Neoadjuvant chemoimmunotherapy for locally advanced esophageal squamous cell carcinoma: Data from literature review and a real‐world analysis

Abstract Background Neoadjuvant chemoimmunotherapy (NCIT) for locally advanced esophageal squamous cell carcinoma (ESCC) is supported by increasing data, but the sample size is limited, and the findings are not completely consistent. We conducted a real‐world study and a meta‐analysis to evaluate the efficacy and safety of NCIT in locally advanced ESCC. Methods We retrospectively assessed the outcomes of patients with locally advanced ESCC who completed NICT and subsequent esophagectomy at our hospital between January 2019 and December 2022, including pathological complete response (pCR) rate, major pathological response (MPR) rate, 1‐, 2‐, and 3‐year overall survival (OS) rates, disease control rate (DCR), objective response rate (ORR), 1‐year recurrence rate, R0 resection rate and adverse events. Moreover, a meta‐analysis of 27 published literatures was also conducted for comparison. Results In the analysis, 128 patients were studied, with 25% achieving pCR, 46.1% MPR, and 99.2% R0 resection. The 1‐, 2‐, and 3‐year OS rates were 91.41% (95% CI: 85.15%–95.63%), 75.00% (95% CI: 66.58%–82.23%) and 64.84% (95% CI: 55.91%–73.07%).ORR and DCR were 31.2% (95% CI: 23.31–39.99) and 64.1% (95% CI: 55.15%–72.38%), and the 1‐year recurrence rate was 26.7% (95% CI: 22.5%–38.1%). Treatment‐related events occurred in 96.1% but were acceptable. In a meta‐analysis of 27 studies with 1734 patients, pooled rates for pCR, MPR, ORR, DCR, and R0 resection were 29%, 52%, 71%, 97%, and 98%, respectively, with a 1‐year recurrence rate of 12%. Conclusion NCIT is safe and provides potential survival benefits for patients with locally advanced ESCC. However, randomized phase 3 trial data is still needed.


INTRODUCTION
Esophageal cancer represents the sixth most common cancer worldwide, with an estimated 20 640 new cases and 16 510 deaths in 2022. 1 The majority of this disease is histologically classified as squamous cell carcinoma and adenocarcinoma.
Esophageal adenocarcinoma develops in a predominantly white, male population.In contrast, African Americans and Caucasian males have the most prevalence of esophageal squamous cell carcinoma (ESCC). 2 About half of esophageal cancer patients have localized and regional disease, which includes spread to regional lymph nodes.
For patients with cT1N0M0 and cT2N0M0 lesions (lowrisk lesions), esophagectomy is the preferred treatment option.Patients with locally advanced thoracic esophageal or esophagogastric junction tumors who have not developed metastases to other organs opt for esophagectomy following neoadjuvant chemotherapy or chemoradiotherapy (NCT/NCRT). 3However, even with NCRT, the high incidence of postoperative recurrence and metastasis contributes to treatment failure.
][6] Recently, more and more studies have reported the use of chemoimmunotherapy in the neoadjuvant treatment of locally advanced ESCC.However, the results remain inconsistent, with major pathological response (MPR) rates of 23.5%-88.9%and pathological complete response (pCR) rates of 7.7%-51.1%.This study sought to pool these studies and analyze the efficacy and safety of neoadjuvant chemoimmunotherapy (NCIT) in clinical studies and to compare it with our real-world experience.

Clinical cohort
In this retrospective study, we identified consecutive adult patients with cT1N1-3 M0 or cT2-4aN0-3 M0 ESCC who underwent NCIT followed by esophagectomy as their standard of care between January 2019 and December 2022.Eligibility criteria included: (1) adult patients with cT1N1-3 M0 or cT2-4aN0-3 M0 ESCC who underwent NCIT followed by esophagectomy; (2) patients who received preoperative intravenous administration of PD-1 inhibitor, taxane and cisplatin or carboplatin for 2-4 cycles and (3) all patients had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1. Patients lost to follow-up were excluded from the study.The flow chart is shown in Figure S20.The study was approved by the Ethnical Review Committee of Fudan University Shanghai Cancer Center, and written informed consents were waived.

Data collection and outcomes evaluation
The data were collected from the medical records of each patient including as follows: age, sex, height, weight, previous medical history, previous history of smoking and alcohol drinking, tumor stage, depth of invasion, lymph node metastasis, the type of neoadjuvant immune checkpoint inhibitor, other details of NCIT (such as therapeutic cycles), imaging outcomes, surgical data (such as surgical approach, resection margin status, postoperative complications), pathological grade, tumor regression grade (TRG), and survival follow-up information.The TRG was assessed according to Chirieac's system. 7,8he primary outcome was pCR (defined as the complete absence of intact tumor cells in the resected specimen and all resected lymph nodes) rate. 9Secondary outcomes included objective response rate (ORR), disease control rate (DCR), MPR (defined as ≤10% viable tumor cells in the resected primary tumor and all resected lymph nodes) rate, R0 resection (defined as no cancer cells at resection margins) rate, 1-, 2-, and 3-years OS rates, 1-year recurrence rate and adverse events (AEs).AEs occurred during NCIT were graded according to the National Cancer Institute Common Terminology criteria for Adverse Events (CTCAE) version 4.03.

Literature search strategy for meta-analysis
This meta-analysis was performed in accordance with the Preferred Reporting Items for Systematic Review and Metaanalysis (PRISMA) criteria (Table S1).We searched for available articles published on PubMed, Embase, Cochrane Library, Web of Science, and abstracts presented at several important international conferences before November 1, 2022, using the Medical Subject Headings (MeSH) terms shown in Table S2.To supplement the electronic search results, a manual review of the reference lists of retrieved studies was conducted to identify any relevant additional studies that might have been omitted by the electronic search.For further analysis, all the additional studies that might be of interest were retrieved.In cases where studies were serially published, only the latest and most comprehensive data was selected for assessment.We excluded studies in languages other than English and/or those involving animals.To minimize potential publication bias and duplication of results, we also excluded reviews, abstracts, case reports, conference papers, editorials and expert opinions.Additionally, we also researched some ongoing clinical studies on academic meeting and clinical trials registry platforms.As the present data were obtained from previously published studies, ethical approval and patient consent were not applicable.

Study selection for meta-analysis
Two coauthors (Yao Zhang, Huiting Li) conducted the screening of the title and abstracts.The full texts of all retrieved studies were assessed based on five dimensions (population, intervention, comparison, outcomes, and study design [PICOS]) in an unbiased manner.Studies that met the following criteria included: (1)

Data extraction for meta-analysis
Data were independently extracted by two coauthors (Yao Zhang, Huiting Li) from texts, tables, and figures of analyzed articles.In case of disagreements, a third author was involved in the discussion or consultation to resolve them.We collected the detailed data of each trial, including first author, year of publication, country of origin, study design, recruitment period, duration of follow-up, number of study centers, patients' demographics and clinical information, and the incidence of AEs.In cases where articles indirectly presented survival data with Kaplan-Meier curves, GetData Graph Digitizer software was utilized to handle and extract time-specific data.We also collected trial data (e.g., trial name, tumor type, number of patients, therapeutic strategies, pCR rate, MPR rate, R0 resection rate, ORR, DCR, clinical trial phase, and study period) from academic abstracts or ongoing prospective studies.

Quality assessment for meta-analysis
Two independent assessors (Jialei Wang, Hui Yu) assessed the strength of evidence for each outcome with the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) system.A table summarizing the outcomes was provided to properly identify and annotate the certainty of all pooled results.Thereafter, each study was retrieved by the Newcastle-Ottawa Scale (NOS), which accounts for selected criteria, such as selection, comparability, and outcome, to assess the quality of the study design.

Bias assessment for meta-analysis
To assess publication and small-study bias, a funnel plot was generated, and then asymmetry Egger's linear regression and Begg's correlation tests were performed to inspect the suspected asymmetry for small-study bias.To assess the risk of single-study bias, all outcomes were analyzed with sensitivity analysis by exclusion.

Statistical analysis
For our real-world study, demographics, clinical data, and safety profile were tabulated using descriptive statistics.Continuous variables are expressed as median (range) or mean ± standard deviation and were compared by one-way analysis of variance (ANOVA) or Wilcoxon sum test as appropriate.Categorical variables are presented as frequency (percentage) and were compared by χ 2 -test.p-values were two-sided, with a significance level of 0.05.The Kaplan-Meier method was used to estimate OS was estimated with the Kaplan-Meier method.
For meta-analysis, heterogeneity across the studies was assessed by Q-test and I 2 statistics.Statistical significance of heterogeneity was considered as p < 0.05 or I 2 > 50%.If significant heterogeneity was observed, a random-effect model was employed; otherwise, a fixed-effect model was utilized.The robustness of the results was checked by sensitivity analysis.Funnel plots were drawn to evaluate publication bias.Statistical analysis was conducted using R 4.0.0 (R Core Team 2020) 10 and GraphPad Prism 6.02 (GraphPad Software Inc., San Diego, CA, USA).High-quality figures were generated using the R packages.

Recurrence rate of ESCC
A total of 48 patients (37.5%) had recurrence until the end of follow-up.The median recurrence interval was 11.7 months (299 days).The main recurrence sites were lymph nodes  F I G U R E 1 Main outcomes of neoadjuvant chemoimmunotherapy for locally advanced esophageal squamous cell carcinoma.DCR, disease control rate; MPR, major pathological response; ORR, objective response rate; pCR, pathological complete response.
The incidence of grade 3 or higher adverse events.RCCEP, reactive cutaneous capillary endothelial proliferation.

Publication bias and sensitivity analysis
The publication bias for the pooled proportion was evaluated with Begg's and Egger's tests (data not shown).Figures S8-S14 show the funnel plots.In general, the distribution of points in the funnel plot was more balanced, suggesting that the bias of each outcome index was slight.All sensitivity analyses related to this meta-analysis indicated stable results (Figures S15-S19).

DISCUSSION
The conventional therapeutic procedure for locally advanced ESCC is NCRT followed by esophagectomy, which has a pCR rate of 29%-43.2%. 31,32Nevertheless, a higher pCR rate in NCRT did not significantly increase long-term survival over NCT.In addition, radiotherapy may increase the difficulty of surgery and surgical complications.Neoadjuvant immunotherapy has been shown to be effective in patients with localized esophageal cancer.However, comprehensive evaluations of the effectiveness and safety of neoadjuvant immunotherapy in ESCC are rare.
Here, we provided evidence on the use of NCIT in ESCC based on our institute's experience involving 128 patients and 27 published articles including 1734 patients.In our retrospective study, the pCR rate was 25.0% (95% CI: 18.3%-33.2%).In our meta-analysis, the pooled pCR rate was 29% (95% CI: 26%-32%), which is inferior to some other systematic reviews with a synthesized pCR of 32.5% 33 and 32.9%. 34he highest pCR rate of 57.1% was from the study reported by Ma et al., 35 which is an abstract reporting the results and was not included in our analysis.The MPR rate for NCIT was 46.1% from our center and the pooled MPR rate was 52%, both higher than the MPR rate of 33.3% for NCT based on recent clinical trials.Moreover, the pooled R0 resection rate was 98%, and a higher R0 resection rate (99.2%) was observed in our real-world analysis, both of which were better than the rate with NCT.These findings support the feasibility of NCIT for ESCC.According to our study, the 1-, 2-, and 3-year OS rates were 91.41% (95% CI: 85.15%-95.63%),75.00% (95% CI: 66.58%-82.23%),and 64.84% (95% CI: 55.91%-73.07%),respectively.One study conducted by Duan and his colleagues reported that the median disease-free survival for patients with NCIT was 13.8 months among 17 patients.Most recent studies have not provided detailed survival data and long-term follow-up findings.Whether NCIT will provide long-term survival benefits remain to be seen.
With regard to safety, immunotherapy did not significantly increase the incidence of TRAEs, according to the KEYNOTE-590 trial 4 and some other clinical trials. 36,37In our meta-analysis and retrospective study, although most patients experienced at least one TRAE, mainly including alopecia, RCCEP, constipation, neutropenia, and rash, these TRAEs were all manageable and did not affect the treatment.NCIT did not increase surgical difficulty in our cases, and there were no surgical complications that resulted in death.Zhang et al. 38 found that compared to nCRT, nCIT achieved comparable pCR and MPR rates, but with significantly more lymph nodes removed during surgery, shorter operation time, reduced intraoperative blood loss, and fewer ICU admissions post-surgery.
For patients with esophageal cancer, there are no available biomarkers to precisely predict the therapeutic outcomes of immunotherapy.The most commonly employed biomarker is PD-L1 expression.However, it is still debatable if PD-L1 expression levels are associated with pathological response in patients treated with neoadjuvant immunotherapy.Several clinical trials 39,40 suggested that patients could benefit from immunotherapy, irrespective of PD-L1 expression levels.Some other studies indicated that the pCR group showed considerably higher level of PD-L1 before treatment when compared to the non-pCR group 28.In our center, according to the post-surgery biopsy, 38 patients were PD-L1 positive with CPS ≥10, and their pCR and MPR rates were 2.6% (1/38) and 18.4% (7/38), respectively.A total of 37 patients were PD-L1 negative with CPS <10, and their pCR and MPR rates were 0% (0/37) and 21.6% (8/37), respectively.No significant differences were found between the two groups in our study.Since the baseline PD-L1 expression level was not evaluated, the association between it and pathological response could not be evaluated in our real-world study.We were also unable to run a meta-regression analysis to include these putative biomarkers due to the lack of research data.
There were several limitations in the present study.First, the disease stage could be a factor for the disparities between the MPR rate and R0 resection rate.However, in our study, we were unable to measure the influence of disease stage on the clinical outcomes.Second, different immune checkpoint inhibitors were involved in our study, which could make the current study's conclusions less generalizable.Third, the most studies involved in our meta-analysis were singlearm studies, which may lead to the instability and bias of the findings.Large randomized controlled trials are still warranted in the future.
In conclusion, NCIT has an acceptable safety profile, obtained pathological downstaging, and may bring survival benefits for locally advanced esophageal squamous cell carcinoma, indicating its application value in resectable esophageal squamous cell carcinoma.
location of the tumor in the esophagus was determined on the basis of endoscopic findings and was divided into three segments: cervical/upper (15-25 cm from the incisor teeth), middle (25-30 cm), and lower (30-40 cm) segments of the esophagus.Location 2 and 3 represent the middle and lower segments of the esophagus.b Descent stage: Descent stage means the decrease in the stage of a patient after neoadjuvant immunochemotherapy.Zero represents that the patient's stage did not change after neoadjuvant immunochemotherapy, while 1 represents the opposite.

F I G U R E 3
The pathological complete response rate of meta-analysis.F I G U R E 4The major pathological response rate of meta-analysis.
T A B L E 2 Postoperative complications.