Impact of the interval between neoadjuvant immunotherapy and surgery on prognosis in esophageal squamous cell carcinoma (ESCC): a real-world study

Background The time interval between neoadjuvant immunotherapy and surgery is 6 weeks for esophageal squamous cell carcinoma (ESCC), but whether delayed surgery affects prognosis remains unclear. Methods Clinical data of locally advanced ESCC who underwent neoadjuvant immunotherapy followed by esophagectomy from November 2019 to December 2022 were collected. The surgery outcomes and prognosis were compared between short-interval (time to surgery ≤ 6 weeks) and long-interval groups (time to surgery > 6 weeks). Results 152 patients were enrolled totally, with a ratio of 91:61 between short-interval and long-interval groups. The rate of pathological complete response in the short-interval and long-interval groups were 34.1% and 24.6% (P = 0.257). Delayed surgery did not have a significantly impact on the number of lymph node dissections (P = 0.133), operative duration (P = 0.689), blood loss (P = 0.837), hospitalization duration (P = 0.293), chest drainage duration (P = 0.886) and postoperative complications (P > 0.050). The 3-year Overall survival (OS) rates were 85.10% in the short-interval group, and 82.07% in the long-interval group (P = 0.435). The 3-year disease-free survival (DFS) rates were 83.41% and 70.86% in the two groups (P = 0.037). Subgroup analysis revealed that patients with a favorable response to immunotherapy (tumor regression grade 0) exhibited inferior 3-year OS (long-interval vs. short-interval: 51.85% vs. 91.08%, P = 0.035) and DFS (long-interval vs. short-interval: 47.40% vs. 91.08%, P = 0.014) in the long-interval group. Conclusions Delayed surgery after neoadjuvant immunotherapy does not further improve pathological response; instead, it resulted in a poorer DFS. Especially for patients with a favorable response to immunotherapy, delayed surgery increases the risk of mortality and recurrence. Supplementary Information The online version contains supplementary material available at 10.1007/s00262-024-03787-2.


Introduction
Esophageal cancer ranks as the seventh most common malignant neoplasm worldwide and the sixth principal cause of cancer-induced mortality.According to the 2020 GLOBOCAN statistics [1], there are approximately 604,000 new cases of esophageal cancer worldwide each year, with around 544,000 new deaths.The incidence and mortality rates remain consistently elevated.China is identified as a high-incidence country for esophageal cancer, with esophageal squamous cell carcinoma (ESCC) constituting over 90% of cases.[2,3].The bulk of cases are typically identified at an locally advanced stage, due to the propensity for early lymph node metastasis.
In recent years, programmed death protein 1 (PD-1) antibodies plus neoadjuvant chemotherapy has been investigated Guozhen Yang and Yutong Hong have contributed equally to this work.
in several clinical trials for locally advanced ESCC [4,5].This neoadjuvant regimen has demonstrated the potential to reduce the size of primary tumors, increase the success rate of surgical resection, achieve favorable rates of pathological complete response (pCR), and exhibit tolerable and manageable toxicity profiles.It offers promising prospects for widespread application in locally advanced ESCC.However, patient's condition after neoadjuvant treatment, local lesion and the systemic inflammatory response, can all impact surgical results.These factors may change over time, indicating that appropriate surgical timing is crucial for optimizing surgical outcomes.Current literature on the timing of surgery following neoadjuvant therapy shows inconsistent conclusions.Kim et al. [6] and Tessier et al. [7] reported that delayed surgery did not affect prognosis, whereas Qin et al. [8] observed that delayed surgery was associated with worse prognosis.Currently, most clinical trials performed surgery within 6 weeks after the end of neoadjuvant immunochemotherapy [9][10][11], which was based on the experience from traditional neoadjuvant chemoradiotherapy [12].In traditional neoadjuvant chemoradiotherapy, surgery is typically scheduled within 4-6 weeks after radiotherapy to avoid exacerbating tissue fibrosis induced by radiation and increasing the surgical difficulty [13].However, PD-1 inhibitor exert completely different mechanisms of action from radiation therapy, by obstructing the interaction between PD-1 and programmed death-ligand 1 (PD-L1), thus alleviating the suppression of T cells, and restoring their capacity to combat tumors.[14][15][16].Additionally, the durable response of immunotherapy could promote tumor continue to shrink even after the drug is stopped for a period of time [17].Therefore, it remains unclear whether delaying surgery in the context of immunotherapy would impact patient prognosis.
Therefore, we conducted a population-based, realworld, retrospective study, which enrolled patients with locally advanced ESCC who underwent neoadjuvant PD-1 inhibitor plus chemotherapy followed by surgery.The patient cohort was divided into short-interval group (time to surgery ≤ 6 weeks) and long-interval group (time to surgery > 6 weeks), to assess whether delayed surgery after immunotherapy affected surgical and pathological outcomes as well as long-term prognosis.

Patients selection
We retrospectively collected patients with ESCC who underwent neoadjuvant PD-1 inhibitor combined chemotherapy followed by surgery at the Department of Thoracic Surgery, Sun Yat-sen University Cancer Center from November 2019 to December 2022.Inclusion criteria were as follows: 1) age ranging from 18 to 80 years; 2) pathologically diagnosed with thoracic ESCC; 3) Clinical stage T1N1-3 or T2-4aN0-3 according to the Union for International Cancer Control/American Joint Committee on Cancer (UICC/AJCC), 8th edition; 4) administration of at least one cycle of PD-1 inhibitor combined chemotherapy prior to surgery; 5) availability of complete clinical, pathological, and follow-up data; 6) eastern cooperative oncology group (ECOG) performance status of 0-1.Exclusion criteria were: (1) adenocarcinoma, sarcoma, or other non-squamous cell carcinoma types; (2) received other neoadjuvant regimen before surgery, such as radiotherapy, targeted therapy; (3) clinical data missing.

Neoadjuvant regimen
In the overall cohort, the median number of neoadjuvant treatment cycles was 3 (range: 2-6).PD-1 inhibitors used in the study comprised camrelizumab, pembrolizumab, sintilimab, tislelizumab, toripalimab, and penpulimab.Chemotherapy regimens consisted of: (1) paclitaxel-based drugs plus platinum-based regimens, and (2) paclitaxelbased drugs plus fluoropyrimidine-based regimens.Detailed treatment protocols were provided in Supplementary Figure 1.Drug dosage and administration details were outlined in Supplementary Table 1.

Surgery procedure
Surgery was conducted after completing neoadjuvant therapy.The interval between neoadjuvant treatment and surgery was defined as the duration from the last cycle of treatment to the surgical procedure.Surgical approaches included traditional thoracotomy, McKeown or Ivor-Lewis minimally invasive esophagectomy (MIE), and roboticassisted thoracoscopic/laparoscopic esophagectomy.Standard two-field lymph node dissection was conducted, with consideration of three-field lymph node dissection if preoperative cervical lymph node ultrasound indicated cervical lymph node metastasis.All surgical procedures were conducted by experienced thoracic surgeon.

Follow-up
The initial postoperative follow-up occurred at 3 months post-surgery, followed by subsequent visits every 3 months during the first two years, and thereafter at intervals of 6-12 months.Follow-up included contrast-enhanced computed tomography (CT) scans of the chest and abdomen, along with ultrasonography of the supraclavicular lymph nodes.If dysphagia or anastomotic leakage was suspected, electronic gastroscopy or endoscopic ultrasound (EUS) was performed.Positron emission tomography/computed tomography (PET-CT) was conducted if distant metastasis was suspected.

Study endpoint
The primary endpoints of the study comprised diseasefree survival (DFS) and overall survival (OS), with secondary endpoints encompassing pathological complete response (pCR), objective response rate (ORR), primary tumor regression, lymph node regression, perioperative complications, and patterns of recurrence.DFS was delineated as the duration from surgery to the occurrence of disease recurrence, metastasis, or mortality, while OS was delineated as the duration from surgery to death from any cause.Evaluation of efficacy adhered to the Response Evaluation Criteria in Solid Tumors (RECIST) criteria (version 1.1).Complete response (CR) denoted the disappearance of all target lesions.Partial response (PR) denoted a reduction in the sum of the longest diameters of all target lesions by 30% or more.Progressive disease (PD) denoted an increase in the sum of the longest diameters of all target lesions by at least 20%.Stable disease (SD) denoted a response between PR and PD.ORR was defined as the proportion of patients achieving CR or PR.Preoperative clinical TNM staging was conducted for all patients via imaging examination.Postoperative pathological TNM staging determined according to pathological findings.Staging adhered to the 8th edition of the UICC/AJCC TNM staging system.

Statistical analysis
Categorical variables were depicted as frequencies (%).Continuous variables following a normal distribution were expressed as mean ± standard deviation, whereas those not following a normal distribution are presented as median (IQR).The Chi-square test or Fisher's exact test was employed for comparisons among categorical variables.The t-test or Wilcoxon test was utilized for comparisons among continuous variables.To mitigate the impact of clinical factors on endpoints, multivariable logistic regression models or multiple linear regression models were employed to adjust for age, sex, tumor location, differentiation, smoking history, alcohol consumption history, ECOG performance status, clinical T stage, clinical N stage, clinical TNM stage, cycles of neoadjuvant treatment, and the regimen of chemotherapy.The Kaplan-Meier method was utilized to estimate DFS and OS, with survival differences compared using the Log-rank test.Univariate and multivariate Cox regression models were applied to analyze the influence of clinical factors on prognosis.P value < 0.05 (both sides) was deemed statistically significant.All analyses were conducted using R version 4.0.4(R Foundation for Statistical Computing, Vienna, Austria).

Baseline characteristics
From November 2019 to December 2022, 152 patients diagnosed with locally advanced ESCC underwent radical esophagectomy following neoadjuvant PD-1 inhibitor combined with chemotherapy at the Department of Thoracic Surgery of the Sun Yat-sen University Cancer Center.The median age of the included patients was 58 years (range: 41.0-78.0years).All patients underwent clinical staging by chest enhanced CT scans and endoscopic examinations, with the majority being stage II-III (stage II: 36 cases, 23.7%; stage III: 90 cases, 59.2%).The median interval from the completion of the last neoadjuvant treatment to surgery was 5.71 weeks (range: 1.29-25.0weeks).There were 91 cases in shortinterval group (time to surgery ≤ 6 weeks), with a median time to surgery of 4.71 weeks (range: 1.29-6.0weeks), and 61 cases in long-interval group (> 6 weeks), with a median time to surgery of 7.29 weeks (range: 6.14-25.0weeks).The clinical baseline characteristics of the two groups were depicted in Table 1.Short-interval group and long-interval group were well-balanced and comparable in terms of sex (P = 0.264), age (P = 0.100), tumor location (P = 0.323), smoking history (P = 0.256), alcohol consumption history (P = 0.286), body mass index (BMI) (P = 0.927), ECOG performance status (P = 0.195), histological differentiation (P = 0.942), clinical T stage (P = 0.470), clinical N stage (P = 0.486), clinical TNM stage (P = 0.315), surgical approach (P = 0.615), and cycles of neoadjuvant treatment (P = 0.830).
In terms of pathological outcomes, the proportions achieving pCR in the short-interval group and longinterval group were 34.1% and 24.6%, respectively.After adjustment, still no statistically significant disparity was

Adverse events of neoadjuvant treatment
The adverse effects of neoadjuvant treatment between shortinterval and long-interval groups were displayed in Table 3.

Surgery outcomes
As presented in Table 1

Recurrence patterns
The recurrence patterns were summarized in Supplementary

Discussion
Although the efficacy and safety of neoadjuvant PD-1 inhibitor combined with chemotherapy in locally advanced ESCC have been demonstrated in some clinical trials, the impact of the interval between immunotherapy and surgery on surgical outcomes and prognosis remains unclear.In this recent study, we compared outcomes between surgery within 6 weeks and surgery after 6 weeks following immunotherapy in the same real-world clinical practice setting.The findings revealed that delayed surgery had no significant impact on tumor downstaging, lymph node downstaging, pCR rate, and postoperative complications.However, delayed surgery was related to poorer prognosis.
Data on the optimal timing to surgery after neoadjuvant immunotherapy for ESCC are limited, but evidence on neoadjuvant chemoradiotherapy is abundant.Since the introduction of neoadjuvant chemoradiotherapy in the 1960s, it has been believed that surgery within 6 weeks is sufficient for tissue repair, resolution of tumor inflammatory response, and without causing tumor progression [18,19].The recommendation is to complete the surgical procedure within 4-6 weeks.Subsequent studies such as the NEOCRTEC5010 [20] and CROSS [21] trials also adopted a surgical regimen within 6 weeks after neoadjuvant treatment.Over the past few decades, a series of retrospective studies have evaluated the impact of timing to surgery after neoadjuvant chemoradiotherapy on histological response and prognosis.In studies by Lee et al. [22] and Azab et al. [23], the pCR rate increased with a prolonged interval between chemoradiotherapy and surgery, but longterm survival did not significantly improved.In a metaanalysis incorporating 10 cohort studies [24], it was found that delayed surgery did not further increase the pCR rate and OS, but rather increased the risk of anastomotic leakage.Klevebro et al. [25] similarly found no evidence to support extending the interval between chemoradiotherapy and surgery for esophageal cancer.Chiu et al. [26] reported that delayed surgery did not reduce surgical risk or increase the pCR rate, and survival was not improved either.The NeoRes II trial [27] is currently the only randomized controlled multi-center clinical trial comparing the standard time to surgery (4-6 weeks) with extended interval to surgery after neoadjuvant chemoradiotherapy for esophageal carcinoma, and the findings revealed that prolonging the surgical interval did not enhance pCR rate.Furthermore, There was a notable inclination towards poorer survival outcomes with an extended duration to surgery.These findings collectively advocate for prudence in deferring surgery following neoadjuvant chemoradiotherapy.
Compared to neoadjuvant chemoradiotherapy, immunotherapy has shown comparable efficacy in ESCC.Additionally, the tumor response patterns after immunotherapy are more diverse, with durable response patterns observed in many tumors [28].This leads us to considered the necessity of delaying surgery in the context of immunotherapy.While clinical trials typically planed for surgery within 6 weeks after completion of immunotherapy, in real-world scenarios, surgery often occurred beyond this timeframe due to patient-related factors or the COVID-19 pandemic.In this study, we retrospectively reviewed information from patients with ESCC who underwent neoadjuvant immunotherapy combined with chemotherapy.The results revealed that delayed surgery failed to further promote primary lesion and lymph node downstaging, instead, the pCR rate was lower, although the difference was not statistically significant.This may be attributed to the relatively small sample size of the study cohort, but this trend is noteworthy, suggested that delaying surgery may entail the risk of further local lession progression.
Liang et al. found severe tissue edema and unclear tissue interspaces occurred after immunotherapy, which may increased surgical difficulty [29].Some studies also reported that a longer waiting time after neoadjuvant treatment could lead to more severe tissue fibrosis [30].In our shortinterval and long-interval groups, there were no disparities observed between the groups concerning surgical timing or blood loss, suggested that delaying surgery did not increase surgical difficulty.Additionally, the long-interval group had numerically more lymph node dissections, which also indicated that the surgical quality of the long-interval group was at least equivalent to that of the short-interval group.Both groups also showed no differences in postoperative complications, hospitalization duration, ICU stay duration and chest drainage duration.In summary, the interval between immunotherapy and surgery did not significantly affect intraoperative procedures and postoperative recovery.
Improving prognosis is the fundamental challenge in cancer treatment.In an in vitro study involving 4T1.2 tumor-bearing mice receiving immunotherapy, prolonging the interval between immunotherapy and surgery resulted in increased mortality among surviving mice [31].This finding underscored the importance of carefully considering the timing of surgery to achieve better oncological outcomes.Similarly, in the follow-up of our study, we found that delaying surgery did not confer any OS improved and instead led to a significant decrease in DFS.We speculated that during the delay in surgery, some lesions may continue to progress, leading to worse outcomes.Additionally, in long-interval group, the pCR rate was only 24.6%, while the pCR rate in short-interval group could reach 34.1%.pCR is a significant factor for better prognosis [32], which may also lead to poorer prognosis in long-interval group.Lastly, the main reasons for delaying surgery were poor physical condition after neoadjuvant therapy or infection with COVID-19, both of which were factors influencing prognosis [33][34][35][36].
Currently, some researchers suggested to defer surgery for patients with complete clinical response (CCR) after neoadjuvant therapy for ESCC [37,38], similar to the watchand-wait strategy used in rectal cancer [39], but this lacks evidence-based support.Our study results did not support this proposition.Subgroup analysis based on TRG revealed that delaying surgery in patients with excellent pathological regression (TRG 0) actually increased the risk of death and disease recurrence.This suggested that early surgery remains the optimal choice for patients who respond well to immunotherapy.Future randomized controlled trials with expanded sample sizes are warranted to further substantiate the rationale behind the watch-and-wait strategy.
To our knowledge, this study represents the most extensive investigation to date regarding the impact of the interval between immunotherapy and surgery on pathological outcomes and prognosis.Moreover, the study is grounded in real-world data, thus mitigating selection bias.Nevertheless, several limitations persist.Firstly, due to the relatively recent initiation of neoadjuvant immunotherapy research for esophageal squamous cell carcinoma, the sample size collected from a single center remains limited, which may impact the generalizability of our conclusions.In the future, it will be necessary to combine data from other medical centers to provide more comprehensive evidence.Secondly, the immunotherapeutic agents included in this study were all PD-1 antibodies, which have similar pharmacological actions and efficacy, resulting in minimal bias.However, there was significant variability in chemotherapy regimens, primarily divided into two categories: paclitaxel plus platinum and paclitaxel plus fluorouracil.This variability arises from differences in chemotherapy drug selection by different physicians when making treatment decisions.Although we adjusted for chemotherapy regimens in our analysis, there may still be some impact on the results.Thirdly, in this study, the number of neoadjuvant treatment cycles was also inconsistent.Most patients received three cycles of treatment, but a small number of patients exceeded three cycles.This inconsistency in treatment cycles may also influence the outcomes.Lastly, the follow-up period was relatively short, necessitating longer-term follow-up to validate the long-term prognosis.

Conclusion
Delayed surgery after neoadjuvant PD-1 inhibitor combined with chemotherapy does not further improve primary tumor and lymph node downstaging or increase pCR rates; instead, it resulted in a poorer DFS.Especially for patients with a favorable response to immunotherapy, delayed surgery increases the risk of mortality and recurrence.Therefore, we recommend surgery within 6 weeks after neoadjuvant immunotherapy as prudent.

Fig. 1
Fig. 1 Kaplan-Meier survival analysis.a Comparison of OS between short-interval group and long-interval group.b Comparison of DFS between short-interval group and long-interval group

Fig. 2
Fig. 2 Cox proportional hazards model for univariate and multivariate analysis.a The association between pretreatment characteristics and OS.b The association between pretreatment characteristics and DFS

Fig. 3
Fig. 3 Kaplan-Meier survival analysis according to the tumor regression grade (TRG).a Comparison of OS between short-interval group and long-interval group for patients with TGR 0 cohort.b Comparison of OS between short-interval group and long-interval group for

Table 1
, clinical TNM stage, cycles of neoadjuvant treatment and regimen of chemotherapy.The results after adjustments still revealed that the interval between neoadjuvant immunotherapy and surgery had no significant impact on primary tumor and lymph node downstaging.
(OR: 0.956, 95%CI: 0.466-1.906,P = 1.000).Considering the impact of clinical factors and control for potential biases, adjustments were made for sex, age, tumor location, smoking status, drinking status, ECOG performance score, differentiation, clinical T stage, clinical N stage

Table 2
The clinical efficacy between short-interval group and long-interval group *The multivariate logistic regression model adjusted age, sex, tumor location, differentiation, smoking history, alcohol consumption history, ECOG performance status, clinical T stage, clinical N stage, clinical TNM stage, cycles of neoadjuvant treatment and the regimen of chemotherapy

Table 3
The adverse effects between short-interval group and long-interval group

Table 2 .
The treatment failure patterns including local recurrence, distant metastasis or both.After adjustment for clinical factors, the short-interval group and long-interval group had 2.20% and 6.56% of patients experienced