18F‐FDG PET/CT predicts the role of neoadjuvant immunochemotherapy in the pathological response of esophageal squamous cell carcinoma

Abstract Background This study aimed to investigate the predictive value of 18F‐FDG PET/CT for pathological response after neoadjuvant immunochemotherapy (NICT) in patients with esophageal squamous cell carcinoma (ESCC). Methods The clinical data of 54 patients with ESCC who underwent two cycles of NICT followed by surgery were retrospectively analyzed. NICT consisted of PD‐1 blockade therapy combined with chemotherapy. 18F‐FDG PET/CT scans were performed before and after NICT. The pathological results after surgery were used to assess the degree of pathological response. The scan parameters of 18F‐FDG PET/CT and their changes before and after NICT were compared with the pathological response. Results Among the 54 patients, 10 (18.5%) achieved complete pathological response (pCR) and 21 (38.9%) achieved major pathological response (MPR). The post‐NICT scan parameters and their changes were significantly associated with the pathological response. In addition, the values of the changes in the scanned parameters before and after treatment can further predict the pathological response of the patient. Conclusion 18F‐FDG PET/CT is a useful tool to evaluate the efficacy of NICT and predict pathological response in patients with ESCC. The post‐NICT scan parameters and their changes can help identify patients who are likely to achieve pCR or MPR.


INTRODUCTION
Previous studies have demonstrated the superiority of neoadjuvant chemoradiotherapy for the treatment of locally advanced esophageal squamous cell carcinoma (ESCC). 1,2 However, this modality is associated with high rates of adverse effects and perioperative mortality. 3 Recently, several studies have reported promising results of neoadjuvant chemotherapy combined with programmed cell death 1 (PD-1) blockade, which achieved significant tumor regression and low perioperative mortality. 4,5 Despite the widespread use of minimally invasive esophagectomy, postoperative mortality remains high and the patients' quality of life is significantly impaired. For patients who achieve a complete pathological response (pCR), surgery or additional neoadjuvant therapy may be unnecessary. Patients who do not achieve a major pathological response (MPR) may be resistant to neoadjuvant immunochemotherapy and require prompt surgery. The Shuohua Wang, Shouyin Di and Jing Lu contributed equally to this work and are co-first authors. identification of patients who achieve pCR and those who do not achieve MPR after neoadjuvant immunochemotherapy remains an unresolved issue. 18 F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) is a common technique for radiological staging of advanced ESCC 6 as it can measure the intensity of FDG uptake and reveal the changes in tumor metabolism and viability regardless of the underlying structural changes. 7 PET imaging technology has improved a lot in recent years, not only in imaging equipment and software, but also in imaging probes and contrast agents. Schwenck et al. reviewed the latest uses of PET imaging in cancer, such as new radioactive tracers, multimodal imaging, artificial intelligence, and personalized medicine and immunotherapy. 8 These improvements may make PET imaging more accurate and useful in cancer diagnosis and treatment. Several studies have explored the role of PET/CT in various aspects of esophageal cancer, such as predicting the response to neoadjuvant chemotherapy, 9 detecting postoperative tumor recurrence, 10 and assessing other outcomes. 11 However, the role of PET/CT in predicting the pathological response after neoadjuvant immunochemotherapy is still under-researched. Wang et al. conducted a retrospective study and confirmed that 18 F-FDG PET/CT parameters can accurately predict pCR after neoadjuvant immunochemotherapy in resectable ESCC. 12 Besides identifying patients who achieve pCR, another clinical challenge is identifying patients who do not achieve MPR. For patients who achieve pCR, surgery may be avoided; for patients who achieve MPR, another two cycles of neoadjuvant immunochemotherapy may be considered; for patients who are resistant to neoadjuvant immunochemotherapy, surgery should be performed as soon as possible. The optimal management of patients who achieve pCR is still a matter of debate, but finding a noninvasive method to assess pathological response is crucial.
This real-world study collected data from patients who underwent neoadjuvant immunochemotherapy and esophagectomy and investigated the role of 18 F-FDG PET/CT parameters in assessing the pathological response of ESCC.

Study design
This was a single center prospective real-world study, approved by the Ethics Committee of the Sixth Medical Center of PLA General Hospital. The study included patients with esophageal squamous cell carcinoma who underwent neoadjuvant immunochemotherapy and esophagectomy at the Department of Thoracic Surgery, the sixth Medical Center of the Chinese PLA General Hospital, from January 2020 to December 2022. The patients underwent PET/CT examinations according to their clinical treatment assessment and provided informed consent for all treatments and examinations. We selected the pathological response after surgery as our primary endpoint, and evaluated it according to the Mandard tumor regression grade (TRG) system. All patients continued treatment and follow-up after the experimental observation endpoint.

Eligibility criteria
The inclusion criteria were: age between 18 and 80 years; histologically or cytologically confirmed esophageal squamous carcinoma; and no prior radiotherapy, chemotherapy, or surgery for the tumor. The exclusion criteria were: prior immunotherapy or known allergy to monoclonal or chemotherapeutic agents; history of other critical illnesses such as multiorgan failure, widespread infection, shock, stroke, or any vascular embolic event within 6 months before enrollment; severe malnutrition; or other primary malignancies. Patients who did not receive a PET/CT scan 1 week before surgery were also excluded from the analysis.

Treatment regimen
The patients received two cycles of neoadjuvant immunochemotherapy (NICT) before surgery, tailored to their individual conditions. The NICT consisted of intravenous lobaplatin (40 mg) and paclitaxel (400 mg) on day 1, and intravenous PD-1 monoclonal antibody (200 mg) on day 2. The PD-1 monoclonal antibodies used were sintilimab, camrelizumab, or tislelizumab. The NICT was repeated on day 15, and surgery was performed 2 weeks after the second cycle. The surgical procedure was esophagectomy with lymph node dissection.

PET/CT imaging
The patients underwent PET/CT imaging using a GE Discovery ST16 scanner with 18 F-FDG as the tracer (provided by China Atomic High Tech). The patients fasted for 6 h before the scan and had their blood glucose levels checked (4.5-6.5 mmol/L). They received an intravenous injection of 18 F-FDG (185-370 MBq) based on their body mass and rested for about an hour before the scan. The scan covered the region from the cranial apex to the mid-femur, starting with a fluoroscopic acquisition, followed by a CT scan (150 mA, 120 kV), and then a 3D emission acquisition (23 min/bed, matrix 128 Â 128). The CT data were used for attenuation correction and reconstructed using ordered subsets expectation maximization (OSEM). The fused images were processed with GE AW workstation software.

PET/CT data analysis
Two experienced nuclear medicine physicians visually assessed the PET/CT images to determine the extent and morphology of the lesions and the areas of increased FDG uptake. The PET/CT scans of patients before neoadjuvant immunochemotherapy (scan 1) and before surgery (scan 2) were recorded and compared. Semi-quantitative analysis was performed using volumetrics to measure the maximum standardized uptake value (SUV max ), mean standardized uptake value (SUV mean ), MTV, and TLG of the lesions at different sites. The SUV max of the blood pool at the aortic arch (10 mm diameter) without including the vessel wall was measured, and the ratio of the SUV max of the lesion to the SUV max of the blood pool was calculated as SUV TBR . The changes in these parameters before and after treatment are expressed as percentages; for example, ΔSUV max % = (Scan 1 SUV max -Scan 2 SUV max )/Scan 1 SUV max Â 100%.

Pathological diagnosis
The pathological response was assessed by measuring the percentage of viable tumors in the primary tumor lesions using hematoxylin-eosin (HE) staining. The specimens were independently analyzed by two senior pathologists. The patients were classified into four groups based on the tumor regression score (TRS) according to the Mandard scoring system: pCR (TRS 1, no residual tumor cells in the primary lesion and lymph nodes), non-pCR (TRS 4-5, minimal or no tumor regression), MPR (TRS 1-2, complete or subtotal tumor regression), and non-MPR (TRS 3-5, partial, minimal, or no tumor regression).

Swimmer plot
We recorded the duration of treatment for each patient, and created a swimmer plot based on their imaging, pathological, and pathological response data using Microsoft Excel. A swimmer plot is a graphical tool that can display the temporal dynamics of individual patients in a clinical trial. The plot can visually show the distribution and association of imaging and pathological results among patients.

Statistical analysis
The data were collected according to the actual condition of the patients and summarized by frequency and proportion for categorical variables. For numerical variables, descriptive statistics are reported as mean ± standard deviation or median (interquartile range), depending on whether the data followed a normal distribution, which was tested by the Shapiro-Wilk normality test. The homogeneity of variance was tested by Levene's test (based on mean). The comparison between two groups was performed by t-test, Welch's ttest, or Wilcoxon test depending on whether the data met the normal distribution and chi-square test assumptions. For categorical variables, if the condition of 5 > theoretical frequency ≥1 and total sample size ≥40 was satisfied, the continuous corrected chi-square test (Yates' correction) was used for comparison between groups. The patients were divided into pCR and non-pCR groups, and MPR and non-MPR groups based on postoperative pathological findings. Receiver operating characteristic (ROC) curves were used to evaluate the diagnostic performance of each metabolic parameter index. The area under the ROC curve (AUC), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy with their 95% F I G U R E 1 The flow chart of the study. ESCC, esophageal squamous cell carcinoma; MPR, major pathological response; pCR, pathological complete response; PET/CT, positron emission tomography/computed tomography.
confidence intervals (95% CI) were calculated. The critical values for ROC curve analysis were calculated based on the Youden index. The differences in ROC curves were tested by DeLong's test. All data analyses were performed using R software (version 4.2.1) and the corresponding R package.

Clinical characteristics of the patients
This study enrolled 63 patients with esophageal squamous cell carcinoma between January 2020 and December 2022 and excluded nine patients who had prior treatment or did not undergo PET/CT before surgery. The final sample comprised 54 patients who received two cycles of neoadjuvant immunochemotherapy followed by esophagectomy ( Figure 1). Tables 1 and 2 display the baseline characteristics of the patients by pathological response groups. No significant differences were observed between the groups in terms of age, gender, smoking, alcohol, tumor site, pretreatment stage, or treatment regimen. The metabolic parameters of PET/CT scans before (scan 1) and after (scan 2) neoadjuvant treatment were analyzed using descriptive and inferential statistics. Tables 3 and 4 present the results of the statistical tests for the metabolic parameters by pathological response groups. No significant differences were found in scan 1 parameters between the groups. However, in scan 2, SUV max , SUV mean , and SUV TBR were significantly lower in the pCR group than in the non-pCR group (p < 0.05), and also lower in the MPR group than in the non-MPR group T A B L E 1 Patient characteristics in the pCR and non-pCR groups.

Characteristics
All patients (n = 54) pCR (n = 10) Non-pCR (n = 44) p-value (p < 0.05). Moreover, the changes in metabolic parameters from scan 1 to scan 2 (Δmetabolic parameters) differed significantly between the groups, except for ΔMTV. ΔSUV max , ΔSUV mean , ΔSUV TBR , and ΔTLG were significantly lower in the pCR group than in the non-pCR group (p < 0.05) and also lower in the MPR group than in the non-MPR group (p < 0.05).

Treatment and analysis of the subjects
In addition, we created a swimmer plot based on the treatment status of 54 patients and found that the tumor regression status measured by PET/CT examination was consistent with the tumor regression status reported by pathological analysis (Figure 2). The swimmer plot shows the time span from each patient's first admission and treatment to discharge after surgery, as well as the start time of the first cycle of neoadjuvant immunotherapy, the tumor regression status after two complete treatment cycles, the time of surgery, and the tumor regression status based on pathological analysis. The tumor regression status is indicated by different colors and shapes: red for complete tumor regression, blue for major tumor regression, and black for no tumor regression.
PET/CT parameters were significantly different in different pathology subgroups The metabolic parameters of PET/CT scans before (scan 1) and after (scan 2) neoadjuvant treatment were compared T A B L E 2 Patient characteristics in MPR and non-MPR groups.

Characteristics
All patients (n = 54) MPR (n = 21) Non-MPR (n = 33) p-value Note: Mean ± SD is used for those with normal distribution, and median (IQR) is used for those who do not conform to normal distribution. MTV, metabolic tumor volume; SUV max , maximum standardized uptake value. SUV mean , mean standardized uptake value. SUV TBR , the tumor-to-blood pool SUV max ratio; TLG, total lesion glycolysis. n a were patients who underwent PET/CT scan before and after neoadjuvant therapy, including six pCR patiernts and 25 non-pCR patients. n b were patients who underwent PET/CT scan before operation, including 10 pCR patiernts and 44 non-pCR patients.
T A B L E 4 The characteristics of PET/CT parameters of major pathological response. Note: Mean ± SD is used for those with normal distribution, and median (IQR) is used for those who do not conform to normal distribution. MTV, metabolic tumor volume; PET/CT, positron emission tomography/computed tomography; SUV max , maximum standardized uptake value; SUV mean , mean standardized uptake value; SUV TBR , the tumorto-blood pool SUV max ratio; TLG, total lesion glycolysis. n a were patients who underwent PET/CT scan before and after neoadjuvant therapy, including 12 MPR patients and 19 non-MPR patients. n b were patients who underwent PET/CT scan before operation, including 21 MPR patients and 33 non-MPR patients.

Characteristics
between different pathological response groups using t-tests. Tables 3 and 4 show the means, standard deviations, and p-values of the metabolic parameters by group. No significant differences in scan 1 parameters were found between the groups (p > 0.05). However, in scan 2, the pCR group had significantly lower SUV max , SUV mean , and SUV TBR than the non-pCR group (p < 0.05), and the MPR group had significantly lower SUV max , SUV mean , and SUV TBR than the non-MPR group ( p < 0.05). The changes in metabolic parameters from scan 1 to scan 2 (Δmetabolic parameters) were also calculated and showed that ΔSUV max , ΔSUV mean , ΔSUV TBR , and ΔTLG were significantly lower in the pCR group than in the non-pCR group (p < 0.05), and also lower in the MPR group than in the non-MPR group ( p < 0.05). The only exception was ΔMTV, which did not differ significantly between the groups (p > 0.05).

Predictive value of metabolic parameters for complete disease remission
To examine the predictive value of metabolic parameters for pCR, ROC analysis was performed using scan 1 and scan 2 parameters (Table 5, Figure 3a-c). In scan 1, only SUV TBR (AUC = 0.760) showed a significant ability to discriminate between pCR and non-pCR patients (Figure 3a). In scan 2, SUV max (AUC = 0.857), SUV mean (AUC = 0.891), and SUV TBR (AUC = 0.870) were all significant predictors of pCR. Among them, SUV mean had the highest AUC value (AUC = 0.891, 95% CI: 0.795-0.987) with a cutoff value of 2.188, resulting in a PPV of (9/54) 47.3% and a NPV of (34/54) 97.1% (Table 3, Figure 3b). No significant differences were found between the ROC curves of these parameters using DeLong's test. The changes in metabolic parameters from scan 1 to scan 2 (Δmetabolic parameters) were also analyzed for their predictive value for pCR (Figure 3c). ΔSUV max (AUC = 0.840), ΔSUV mean (AUC = 0.913), and ΔSUV TBR (AUC = 0.893) were all significant predictors of pCR.

Predictive value of metabolic parameters for disease remission
The ROC analysis for identifying MPR yielded similar results to those for identifying pCR ( Table 6, Figure 5d-f).

DISCUSSION
This study investigated the prognostic significance of 18 F-FDG PET/CT for the pathological response of ESCC following neoadjuvant immunochemotherapy. In contrast to Wang et al., this study employed multiple PD-1 inhibitors, which enhanced its validity. The study retrospectively analyzed 54 patients and examined the association between 18 F-FDG PET/CT and pathological response, with the aim of identifying patients with T A B L E 5 Different metabolic parameters on predicting tumor pathological complete response.  pCR, MPR, and non-MPR after neoadjuvant immunochemotherapy. The results indicated that SUV max , SUV mean in scan 2 and ΔSUV max , ΔSUV mean , ΔSUV TBR had high predictive accuracy for the pCR and MPR of ESCC. This study showed that SUV max (AUC 0.857, cutoff value 3.20), SUV mean (AUC 0.891, cutoff value 2.188) in scan 2, and ΔSUV max (AUC 0.84, cutoff value 74.826), ΔSUV mean (AUC 0.913, cutoff value 81.568), and ΔSUV TBR (AUC 0.893, cutoff value 79.931) had a significant association with pCR with sensitivity of 90%, 90%,100%, 100%, 100%, specificity of 81.8%, 77.3%, 72.0%, 80.0%, 80.0%, and PPV of 52.9%, 47.4%,46.2%, 54.5% 54.5%, respectively. These results confirmed that 18 F-FDG PET/CT is a reliable method to evaluate pathological response after NICT. The role of SUV max was inconsistent in previous studies. Some studies reported that SUV max of 18 F-FDG PET/CT was not a strong predictor of pathological response of ESCC after NCRT due to esophagitis. 13 However, Sasaki et al. found a correlation between pathological response and SUV max after nCRT. 14 Wang et al. concluded that SUV max and SUV mean were adequate predictors of pCR after NICT (AUC 0.848), 12 which was in agreement with the present study (AUC 0.857). Regarding SUV mean , Hu et al. found that SUV mean was significantly higher in nonresponders than in responders. Receiver operating characteristics curve analysis identified SUV mean (AUC = 0.870, p = 0.010) as significant predictors of the response to concurrent chemoradiotherapy. 11 Volumebased parameter SUV TBR might have better predictive performance in ESCC after NICT. To the best of our knowledge, only the study by Wang et al. investigated SUV TBR in ESCC. In their study, SUV TBR (AUC 0.860) was better than ΔSUV TBR (AUC 0.668). However, our study showed that ΔSUV TBR (AUC 0.893) was better than SUV TBR (AUC 0.870) because of higher specificity (80% vs. 68.2%). The above differences between the study by Wang et al. and our study might be due to three aspects. 12 First, our study was retrospective and the study by Wang et al. was prospective. Second, we enrolled patients receiving sarilumab, tirelizumab and camrelizumab, while only camrelizumab was used in the study by Wang et al. NICT has brought remarkable changes to ESCC treatment for its impressive pCR and MPR rate and tolerable adverse effect rate. [15][16][17] For patients with MPR after two cycles of NICT, pCR may be achieved after another two cycles of NICT. Distinguishing MPR from non-MPR is another clinical challenge. To the best of our knowledge, this is the first study to report the role of 18 F-FDG PET/CT in predicting the major pathological response to NICT of resectable ESCC. In this study, we further evaluated the predictive performance of 18   Previous studies have examined the association between PET/CT and major pathological response. 18,19 However, the reported accuracy to predict major pathological response differs significantly among studies. In most studies, the SUV max on pre-or post-treatment 18 F-FDG PET/CT, as well as a (relative) change in SUV max , were analyzed. However, SUV max or reduction of SUV max had poor predictive performance in ESCC patients. [20][21][22] Simoni et al. found that MTV and TLG before IC (scan 1), and SUV mean and TLG after IC (scan 2) were related to MPR in ESCC patients who received NCRT. 23 The study by Stiekema et al. showed that SUV max , MTV, and TLG played a role in identifying MPR, but the AUC of ROC in predicting MPR was 0.70 (95% CI: 0.65-0.92) for ΔSUV max , and 0.73 (95% CI: 0.58-0.88) for both MTV and TLG. 24 The study by Stiekema et al. concluded that the accuracy for predicting a complete or major pathological response was limited. 24 The conflicting results might be due to various factors, such as small sample size, esophagitis after NCRT and systemic errors. The study by Wang et al. focused on pCR and lymph nodes. 12 Few studies have investigated the MPR predictive performance of 18 F-FDG PET/CT in ESCC patients who received NICT.
Identifying pCR and MPR is a clinical challenge. For patients with pCR, active surveillance may be sufficient. For patients with MPR, they may achieve pCR after another 1-2 cycles of NICT. Whether surgery is necessary for patients with pCR should be confirmed by further randomized controlled trials. However, it is crucial to find an accurate assessment of residual disease after NICT. A previous study and our study demonstrated the excellent performance of 18 F-FDG PET/CT for predicting pCR in resectable ESCC after NICT. 12 Moreover, 18 F-FDG PET/CT is also a good predictor of MPR as shown in this study. Future clinical trials are needed to determine SUV threshold and cutoff value for pCR and MPR assessment.
Furthermore, we noticed that our analysis results were not entirely consistent with some previously published articles. 20,25,26 We concluded that was related to the selection of the study population and the treatment regimen. We only included Chinese patients with ESCC in our study and excluded adenocarcinoma patients. In addition, we inferred that radiotherapy-induced esophagitis and inflammation may cause false-positive PET/CT results and statistical errors. [27][28][29] There are some limitations in this study. First, it was a single-center retrospective study with a small sample size. Second, we enrolled patients who received multiple PD-1 inhibitors, including sarilumab, tirelizumab and camrelizumab. A univariate randomized controlled study would make the conclusions more convincing. Third, predicting lymphatic metastasis is important for ESCC staging. We will evaluate the predictive performance of 18 F-FDG PET/CT for lymphatic metastasis in future research.
In summary, this study showed that 18 F-FDG PET/CT is an excellent tool to predict pCR and MPR after NICT in patients with locally advanced ESCC. Future studies are still required to investigate the detailed SUV threshold and cutoff value of 18 F-FDG PET/CT. Additionally, studies with longer follow-up time will elucidate the role of 18 F-FDG PET/CT in predicting the overall survival of ESCC after NICT. We also expect that more high-quality and highsensitivity PET imaging probes and methods will provide more accurate and comprehensive diagnosis and evaluation for ESCC patients in the future.
AUTHOR CONTRIBUTIONS SW, SD and JL: designed the study, analyzed experimental data and drafted the manuscript. SX: participated in the experiment and manuscript preparation. ZY: analyzed experimental data. TG and YL: conceived the study, evaluated the data, and prepared the manuscript. All authors contributed to the article and approved the submitted version.