of differences in study design and patient characteristics. Although the proportion of patients with M0 disease in our cohort (15.2%) was slightly higher than that reported in prior studies (2%–11.8%),^{7 9} subgroup analyses demonstrated maintained clinical activity in the metastatic (M1) population, with an ORR of 30.8%, median PFS of 8.48 months, and median OS of 12.55 months. These findings support the potential of combined EGFR inhibition and immunotherapy to enhance clinical efficacy beyond that achieved with PD-1 blockade alone in this setting.

The observed clinical activity may be explained by the complementary antitumor mechanisms of EGFR inhibition and PD-1 blockade. EGFR-targeted monoclonal antibodies, such as nimotuzumab, bind to the extracellular domain of EGFR, thereby inhibiting downstream signaling pathways involved in tumor proliferation, angiogenesis, and survival.²⁵ Beyond direct tumor inhibition, EGFR blockade may modulate the tumor microenvironment by enhancing antigen presentation, reducing immunosuppressive cell populations, and promoting T-cell infiltration and cytotoxicity. Preclinical studies further suggest that EGFR inhibition can augment major histocompatibility complex class I expression and sensitize tumor cells to CD8+ T cell-mediated killing. In addition, nimotuzumab has been shown to activate natural killer cells, facilitating dendritic cell maturation, and the priming of tumor-specific T-cell responses,²⁶ while restoring human leukocyte antigen class I expression on tumor cells to enhance immune recognition.²⁷ Collectively, these effects provide a biological rationale for combining EGFR-targeted therapy with ICI, supporting the potential for synergistic antitumor activity observed in this study.

Our findings suggest that EGFR amplification may represent a clinically relevant biomarker of response to combined EGFR-targeted therapy and PD-1 blockade. The prevalence of EGFR amplification in our cohort (40.5%) was consistent with prior reports (37.7%).²⁸ Patients with EGFR amplification demonstrated a higher response rate and a numerically prolonged OS compared with those without amplification, supporting a potential predictive role for EGFR-driven tumors. Similar trends have been observed in other anti-EGFR studies in ESCC, including a trial of cetuximab in which improved outcomes were reported both in the overall population and more prominently in the EGFR-amplified subgroup.²⁹ Although PFS was comparable between EGFR-amplified and non-amplified groups in our study, both appeared improved relative to historical outcomes with chemotherapy, suggesting a broader benefit of this combination regimen. Mechanistically, EGFR inhibition may enhance antitumor immunity by reducing immunosuppressive signaling, promoting immune cell infiltration, and facilitating the transition from an immunologically ‘cold’ to ‘hot’ tumor microenvironment. In combination with ICI, these effects may augment antigen presentation, induce immunogenic cell death, and reinforce the cancer-immunity cycle, resulting in more durable tumor control.^{30 31}

Clinical benefit was observed across key subgroups, including both immunotherapy-naïve and previously treated patients. Notably, outcomes in the immunotherapy-exposed population appeared more favorable than those reported in the CAP02 re-challenge study, with higher response rates and longer survival (ORR 33.3% vs 10.2%; median OS 14.52 vs 7.5 months; median PFS 8.80 vs 4.6 months).¹⁰ In our cohort, efficacy in previously treated patients was comparable to that in immunotherapy-naïve patients, suggesting that this combination regimen may partially overcome resistance to prior PD-1 blockade. When compared with other combination strategies, including camrelizumab plus apatinib in the CAP-02 study, the observed response rate was similar, with a modest improvement in PFS (ORR 32.6% vs 34.6%; median PFS 9.13 vs 6.8 months).³² Consistent trends were also noted when compared with other regimens in previously treated populations, with numerically higher response and survival outcomes.^{10 33} However, cross-trial comparisons should be interpreted with caution given differences in study design and patient populations. Importantly, treatment activity appeared independent of PD-L1 expression, with comparable efficacy observed between patients with CPS <1 and CPS ≥1. This finding is consistent with prior evidence from the ATTRACTION-3 trial, in which OS was similar regardless of PD-L1 status. Collectively, these observations suggest that the therapeutic benefit of this combination may extend across clinically relevant subgroups, irrespective of prior immunotherapy exposure or PD-L1 expression.

The safety profile of the combination regimen was acceptable, with most TRAEs being low grade and manageable, and no new safety signals identified. irAEs, including reactive cutaneous capillary endothelial proliferation and hypothyroidism, were consistent with the known toxicity profile of camrelizumab. Compared with the ESCORT trial, the incidence of irAEs (61% vs 94%), serious AEs (2% vs 16%), and grade ≥3 irAEs (8.7% vs 15%) appeared lower in the present study.⁵ Importantly, the addition of nimotuzumab did not result in increased toxicity. This favorable safety profile may be attributable to the intermediate affinity of nimotuzumab, which preferentially targets tumor cells with high EGFR expression while sparing normal tissues.

Exploratory biomarker analyses identified recurrent alterations in MUC16, NOTCH3, and MTAP that may be associated with treatment response and tumor immune contexture. The frequency of MUC16 mutations in our cohort (28.6%) was higher than previously reported in ESCC (~17%),³⁴ potentially reflecting population heterogeneity. Patients harboring MUC16 mutations appeared to exhibit reduced antitumor immune activity, which may be related to alterations in the tumor microenvironment, including decreased CD8+ T-cell infiltration and changes in macrophage and follicular helper T-cell populations.^{35 36} In contrast, NOTCH3, a key component of the NOTCH signaling pathway implicated in inflammatory regulation, has been associated with improved clinical outcomes and may represent a potential immunoregulatory target.^{37 38} In our analysis, NOTCH3 mutations were associated with reduced infiltration of M2 macrophages, suggesting a possible role in limiting tumor immune evasion, although the underlying mechanisms remain to be elucidated. Loss of MTAP has been linked to adverse prognosis and immune suppression across multiple tumor types.³⁹ Preclinical studies indicate that MTAP deficiency may promote immune escape through upregulation of PD-L1 and induction of T-cell dysfunction.^{40 41} Consistent with these observations, tumors with MTAP deletions in our cohort demonstrated increased infiltration of regulatory T cells, B cells, neutrophils, and natural killer cells, which may reflect a complex, potentially immunosuppressive microenvironment. GSEA further suggested upregulation of epithelial-mesenchymal transition pathways and downregulation of DNA damage repair, cell cycle regulation, interferon signaling, and other immune-related pathways. Similar biological patterns have been reported in other malignancies, including urothelial carcinoma, pancreatic cancer, and osteosarcoma.^42–44 Taken together, these findings should be considered hypothesis-generating and warrant further validation in larger, prospective studies.

This study has several limitations. First, the single-arm design and relatively small sample size limit the strength of causal inference, and randomized controlled trials are required to confirm the clinical efficacy of this combination regimen. Second, all biomarker analyses in this study were exploratory and should be interpreted cautiously. The relatively small sample size of several biomarker-defined subgroups and the absence of adjustment for multiple comparisons increase the risk of false-positive findings. Therefore, the predictive value of the identified biomarkers should be considered hypothesis-generating and requires confirmation in independent cohorts. Third, multiplex immunofluorescence analyses were not performed, limiting direct characterization of immune cell composition and interactions within the tumor microenvironment and precluding a comprehensive assessment of the immunomodulatory effects of nimotuzumab. In addition, most tumor specimens were obtained during first-line treatment or from postoperative settings rather than immediately before second-line therapy, limiting their ability to reflect the baseline immune microenvironment at study entry or capture dynamic immune changes during treatment. Consequently, NGS-based molecular profiling was used as an exploratory surrogate approach to investigate potential biological correlates of treatment response. Fourth, the lack of EGFR immunohistochemistry (IHC) data represents a limitation. This gap precludes the correlation of NGS-detected EGFR gene amplification with EGFR protein expression and the analysis of their relationship with clinical outcomes. Importantly, gene copy number variations do not necessarily correlate linearly with functional protein overexpression in ESCC.⁴⁵ Complex post-transcriptional and post-translational mechanisms may decouple genomic amplification from the actual expression of biologically active proteins in tumor cells.⁴⁶ While NGS-based screening identifies genomic alterations that may indicate EGFR pathway dependence, IHC directly assesses protein expression and target availability. The relative utility of these approaches for patient selection remains unclear. Future studies integrating NGS and IHC are therefore warranted to better define the relationship between EGFR amplification, protein expression, and treatment response, and to optimize biomarker-guided use of nimotuzumab-based combination therapy in advanced ESCC.^{45 47} Finally, the external validation analyses were based on the TCGA ESCC cohort, which predominantly comprises treatment-naïve tumors. Given that most patients in our study had received prior chemotherapy and/or immunotherapy, treatment-induced alterations in the tumor microenvironment may limit the direct comparability of immune infiltration patterns between the two cohorts. Therefore, the TCGA analyses should be interpreted as providing mechanistic context rather than direct biological validation of our findings.

In summary, the combination of camrelizumab and nimotuzumab demonstrated encouraging antitumor activity with a manageable safety profile in patients with advanced ESCC, with clinical benefit observed irrespective of prior immunotherapy exposure or PD-L1 expression status. These findings also suggest a potential role for EGFR amplification in identifying patients who may derive greater benefit from EGFR-targeted combination strategies. Exploratory biomarker analyses further indicate that alterations in MUC16, NOTCH3, and MTAP may be associated with treatment outcomes. However, these observations remain hypothesis-generating and warrant validation in randomized controlled studies.

Data availability statement

Data are available on reasonable request.