The area of residual tumor predicts esophageal squamous cell carcinoma prognosis following neoadjuvant chemotherapy

To clarify the utility of the area of residual tumor for patients with esophageal squamous cell cancer treated with neoadjuvant chemotherapy. We enrolled 186 patients with esophageal squamous cell cancer who underwent surgical resection following neoadjuvant chemotherapy at our hospital. Using digital images, we measured the area of residual tumor at the maximum plane of the specimen and divided the patient into three groups as follows: 0 (area = 0 mm2), low (area = 0–40 mm2), and high (area ≥ 40 mm2). The clinicopathological factors and prognosis were compared among these groups. The median area of the residual tumor was 15.0 mm2　(range 0–1,448.8 mm2). Compared with the 0 and low group, the high group was significantly associated with poorer recurrence-free survival (all P < .001) and overall survival (P < .001 [vs. 0] and P = .017 [vs low]). The area of residual tumor, ypN, tumor regression grade, and lymphovascular invasion were independent predictors of recurrence-free survival. By dividing the patients using a combination of the area of residual tumor and lymphovascular invasion, the high and/or lymphovascular invasion ( +) group displayed significantly poor recurrence-free survival than the 0 group and low/lymphovascular invasion ( −) group. However, there was no significant difference in the recurrence-free survival between the 0 group and low/lymphovascular invasion ( −) group. The area of residual tumor is a promising histopathological prognostic factor for patients with esophageal squamous cell cancer treated with neoadjuvant chemotherapy. Moreover, it is a possible candidate histopathological factor for postoperative chemotherapy selection.


Introduction
Esophageal cancer is the seventh most common cancer worldwide (Kelly et al. 2021). It causes more than half a million deaths in a year globally (Doki et al. 2022). The standard treatment for esophageal cancer involves multidisciplinary therapy, predominantly surgery combined with chemoradiation therapy (CRT) or chemotherapy (Shah et al. 2020). For the postoperative treatment of esophageal cancer, the CheckMate 577 trial reported on the efficacy of adjuvant therapy with nivolumab (Kelly et al. 2021). It was performed for patients with residual disease following neoadjuvant CRT, and those treated with neoadjuvant chemotherapy (NAC) were ineligible for this trial (Kelly et al. 2021). Therefore, it is unclear which patients treated with NAC would benefit from adjuvant therapy.
It is essential to predict the risk of recurrence by histopathologic evaluation following neoadjuvant therapy in the selection of adjuvant therapy for patients with esophageal cancer. The pathological complete response (pCR) of patients treated with neoadjuvant CRT is associated with better survival (Murphy et al. 2017). Meanwhile, patients with residual pathological diseases are at a high risk of recurrence, and postoperative treatment is suggested for those treated with neoadjuvant CRT (Kelly et al. 2021).
For patients with esophageal cancer, neoadjuvant pathologic TNM (ypTNM) classification or tumor regression grade (TRG) is used to evaluate the therapeutic effect of neoadjuvant therapy (Chirieac et al. 2005;Donohoe et al. 2013;Hatogai et al. 2016;Mandard et al. 1994;Yun et al. 2022). ypTNM classification is the most frequently used method (Yun et al. 2022). However, it does not reflect the prognosis (Oguma et al. 2022;Rice et al. 2016;Yun et al. 2022). By contrast, the TRG method determines the estimated pre-treatment tumor volume with the remaining post-treatment volume from surgical specimens (Nakao et al. 2019;Sakuyama et al. 2018). It is difficult to estimate the pre-treatment tumor volume, and TRG has interobserver variety (Nakao et al. 2019;Sakuyama et al. 2018). Thus, there is no consensus on the cut-off value for TRG (Yun et al. 2022). Recently, researchers have reported the residual cancer volume (RCV) and pattern of response as an alternative assessment method for neoadjuvant response (Martinez et al. 2022;Nakao et al. 2019). The RCV is calculated as the sum of all residual tumor areas. However, its application in practice is difficult because it requires the measurement of the tumor area on multiple slides.
The area of residual tumor (ART) has been reported as a prognostic factor in patients treated with neoadjuvant therapy for rectal, lung, pancreatic, and breast cancer (Eguchi et al. 2022;Kojima et al. 2009;Okubo et al. 2019;Sakuyama et al. 2018;Yamane et al, 2010). ART measures the area of residual tumor in surgical specimens (Kojima et al. 2009). It is an objective and quantitative pathological assessment method (Eguchi et al. 2022;Kameyama et al. 2021). Researchers have demonstrated the utility of ART for postoperative therapy (Eguchi et al. 2022). For patients with breast cancer treated with NAC, Eguchi et al. (2022) reported on ART-low (ART ≤ 136 mm 2 ) as well as ART-0 (ART = 0 mm 2 ) as favorable prognostic factors, and indicated that postoperative chemotherapy could be omitted in those with ART-0 and ART-low without lymph node metastasis.
We aimed to evaluate the prognostic significance of ART as a method of histopathological assessment for patients with ESCC treated with NAC and radical surgery.

Patients
We retrospectively reviewed the medical records of 233 patients with ESCC who underwent radical surgical resection following NAC between January 2015 and December 2017 at the National Cancer Center Hospital East. The exclusion criteria were as follows: chemotherapy regimen other than cisplatin plus 5-fluorouracil (CF) or docetaxel plus CF (DCF); incomplete resection; distant metastasis excluding supraclavicular lymph node metastasis; adjuvant chemotherapy; severe postoperative complications; or other advanced cancer. We enrolled 186 patients with ESCC in this retrospective study (Supplementary Information). Their clinicopathological characteristics are summarized in the Supplementary Information. The study was approved by the institutional review board of the National Cancer Center Hospital East (approval number 2020-292).

Neoadjuvant chemotherapy
The NAC consisted of either two courses of CF or three courses of DCF, with each course beginning every 3 weeks. The remaining chemotherapy was discontinued in patients without obvious treatment effects or with severe toxicity related to chemotherapy, following which they underwent surgery (Hatogai et al. 2016).

Pathological analysis
The surgically resected ESCC specimens were fixed with 10% formalin. They were cut into 5 mm slices, parallel to the long axis of the esophagus, and embedded in paraffin. We subsequently made thin sections from the paraffin-embedded block. We evaluated the hematoxylin-eosin-stained sections.

3
All residual tumors were pathologically staged according to the TNM classification of the American Joint Committee on Cancer 8th edition (Rice et al. 2017). All tumors were evaluated for lymphovascular invasion (LVI) and tumor regression grade (TRG). The TRG was classified into five categories according to the Mandard grading system (Rice et al., 2016).

ART measurement
ART was defined as the residual tumor area at the largest tumor slice on the surgical specimen, in accordance with previous reports (Kojima et al. 2009;Sakuyama et al. 2018). It was measured using NanoZoomer Digital Pathology Virtual Slide Viewer (Hamamatsu Photonics KK, Hamamatsu, Japan), an automatic digital slide scanner, and its viewer. We included degenerated tumor cells with a nucleus and cytoplasm as the residual tumor cells (Eguchi et al. 2022;Kojima et al. 2009). Moreover, we excluded intraepithelial lesions and necrotic tumor cells. If a group of residual tumor cells was 1 mm from the neighboring group, it was considered a separate residual tumor (Fig. 1a). Lesions within 1 mm were counted together as a single lesion (Fig. 1b). ART was obtained by summing all tumor nests on the slice with the largest tumor area. We plotted the receiver operator characteristic (ROC) curve to predict patient recurrence. The cutoff value for the association between recurrence and ART was based on the lower distances to the top-left corner of the ROC curve, which excluded patients with ART = 0 mm 2 . Based on the ROC curve for recurrence-free survival (RFS), the ART cut-off value was set at 40 mm 2 . We divided the patients into the following three groups according to their ART values: ART-0 (ART = 0 mm 2 ), ART-low (ART = 0 mm 2 to 40 mm 2 ) and ART-high (ART > 40 mm 2 ).

Statistical analysis
We performed between-group comparisons using Fisher's exact test and t-test for categorical characteristics and continuous variables, respectively. We used the RFS and overall survival (OS) to access the effect of ART. RFS was defined as the time from surgery to recurrence or death for any reason. OS was defined as the time between surgery and death for any reason. We compared the RFS and OS rates using the Kaplan-Meier method and Log-rank test. For the threearm comparisons, we performed Bonferroni correction and the P values were calculated using the log-rank test. We examined the univariate and multivariate between-group differences in survival using the Cox regression models. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria) (Kanda 2013). A P value < 0.05 was considered statistically significant.

ART and clinicopathological features
There were 54 patients (29.0%) with ART = 0 mm 2 . The median ART was 15.0 mm 2 (range 0-1,448.8 mm 2 ) (Fig. 2a). Figure 2b depicts the ROC curve for the RFS, excluding the patients with ART = 0 mm 2 . Figure 2c depicts the representative case around the cut-off value, ART = 42.6 mm 2 . The clinicopathological features of each ART group are shown in Table 1. The ART was significantly related to the ypT stage, ypN stage, ypM stage, ypStage, LVI, and TRG.

Clinical outcome of ESCC following NAC
The 3 year RFS rate of all patients was 67.9%. The Kaplan-Meier curves on the RFS according to the ypT stage, ypN stage, ypStage and TRG are presented in the Supplementary Information. There were significant between-group differences in the RFS among the groups in the ypT stage, ypN stage, ypStage, and TRG. The 3 year RFS rate for the ART-0, ART-low, and ART-high groups were 82.4% (95% confidence interval (CI), 68.8-90.5%), 79.9% (95% CI, 66.5-88.3%), and 43.0% (95% CI, 31.6-53.9%), respectively. The 3 year OS rate for the ART-0, ART-low, and ART-high groups were 91.9% (95% CI, 79.7-96.9%), 81.6% (95% CI, 68.4-89.7%) and 56.7% (95% CI, 43.8-67.7%), respectively. Compared with the ART-0 and ART-low groups, the ART-high group was significantly associated with poorer RFS (all P < 0.001) and OS (P < 0.001 [vs. ART-0] and P = 0.017 [vs. ART-low]) in all patients. Figure 3a and b depict the Kaplan-Meier curves for the ART on the RFS and OS in all patients. In patients with ypStage 0-II, there were no statistically significant between-group differences in the RFS and OS (Fig. 3c, d). In those with ypStage III-IV, the ART-high group displayed significantly shorter RFS than the ARTlow group (P = 0.008) (Fig. 3e). There were no significant differences in the RFS and OS between the ART-0 and ART-low groups in any patient subgroup (Fig. 3).

Prognostic factors in the univariate and multivariate analysis
The univariate analysis revealed the ART, ypT, ypN, ypM, ypStage, LVI, and TRG as the significant prognostic factors for RFS. The multivariate model that included the ART, ypT, ypN, LVI, and TRG demonstrated that ART, 1 3 ypN, TRG, and LVI were significant independent prognostic factors for RFS (Table 2). However, the hazard ratio for TRG was less than 1.0, which was reversed in the univariate analysis. It also demonstrated that the ypT was not a significant independent prognostic factor for RFS.

Clinical outcomes of ART combined with LVI
To analyze the prognosis of patients with pCR, which corresponded to ART-0, we divided the patients into two groups as follows: (i) patients in the ART-0 group and (ii) those in the ART-low or -high groups. Patients in the ART-0 group displayed significantly better RFS than those with ART-low or high (P = 0.002) (Fig. 4a).
Subsequently, we divided the patients into the following three groups using a combination of ART and LVI: group 0 (ART-0), group I (ART-low and LVI [−]), and group II (ART-high and/or LVI [ +]). Compared with group II, groups 0 and I displayed significantly better RFS (P < 0.001) (Fig. 4b); however, there were no significant differences in the RFS between groups 0 and I (P = 1.0).

Discussion
In this study, ART ≥ 40 mm 2 was significantly associated with poorer RFS and OS for patients with ESCC following NAC. The multivariate analysis indicated ART as an independent prognostic factor for RFS as well as LVI. In addition, we could perform further prognostic stratification upon using ART in conjunction with LVI. ART was useful,   particularly in identifying groups with favorable prognoses for patients with ESCC following NAC. For patients with ESCC treated with neoadjuvant therapy, pCR has been associated with better OS and RFS (Murphy et al. 2017). Recently, researchers have investigated adjuvant therapies, such as nivolumab, for these patients (Kelly et al. 2021). Kelly et al. (2021) reported on the efficacy of nivolumab for patients with resected esophageal cancer who received neoadjuvant CRT in the CheckMate 577 trial. In this trial, patients with pCR were ineligible for adjuvant treatment (Kelly et al. 2021). In contrast to patients who had received neoadjuvant CRT, it is unclear which patients with ESCC who received NAC would benefit from adjuvant therapy (Shah et al. 2021). In our study, the ART-0 group, predominantly comprising patients with pCR, displayed a significantly better prognosis in the RFS than others, similar to a previous report (Murphy et al. 2017). Furthermore, there was no significant difference between the ART-low group without LVI and the ART-0 group. Thus, patients in the ART-low group without LVI and those in the ART-0 group exhibited a low risk of recurrence. Moreover, adjuvant treatment can be omitted for these patients.
TRG is reportedly a method to evaluate the histopathological response and predict prognosis in patients with 1 3 ESCC treated with neoadjuvant therapy (Chirieac et al. 2005;Donohoe et al. 2013;Hatogai et al. 2016;Mandard et al. 1994). It is evaluated by the degree to which the tumor shrinks from its original size, which is assumed from the area of fibrosis (Nakao et al. 2019;Sakuyama et al. 2018). However, fibrosis may not reflect the original tumor area, thereby making an accurate assessment difficult and causing interobserver variability (Nakao et al. 2019;Sakuyama et al. 2018). In contrast to TRG, ART does not require assuming the area of the original tumor and is simple to evaluate. In addition, ART results are quantified, and there is less variation among examiners (Kameyama et al. 2021). Thus, we considered ART as a useful indicator of prognosis because it can be easily evaluated and generates less interobserver variability.
Researchers can evaluate the ypT stage for the primary tumor of the surgical specimen (Rice et al. 2017). However, Schneider et al. (2005) reported that the ypT stage does not predict the survival of patients with ESCC following neoadjuvant therapy. This is because the original T stage remains preserved, despite the tumor regression response (Schneider et al. 2005). These patients would be expected to have lower ART reflecting tumor shrinkage, but advanced ypT stage owing to the depth of the residual tumor (Schneider et al. 2005). In our study, 12 of 66 (18.2%) patients with ypT3 displayed ART < 40 mm 2 , thus indicating they may have a good prognosis, despite their ypT stage (Supplementary Information). The ART would adequately stratify the prognosis, particularly for patients with advanced ypT.
The disadvantage of ART is that it is difficult to be measured under a microscope; however, Matsuda et al. indicated an assessment method of ART using the number of field views at 40 × magnification (Matsuda et al. 2020). Contrarily, ART can be easily measured using a digital pathology image. Digital pathology has considerably progressed in recent years (Bankhead 2022); the importance of ART, which can be easily measured on digital slides, will presumably increase in the future.
This study had several limitations. First, NAC was the type of neoadjuvant therapy in this study. CRT is widely used as the standard neoadjuvant therapy for ESCC (Kelly et al. 2021). However, the JCOG1109 trial reported on the utility of NAC, indicating NAC may become the standard neoadjuvant therapy for patients with ESCC . Second, it was based on a retrospective single-center design. Third, the median follow-up of 35 months was relatively short, despite most recurrences patients occurring within the first 2 years following surgery (Hatogai et al. 2016). This necessitates additional follow-up to obtain more reliable results. Therefore, further multicenter studies are necessary to clarify the utility of ART in these patients.
In conclusion, our present findings indicated that ART can be a useful prognostic factor for patients with ESCC following NAC. This novel study revealed the significant prognostic impact of ART in patients with ESCC. In addition, further stratification of prognosis was possible upon combining ART with LVI. Patients in the ART-low group without LVI and those in the ART-0 group displayed a favorable prognosis, which was useful in selecting adjuvant therapy in these patients following NAC. We did not identify the patients requiring postoperative treatment following NAC; this necessitates histopathological evaluation for determining the risk of recurrence.