Utility of endoscopic ultrasonography elastography as a predictive tool for post-treatment recurrence of resectable pancreatic cancer treated with neoadjuvant and adjuvant chemotherapy: a retrospective study in Japan Authors: Yuichi Kano, Takuya Ishikawa, Kentaro Yamao, Yasuyuki Mizutani, Tadashi Iida, Yoshihisa Takada, Takeshi Yamamura, Kazuhiro Furukawa, Hiroki Kawashima Published: 2026 ABSTRACT AbstractBackground/AimsPost-treatment recurrence frequently occurs even in patients with resectable pancreatic cancer (PC) who receive neoadjuvant chemotherapy (NAC), surgery, and adjuvant chemotherapy (AC). This study aimed to evaluate whether endoscopic ultrasonography elastography (EUS-EG) of PC lesions is useful in predicting post-treatment recurrence in such patients.MethodsWe retrospectively analyzed 44 of 164 consecutive patients with resectable PC who underwent NAC, surgery, and AC after EUS-EG. The endpoints were associations between EUS-EG stiffness classification (soft/stiff) and pathological response, post-treatment recurrence, and recurrence-free survival (RFS).ResultsIn the soft vs. stiff groups, the median pathological response rate was 30% (20%–75%) vs. 20% (10%–65%) (p=0.022), post-treatment recurrence rate was 25% vs. 59% (p=0.042), and the median RFS was NR (not reached) days vs. 683 days (p=0.039). In the multivariate analysis, stiffness classification was the only factor significantly associated with RFS (hazard ratio, 4.2; p=0.025).ConclusionsTo the best of our knowledge, this is the first study to indicate that EUS-EG stiffness classification may predict recurrence in resectable PC treated with NAC, surgery, and AC. Patients with stiff lesions on EUS-EG tend to have a higher risk of recurrence and may require more careful post-treatment follow-up and novel treatment strategies.Keywords: Adjuvant chemotherapy, Elasticity imaging techniques, Endoscopic ultrasonography, Neoadjuvant therapy, Pancreatic neoplasms, Recurrence Graphical abstract INTRODUCTION Post-treatment recurrence is frequently observed in patients with resectable pancreatic cancer (PC) receiving adjuvant chemotherapy (AC). Recently, neoadjuvant chemotherapy (NAC) has been increasingly adopted in real-world clinical practice to treat resectable PC.1 However, NAC is not uniformly effective, and some patients experience relapse despite completing planned treatment with NAC, surgery, and AC.2 Therefore, identifying the risk factors for recurrence is critical; nonetheless, they remain poorly defined. Recent studies indicate that carbohydrate antigen 19-9 response and post-treatment tumor size after NAC may predict recurrence; however, the available evidence remains limited.3,4 Contrast-enhanced computed tomography (CE-CT) is routinely used for the diagnosis and management of PC; however, its association with post-treatment recurrence has not been fully investigated. Endoscopic ultrasonography elastography (EUS-EG) has recently emerged as a useful tool for differentiating between benign and malignant pancreatic masses. However, no studies have evaluated whether EUS-EG findings in resectable PC lesions are predictive of post-treatment recurrence. Accordingly, this study aimed to investigate whether the EUS-EG stiffness classification (stiff/soft lesion) could serve as a predictor of recurrence in resectable PC treated with NAC, surgery, and AC, and to compare its predictive value with the contrast enhancement pattern on CE-CT. METHODS Study design This single-center retrospective study was conducted at Nagoya University Hospital. Patients Patients with resectable PC who underwent EUS at Nagoya University Hospital between April 2015 and August 2023 were included. Those who received EUS-EG prior to NAC and subsequently underwent NAC, surgery, and AC were included. All included patients were pathologically diagnosed with PC based on surgical specimens. This study focused exclusively on conventional pancreatic adenocarcinoma, excluding non-adenocarcinoma histologies such as squamous cell carcinoma, acinar cell carcinoma, and invasive carcinoma arising from an intraductal papillary mucinous neoplasm (IPMN). Resectability before NAC was determined using CE-CT according to the National Comprehensive Cancer Network guidelines.5 Tumor staging was based on the tumor-node-metastasis classification of the Union for International Cancer Control (UICC).6 Tumor response to NAC was evaluated using the response evaluation criteria in solid tumours criteria on CE-CT.7 Inclusion criteria were defined as patients who received two cycles of NAC with gemcitabine plus S-1 (GS) and four cycles of AC with S-1. The exclusion criteria included patients who underwent chemoradiotherapy prior to surgery, received NAC regimens other than GS, received fewer than two cycles of NAC owing to adverse events, underwent three or more cycles of NAC, received AC for less than 3 months or more than 1 year, or had other concomitant malignancies. The GS regimen was selected in accordance with the recommendations of the Japan Pancreas Society guidelines8; patients receiving alternative NAC regimens were excluded. CE-CT Multiphasic multidetector computed tomography was performed using a 64-slice CT scanner (Aquilion 64; Toshiba Medical Systems Corporation). Scans were obtained before contrast administration and during the arterial-dominant, pancreatic parenchymal, portal venous, and delayed phases at 25, 40, 70, and 180 seconds, respectively, following the start of contrast injection. The CT images of each patient were independently evaluated by two radiologists with 5 and 10 years of experience in pancreatic imaging, respectively. PC is generally hypovascular; thus, pancreatic tumors typically appear as hypovascular lesions during the pancreatic parenchymal phase (PPP). In contrast, the delayed phase often demonstrates enhancement due to the retention of contrast in the abundant fibrous stroma. Given that PPP reflects intratumoral blood flow, the lesions in this study were categorized as hypervascular, isovascular, or hypovascular relative to the surrounding pancreatic parenchyma based on their appearance during PPP. Procedure of EUS and EUS-EG EUS was performed by experts with over 10 years of experience or by trainees under expert supervision. The vital signs of patients were continuously monitored in the left lateral position under sedation with midazolam. The lesion diameter was measured based on EUS images. After conventional EUS, strain elastography was performed using an ELST system (Olympus). The region of interest (ROI) was defined to encompass the entire lesion along with sufficient surrounding pancreatic parenchyma, ensuring that the lesion occupied less than 50% of the total ROI. This threshold was adopted because strain elastography requires adequate reference tissue surrounding the target to enable accurate calculation of the relative strain, and similar criteria have been used in prior pancreatic elastography studies. Strain elastography assessed relative tissue stiffness based on the strain generated by the aortic pulse wave. The stiff, intermediate, and soft tissues were color-coded as blue, green, and red, respectively. Because the EUS-EG color map changes dynamically, videos of a few seconds were recorded, from which still images of the lesions were extracted. This process was repeated as required. Consistent with our previous study,9 each EUS-EG image was classified as blue-dominant, equivalent, or green-dominant based on the blue-to-green ratio within the lesion. In PC, which is typically characterized by abundant fibrosis, comparisons were made between blue- and green-dominant images, rather than between blue- and red-dominant images. Lesions were classified as stiff if the majority of EUS-EG images were blue-dominant and all remaining lesions were classified as soft. Quantitative elastography methods such as strain ratio, histogram analysis, or shear-wave measurements were not used because they were not routinely available at our institution during the study period and are known to be influenced by ROI placement and respiratory motion in deep abdominal organs. To minimize subjectivity, the EUS-EG classifications were independently reviewed by two experienced endosonographers, and discrepancies were resolved by consensus. Outcome measurement and definition The endpoints of this study were the correlation between the EUS-EG stiffness classification (stiff/soft lesion) of PC lesions and the pathological response rate, post-treatment recurrence rate, and recurrence-free survival (RFS), as well as the identification of factors associated with RFS. The relationship between CE-CT findings and post-treatment recurrence was also evaluated. The pathological response rate was defined as the proportion of tumor cells that were necrotic or absent following chemotherapy, according to the surgical pathology reports assessed by two experienced pathologists. The pathological response to chemotherapy was further categorized according to Evans classification: grade I (<10% tumor cell destruction), grade IIa (10%–50% tumor cell destruction), grade IIb (51%–90% tumor cell destruction), grade III (<10% viable tumor cells), and grade IV (no viable tumor cells). Post-treatment recurrence was defined as recurrence occurring after completion of the entire treatment sequence, including NAC, surgery, and AC. RFS was defined as the interval between the date of surgery and the date of recurrence. The observation period began on the date of the surgery. For patients without recurrence at the end of follow-up, the observation period was censored on June 30, 2024. For patients who experienced recurrence during the follow-up period, the date of recurrence marked the end of the observation period. The date of transfer was used as the observational endpoint for patients transferred to other institutions during the follow-up period. Chemotherapy NAC and AC were administered in accordance with the guidelines of the Japan Pancreas Society.8 For NAC, gemcitabine (1,000 mg/m²) was administered intravenously on days 1 and 8 of each cycle, and S-1 (80 mg/m²/day) was orally administered for 14 consecutive days (days 1–14) every 3 weeks, with the cycle repeated twice. For AC, S-1 (80 mg/m²/day) was orally administered for 28 consecutive days (days 1–28) every 6 weeks, and the cycle was repeated four times. Evaluation of EUS-EG images EUS-EG images were independently reviewed by two readers (Y.K. and T.I.), who were blinded to the patients’ clinical backgrounds, radiographic findings, and pathological data. Evaluation was based on still images. Interobserver variability in the EUS-EG stiffness classification (stiff/soft lesions) was assessed using the kappa coefficient. Agreement was interpreted as follows: minor (0.01–0.20), fair (0.21–0.40), moderate (0.41–0.60), good (0.61–0.80), or excellent (0.81–1.00). Images with discrepant assessments were jointly reviewed by the two reviewers to reach a consensus. Statistical analysis To evaluate differences in patient characteristics, continuous variables were analyzed using the Mann-Whitney U-test because of their non-normal distribution, and categorical variables were analyzed using the chi-square test. RFS was estimated using the Kaplan-Meier method, and comparisons were performed using the log-rank test. Univariate and multivariate analyses were conducted to identify the factors significantly associated with RFS using the Cox proportional hazards model. Variables with a p-value <0.1 in the univariate analysis were included in the multivariate analysis. Statistical significance was defined as a two-sided p-value of <0.05. Statistical analyses were performed using EZR software ver. 1.56 (Saitama Medical Center, Jichi Medical University).10 Ethics statements This study was approved by the Ethics Review Committee of Nagoya University Hospital (approval number: 2015-0316) and was performed in accordance with the Declaration of Helsinki. Informed consent was obtained using an opt-out approach, which was publicly posted on the hospital website to allow patients the opportunity to decline participation. RESULTS In total, 164 consecutive patients were included in this study. Of these, 120 patients were excluded for the following reasons: 25 patients received chemoradiotherapy before surgery, 61 underwent NAC regimens other than GS, 18 received fewer than two cycles of NAC owing to adverse events, one received three or more cycles of NAC, nine underwent AC for 3 months or less, four underwent AC for 1 year or longer, and two had other concomitant cancers. Ultimately, 44 patients were included in the analysis. A total of 230 EUS-EG images were analyzed, with a median of five images acquired per patient. Based on the EUS-EG classification (stiff/soft lesion), 12 patients were categorized as soft and 32 as stiff lesions (Fig. 1). The interobserver agreement between the two readers for the EUS-EG stiffness classification was excellent, with a kappa coefficient of 0.823. Among the 44 patients, IPMN was present in 11 patients (25%), all of whom had low-grade malignancy based on imaging studies. Flow chart of patients recruited in this study. PC, pancreatic cancer; EUS-EG, endoscopic ultrasonography elastography; NAC, neoadjuvant chemotherapy; AC, adjuvant chemotherapy; GS, gemcitabine. Because EUS-EG was performed in all 164 patients, we compared the baseline characteristics between the included and excluded cohorts. As shown in Table 1, no significant differences were observed across all baseline variables, suggesting a minimal selection bias between the two groups. Among the 44 patients included in this study, 12 (27.3%) were classified as having soft lesions, and 32 (72.7%) as having stiff lesions. Among the patients excluded because of treatment intolerance, 18 were excluded because of incomplete NAC and 9 because of incomplete AC. Of the 18 patients with incomplete NAC, five (27.8%) were classified as having soft lesions and 13 (72.2%) as having stiff lesions. Similarly, among the nine patients with incomplete AC, three (33.3%) had soft lesions and six (66.7%) had stiff lesions. The distribution of soft and stiff lesions among the excluded patients was comparable to that among the included patients. Table 1.Baseline characteristics between included vs. excluded patients in this studyCharacteristicIncluded (n=44)Excluded (n=120)p-valueAge (yr)72.5 (38–82)71 (44–88)0.332Sex0.829 Male26 (59.1)75 (62.5) Female18 (40.9)45 (37.5)Tumor location0.63 Head29 (65.9)72 (60.0) Body12 (27.3)34 (28.3) Tail3 (6.8)14 (11.7)Tumor size (mm)20 (11–36)21 (9–32)0.25CEA (ng/mL)2.6 (0.5–15.8)2.75 (0.7–108.6)0.653CA19-9 (U/mL)73.5 (2–1,721)142 (1–4,911)0.264Jaundice10 (22.7)18 (15.0)0.352Resectability before NAC1 R44 (100.0)120 (100.0) BR0 (0)0 (0) UR0 (0)0 (0)Stage before NAC (IA/IB/IIA/IIB/III/IV)25/19/0/0/0/057/63/0/0/0/00.378Stiffness classification (soft/stiff)12/3241/790.517Values are presented as median (range) or number (%).CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; NAC, neoadjuvant chemotherapy; R, resectable; BR, borderline resectable; UR, unresectable. Baseline characteristics between included vs. excluded patients in this study Values are presented as median (range) or number (%). CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; NAC, neoadjuvant chemotherapy; R, resectable; BR, borderline resectable; UR, unresectable. Regarding patient characteristics for patients included in this study, no significant differences were observed between the soft and stiff groups (Table 2). The pathological response rate of resected specimens after NAC was significantly higher in the soft group than in the stiff group, with median values of 30% (range, 20%–75%) and 20% (range, 10%–65%), respectively (p=0.022) (Table 3). Although the rate of R0 resection did not differ between groups, lymph node metastasis was significantly less frequent in the soft group than in the stiff group (2/12 [16.7%] vs. 23/32 [71.9%]; p=0.003). Post-treatment recurrence was also significantly lower in the soft group (3/12 [25.0%] vs. 19/32 [59.4%]; p=0.042) (Table 3). Overall, 22 of 44 patients (50%) experienced post-treatment recurrence (three in the soft group and 19 in the stiff group). In the soft group, recurrence sites included the liver (n=1), para-aortic lymph node (n=1), and pancreatic transection margin (n=1). In the stiff group, recurrence sites were the liver (n=4), lung (n=2), peritoneum (n=6), and para-aortic lymph node (n=7). Table 2.Characteristics of the patients analyzed in this studyCharacteristicSoft (n=12)Stiff (n=32)p-valueAge (yr)72 (51–81)72.5 (38–82)0.843Sex0.684 Male6 (50.0)20 (62.5) Female6 (50.0)12 (37.5)Tumor location0.379 Head6 (50.0)23 (71.9) Body5 (41.7)7 (21.9) Tail1 (8.3)2 (6.3)Tumor size (mm)21.5 (14–31)19 (11–36)0.133CEA (ng/mL)2.6 (1.2–15.8)2.75 (0.5–12.3)0.742CA19-9 (U/mL)108 (9–1,493)64.5 (2–1,721)0.562Jaundice1 (8.3)9 (28.1)0.322Resectability before NAC1 R12 (100.0)32 (100.0) BR0 (0)0 (0) UR0 (0)0 (0)Stage before NAC (IA/IB/IIA/IIB/III/IV)6/6/0/0/0/019/13/0/0/0/00.828Observation period (day)665.5 (225–1,516)416 (224–1,578)0.108Values are presented as median (range) or number (%).CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; NAC, neoadjuvant chemotherapy; R, resectable; BR, borderline resectable; UR, unresectable. Characteristics of the patients analyzed in this study Values are presented as median (range) or number (%). CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; NAC, neoadjuvant chemotherapy; R, resectable; BR, borderline resectable; UR, unresectable. Table 3.RECIST classification and pathological findings of resected specimens after neoadjuvant chemotherapy and post-treatment recurrenceCharacteristicSoft (n=12)Stiff (n=32)p-valueRECIST classification (CR/PR/SD/PD)0/7/5/00/3/29/00.002Differentiation0.73 Well1 (8.3)1 (3.1) Moderate10 (83.3)29 (90.6) Poor1 (8.3)2 (6.3)Pathological response, median, (range), %30 (20–75)20 (10–65)0.022Evans’ classification (I/IIa/IIb/III/IV)0/8/4/0/00/28/4/0/00.111R0 resection0.345 RX1 (8.3)2 (6.3) R011 (91.7)25 (78.1) R10 (0)5 (15.6) R20 (0)0 (0)Lymph node metastasis2 (16.7)23 (71.9)0.003Interstitial (med/int/sci)0/10/20/29/30.884Infiltration (INFa/INFb/INFc)0/10/20/29/30.884Lymphatic invasion (ly0/ly1/ly2/ly3)7/5/0/011/18/3/00.259Venous invasion (v0/v1/v2/v3)7/5/0/014/14/3/10.594Neural invasion (ne0/ne1/ne2/ne3)3/6/3/07/15/9/10.927Post-treatment recurrence3 (25.0)19 (59.4)0.042Values are presented as number (%) or median (range).RECIST, response evaluation criteria in solid tumours; CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease. RECIST classification and pathological findings of resected specimens after neoadjuvant chemotherapy and post-treatment recurrence Values are presented as number (%) or median (range). RECIST, response evaluation criteria in solid tumours; CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease. Regarding CE-CT findings during PPP, four of the 44 patients were classified as having an isovascular pattern, while 40 patients were classified as having a hypovascular pattern. The pathological response rate was 35% (range, 10%–60%) in the isovascular group and 20% (range, 10%–75%) in the hypovascular group, with no significant difference (p=0.951). Post-treatment recurrence occurred in two of four patients (50%) in the isovascular group and 20 of 40 patients (50.0%) in the hypovascular group, showing no statistically significant difference (p=1). The median RFS was 683 (95% confidence interval [CI], 370–1,201) days in the stiff group and not reached (NR; 95% CI, 567–not available) in the soft group. RFS was significantly longer in patients with soft lesions than in those with stiff lesions (p=0.039, log-rank test) (Fig. 2A). In contrast, RFS did not differ significantly according to CE-CT findings (Fig. 2B). Univariate analysis identified sex (p=0.082), tumor location (p=0.080), EUS-EG stiffness classification (p=0.052), and R0 resection (p=0.083) as potential factors associated with RFS. In the multivariate analysis, EUS-EG stiffness classification was the only factor significantly associated with RFS (hazard ratio, 4.2; p=0.025) (Table 4). (A) The median recurrence-free survival (RFS) of patients in the stiff group and soft group was 683 (95% confidence interval [CI], 370–1,201) days and not reached days (95% CI, 567–not available), respectively. RFS was significantly longer for patients with soft lesion (p=0.039, log-rank test). (B) The median RFS of patients in the isovascular group and hypovascular group at pancreatic parenchymal phase on computed tomography was 779 days (95% CI, 224–not available) and 885 days (95% CI, 423–not available), respectively (p=0.843, log-rank test). Table 4.Univariate and multivariate analyses of factors associated with recurrence-free survivalRecurrence-free survivalUnivariateMultivariateHR (95% CI)p-valueHR (95% CI)p-valueAge (yr) <651.00 ≥651.20 (0.50–2.87)0.689Sex Male1.001.00 Female2.57 (0.86–6.62)0.0822.44 (0.99–6.00)0.052Tumor size before NAC <20 mm1.13 (0.51–2.52)0.766 ≥20 mm1.00Tumor location Head1.001.00 Body-tail2.58 (0.89–7.48)0.081.94 (0.77–4.90)0.16CA19-9 before NAC <37 U/mL1.00 ≥37 U/mL1.50 (0.45–5.03)0.509CT findings in PPP Isovascular1.16 (0.27–5.04)0.843 Hypovascular1.00Stiffness classification Stiff3.38 (0.99–11.51)0.0524.20 (1.20–14.70)0.025 Soft1.001.00R0 resection R01.001.00 RX–R12.31 (0.89–5.96)0.0832.70 (0.89–8.23)0.081Lymph node metastasis Yes1.23 (0.41–3.72)0.714 No1.00HR, hazard ratio; CI, confidence interval; CA19-9, carbohydrate antigen 19-9; NAC, neoadjuvant chemotherapy; CT, computed tomography; PPP, pancreatic parenchymal phase. Univariate and multivariate analyses of factors associated with recurrence-free survival HR, hazard ratio; CI, confidence interval; CA19-9, carbohydrate antigen 19-9; NAC, neoadjuvant chemotherapy; CT, computed tomography; PPP, pancreatic parenchymal phase. We further assessed whether the exclusion of patients who did not complete NAC or AC or those who received non-GS NAC regimens might have affected the association with RFS. Univariate analysis comparing the analyzed patients (n=44) with the excluded patients who did not complete NAC or AC (n=27) showed no significant association between treatment completion and RFS (p=0.105). Similarly, comparison with excluded patients who received non-GS NAC regimens (n=61) revealed that the regimen type (GS vs. non-GS) was not associated with RFS (p=0.923). These findings suggest that the exclusion of these patients was unlikely to have caused substantial selection bias in the present analysis. Representative EUS images of patients with PC with stiff and soft lesions are shown in Fig. 3, respectively. (A) A case of pancreatic cancer (PC) with stiff lesion. The patient was an 82-year-old woman. Endoscopic ultrasonography (EUS) revealed a 12-mm hypoechoic lesion at the pancreatic head. EUS elastography showed blue-dominant patterns, and the lesion was classified as stiff. Response evaluation criteria in solid tumours (RECIST) classification after neoadjuvant chemotherapy (NAC) was stable disease. The lesion was diagnosed as a moderately differentiated PC based on the pathological results of surgery after NAC, with R0 resection and pathological response rate of 10%. The patient had a recurrence 224 days after surgery. (B) A case of PC with soft lesion. The patient was a 78-year-old man. EUS revealed a 23-mm hypoechoic lesion at the pancreatic head. EUS elastography showed green-dominant patterns, and the lesion was classified as soft. RECIST classification after NAC was partial response. The lesion was diagnosed as a moderately differentiated PC based on the pathological results of surgery after NAC, with R0 resection and pathological response rate of 75%. The patient had no recurrence during follow-up of 1,441 days after surgery. DISCUSSION This study demonstrated that EUS-EG may serve as a predictive tool for recurrence in patients with resectable PC undergoing surgery in combination with NAC and AC. Significant differences in pathological response rates and RECIST classifications after NAC, stratified by EUS-EG stiffness classification, indicated that the therapeutic efficacy of NAC varies according to lesion stiffness. Furthermore, significant differences in post-treatment recurrence rates and RFS based on EUS-EG findings suggest that these imaging characteristics may reflect variations in the efficacy of AC in addition to NAC. Based on the results of this study, the contrast enhancement pattern during the PPP on CE-CT was not significantly associated with the pathological response rate or post-treatment recurrence. These findings suggest that in patients with resectable PC, vascularity assessments obtained from CE-CT, a standard diagnostic modality for PC, may be insufficient to predict post-treatment recurrence. Therefore, novel imaging modalities are required to improve recurrence prediction in this patient population. In recent years, the utility of EUS-EG and transabdominal ultrasonography elastography for pancreatic solid masses has been increasingly reported, particularly for differentiating benign from malignant lesions.11-21 This likely reflects the fact that PC often present as stiff lesions owing to abundant fibrosis. In addition to its diagnostic applications, EUS-EG has gained attention as a method for estimating pathological features by assessing lesion stiffness. Our previous pathological research demonstrated that meflin-positive cancer-associated fibroblasts are associated with tissue stiffness in PC.22 Furthermore, we compared PC with non-PC lesions using the EUS-EG stiffness classification and reported that PC lesions were significantly stiffer owing to abundant fibrosis.9 Notably, soft lesions were also observed in the PC group, indicating that the degree of fibrosis was variable, with stiff lesions containing abundant fibrosis and soft lesions containing relatively sparse fibrosis. Although shear-wave measurements and strain elastography with histogram analysis can provide a semiquantitative evaluation of targeted pancreatic areas,16,20 these methods are not always readily available and are procedurally complex, which may affect reproducibility. In contrast, our qualitative approach using EUS-EG color pattern classification is simple and feasible in routine practice, although we acknowledge its inherent subjectivity. To address this, image interpretation was performed by two experienced readers. Our previous study9 demonstrated acceptable interobserver agreement. In addition, the qualitative EUS-EG stiffness classification applied in this study was based on the original strain elastography framework proposed by Iglesias-Garcia et al.,23 in which blue-dominant patterns indicate increased tissue stiffness and were subsequently modified for clinical application in pancreatic solid lesions. However, external validation of this specific categorization remains limited, and further multicenter studies incorporating standardized quantitative elastography parameters are warranted to confirm its generalizability. To the best of our knowledge, this study is the first to investigate the correlation between the EUS-EG stiffness classification and clinical outcomes in patients with resectable PC undergoing surgery in combination with NAC and AC, suggesting that EUS-EG may serve as a predictive tool for post-treatment recurrence in this population. Notably, the pathological response rates after NAC differed significantly according to the EUS-EG findings, which provides pathological evidence that the therapeutic efficacy of chemotherapy varies according to lesion stiffness. Several mechanisms may explain why the effects of NAC and AC differ according to the EUS-EG stiffness classification. The difference in stiffness between PC lesions likely reflects the variations in the degree of intratumoral fibrosis. In the present study, accurately correlating the EUS findings before NAC with the degree of fibrosis in surgical specimens was challenging owing to chemotherapy-induced modifications. However, in our previous study, we demonstrated that stiff lesions on EUS-EG exhibit abundant fibrosis, whereas soft lesions exhibit relatively little fibrosis.9 In tumors with extensive fibrosis, intratumoral blood vessels may be compressed, or the distance between the vessels and cancer cells may be increased by the abundant stroma, potentially reducing the delivery of anticancer drugs. This mechanism likely accounts for the significantly lower pathological response rates observed in stiff lesions compared to that in soft lesions. Furthermore, because most recurrences in this study were distant metastases, these events likely resulted from the proliferation of microscopic cancer cells located outside the surgical resection area that were undetectable on imaging at the time of surgery. Given the biological similarity between primary tumors and recurrent lesions in PC, variations in drug delivery to recurrent lesions may also be inferred from the EUS-EG findings of the primary tumor. Therefore, both NAC and AC likely contribute to suppression of recurrence, and the therapeutic efficacy of AC, in addition to NAC, may vary according to the EUS-EG stiffness classification. In contrast, the pathological examination revealed a significantly higher incidence of lymph node metastasis in the stiff group. Although lymph node metastasis was not identified as a significant predictor of RFS in univariate analysis, this finding suggests that stiff lesions may possess more aggressive biological features. Therefore, the observed association between stiffness and recurrence is likely influenced by both differential chemosensitivity and the underlying tumor biology. Further studies incorporating molecular and histopathological analyses are required to clarify this relationship. In real-world clinical practice, post-treatment recurrence frequently occurs, even in patients with PC who undergo surgery in combination with NAC and AC. The identification of risk factors for recurrence could facilitate the development of more individualized treatment strategies based on patient-specific risk profiles. Although recurrence risk factors have not been clearly established to date, the findings of this study suggest that the EUS-EG stiffness classification may be useful for predicting post-treatment recurrence in patients with resectable PC undergoing surgery with NAC and AC. Specifically, patients with stiff lesions are at a higher risk of postoperative recurrence despite receiving both NAC and AC, highlighting the need for more vigilant post-treatment monitoring. These patients may also benefit from alternative therapeutic strategies, including the addition of radiotherapy, modification of chemotherapy regimens, or adjustment of treatment duration. This study had some limitations. First, this was a single-center retrospective study with a relatively small sample size. Second, owing to the retrospective design of the study, the indications for EUS-EG, NAC, and AC were not uniformly established. Third, the study population was rigorously selected and included only patients who were able to complete the entire treatment sequence of NAC, surgery, and AC, as scheduled. Fourth, this study included only patients who received GS as part of the NAC regimen. However, in Western countries, mFOLFIRINOX has recently become the standard regimen instead of GS. Therefore, the inclusion of GS-only cases limited the applicability of our findings to real-world international clinical practice. Thus, a global consensus on the optimal NAC regimen is needed. In conclusion, this single-center retrospective study suggests that the EUS-EG stiffness classification may serve as a potential predictive tool for recurrence in patients with resectable PC undergoing NAC, surgery, and AC. Patients with stiff lesions on EUS-EG before NAC were more likely to experience recurrence, highlighting the need for closer post-treatment surveillance and consideration of alternative therapeutic strategies. Larger multicenter studies are warranted to validate the clinical utility of EUS-EG as a predictive tool. Footnotes Conflicts of Interest The authors have no potential conflicts of interest. Funding This study was supported by JST (Moonshot R&D) (grant number: JPMJMS2214-11). Author Contributions Conceptualization: YK; Analysis and interpretation of the data: Kano Y, Takada Y, Yamamura T, and Furukawa K; Drafting of the article: Kano Y, Ishikawa T, Yamao K, Mizutani Y, Iida T, and Kawashima H; Critical revision of the article for important intellectual content: all authors; Writing–review & editing: all authors. REFERENCES