Prediction of tumor regression grade and identification of prognostic factors using CT and biological features in patients with pancreatic cancer who underwent surgery after neoadjuvant therapy

Abstract

Background

Radiological response of pancreatic ductal adenocarcinoma (PDAC) to neoadjuvant chemoradiation therapy (CRT) is challenging to assess.

Purpose

To evaluate whether computed tomography (CT) and biological features can predict tumor regression grade (TRG), recurrence-free survival (RFS), and overall survival (OS) of patients who undergo surgery after CRT for PDAC.

Material and Methods

This retrospective study included 125 patients who underwent surgery after CRT for non-metastatic PDAC between January 2013 and March 2021. Two board-certified radiologists independently reviewed initial and post-CRT CT images and assessed the primary tumor extent and regional lymph node metastasis. Another board-certified radiologist quantitatively assessed the primary tumor on pre- and post-CRT diffusion-weighted and positron emission tomography images. Logistic regression and Cox regression analyses were performed to identify predictors of TRG 0/1, RFS, and OS.

Results

In total, 44 (35.2%) patients had a TRG of 0/1. The normalized post-CRT carbohydrate antigen (CA) 19–9 level (<37 IU) (odds ratio [OR] = 3.69; P = 0.024) and adjacent organ invasion on post-CRT CT images (OR = 0.24; P = 0.042) were independent predictors of TRG 0/1. During follow-up (mean = 33.6 months), 68 (54.4%) patients experienced tumor recurrence and 65 (52.0%) died. The normalized post-CRT CA 19-9 level (<37 IU) (hazard ratio [HR] = 0.51; P = 0.028) was a significant predictor of RFS, and size change (%) after CRT (HR = 0.24; P = 0.044) was an independent predictor of OS.

Conclusion

The normalized post-CRT CA 19-9 level and adjacent organ invasion on post-CRT CT images predicted TRG. The normalized post-CRT CA 19-9 level was associated with RFS, whereas size change was an independent predictor of OS.

Keywords

Abdomen/gastrointestinal computed tomography positron emission tomography magnetic resonance imaging pancreas structures primary neoplasms topics

Introduction

Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive malignancy with a poor prognosis. Neoadjuvant therapy (NAT), which includes chemotherapy with or without radiation, has become a standard approach to improve surgical outcomes and survival rates among patients with borderline resectable and locally advanced PDAC. Accurate assessment of the tumor response to NAT is crucial for predicting tumor resectability and patient survival, as well as for planning subsequent treatments (1).

Pathological assessment of tumor regression may be useful for predicting patient outcomes after post-NAT resection to treat PDAC. Currently, multiple systems are available to assess tumor regression, including the College of American Pathologists (CAP), Evans’, and M. D. Anderson Cancer Center systems (2). According to a previous study, the CAP tumor regression grade (TRG) system can predict overall survival (OS) of patients with locally advanced PDAC (3). However, since information about the pathologic TRG is unavailable preoperatively, efforts have been made to predict prognosis or treatment response with imaging features (4 –6).

Computed tomography (CT) is recommended as the primary imaging modality to assess resectability according to the National Comprehensive Cancer Network and European Society for Medical Oncology guidelines (7,8). However, the radiological response of PDAC to NAT is challenging to assess since NAT may induce fibrosis, edema, or inflammation (9). In addition, carbohydrate antigen (CA) 19-9 is a widely used tool for evaluating PDAC since it has been shown to predict tumor resectability and patient survival after NAT (10,11). However, the combined value of CT and CA 19-9 for predicting pathological TRG and survival after NAT has not been fully elucidated.

Therefore, the aim of the present study was to evaluate whether CT imaging, tumor marker levels, and pathology can predict TRG, recurrence-free survival (RFS), and OS in patients who undergo surgery after neoadjuvant chemoradiation therapy (CRT) to treat pancreatic cancer.

Material and Methods

This retrospective study was approved by Seoul National University Hospital institutional review board (IRB No. 2112-037-1279). Due to the retrospective nature of the study, the requirement for written informed consent was waived by Institutional Review Board of Seoul National University Hospital.

Patients

A search of the pathological database at our institution revealed 235 patients who underwent curative-intent surgery for PDAC between January 2013 and March 2021. The inclusion criteria were as follows: (i) patients who underwent surgery after neoadjuvant therapy to treat non-metastatic PDAC; (ii) patients who underwent PET and contrast-enhanced CT after neoadjuvant therapy for preoperative workup; and (iii) patients aged >18 years. According to the inclusion criteria, 146 patients were eligible for enrollment in the study. The exclusion criteria were as follows: (i) patients who underwent contrast-enhanced CT without arterial phase or portal venous phase images (n = 7); (ii) patients with an interval >2 months between contrast-enhanced CT and surgery (n = 4); and (iii) patients with an interval >2 months between PET and surgery (n = 10). Ultimately, 125 patients (66 women [52.8%]; mean age = 62.7 years ± 9.8; 66) were included in the study population (Fig. 1).

Fig. 1.

Patient enrollment process.

NAT regimens included FOLFIRINOX (a combination of 5-fluorouracil, oxaliplatin, irinotecan, and leucovorin) or gemcitabine-based chemotherapy with or without radiotherapy (see supplementary material).

Image acquisition

CT protocol. All patients underwent initial and post-CRT multiphasic CT scans, including precontrast, arterial, and portal venous phases, using various multidetector CT scanners (Supplementary Tables E1 and E4 and supplementary material).

PET protocol. Whole-body PET images were acquired using PET/CT or 3-T simultaneous PET/MRI scanners (Supplementary Tables E2 and E4) 1 h after the intravenous injection of 18F-FDG with a radioactivity of 5.55 MBq/kg (0.15 mCi/kg) of body weight for initial (n = 122) or post-CRT, preoperative (n = 125) workup (see supplementary material).

MRI protocol. Contrast-enhanced MRI was performed using either 1.5-T or 3-T scanners (Supplementary Tables E3 and E4) for initial (n = 97) or post-CRT, preoperative (n = 96) workup (see supplementary material).

Image analysis

Two board-certified radiologists (J. H. K. and S. P. with 25 and 9 years of experience in abdominal imaging, respectively) independently reviewed the initial CT first, followed by the post-CRT CT with intervals >4 weeks to minimize recall bias. The two radiologists were blinded to the clinical, surgical, and pathological data, but they were aware that patients had undergone neoadjuvant therapy and surgery to treat pancreatic cancer. They were asked to determine the maximal tumor size on the axial plane, presence of rim enhancement, extrapancreatic perineural invasion, adjacent organ invasion, tumor vessel (celiac axis, common hepatic artery, superior mesenteric artery, and superior mesenteric/main portal vein) contact (no contact, ≤180°, or >180°), and regional LN metastasis on the initial and post-CRT CT images. Rim enhancement was included for assessment because several previous studies reported its role as a potential imaging biomarker of aggressive tumor biology in PDAC, serving as a predictor of resectability and survival (12,13). Regional LN metastasis was determined based on a short-axis diameter ≥10 mm and morphological features such as round shape, heterogeneous enhancement, or central necrosis (14). The presence of perineural invasion was determined in patients with soft tissue infiltration extending contiguously along well-established pathways (plexus pancreaticus capitalis 1 and 2, anterior, root of mesentery, and splenic plexus) or perivascular soft tissue attenuation >2 mm (15 –17). Overall resectability was categorized on initial and post-CRT CT images, respectively, according to National Comprehensive Cancer Network Guidelines Version 2.2025 (18). Disagreements between the two radiologists were resolved by a consensus meeting.

One board-certified radiologist (J. Y. with 10 years of experience in abdominal imaging) quantitatively assessed the apparent diffusion coefficient (ADC) and maximum standardized uptake value (SUV_max) of the primary tumor on post-CRT diffusion-weighted and PET images, and effort was made to avoid necrotic portions of the tumor. The mean ADC or SUV_max values from two separate sessions were calculated and used for statistical analysis, and an interval of at least 4 weeks was included between each session to reduce recall bias.

Clinical and pathological data collection

Clinical information was acquired from electronic medical records. The serum CA 19-9 level that was measured before CRT was defined as the initial CA 19-9 level. In addition, the CA 19-9 level measured after CRT was considered the post-CRT CA 19-9 level.

The reference standard for tumor resectability was based on the clinical, surgical, and pathological findings. Resectability was classified according to surgical records and pathology reports as follows: R0 = absence of cancer cells within 1 mm of all resection margins; (19) or non-R0 = micro/macroscopic residual tumor. The following information was extracted from pathology reports: data on differentiation; T and N staging; lymphatic, vascular, and perineural invasion; and margin status. To evaluate the pathological TRG system, the hematoxylin and eosin-stained slides of all patients were reviewed by one of two pathologists (K. B. L. and H. K. with 23 and 20 years of clinical experience, respectively). The pathological TRG grading system was implemented according to the College of American Pathologists (20) (see supplementary material).

Follow-up and definitions of recurrence and overall survival

After curative resection, each patient underwent a standardized follow-up protocol, which included contrast-enhanced CT or MRI scans every 6–12 months, along with monitoring of CA19-9 levels. Recurrence was determined based on histopathological analyses or comparisons between previous and follow-up images and tumor marker (CA 19-9) levels.

OS was calculated as the interval between the date of surgery and the date of death or the last follow-up. The survival data of the study population were acquired from the national registry data from the Korean Ministry of Interior and Safety. RFS was measured from the date of surgery to the date of first recurrence or death from any cause. The data cutoff date was 30 September 2023.

Statistical analysis

Univariate and multivariate logistic regression analyses were performed to identify factors that were significantly associated with TRG 0/1. All variables with P values <0.05 in the univariate analyses were included in the multivariate analysis using stepwise selection. Changes (%) in ADC values, CA 19-9 levels, and tumor sizes measured on CT images were defined as the subtraction of the initial value from the post-NAT value divided by the initial value and then multiplied by 100. The Kaplan–Meier method was used to estimate RFS and OS. Prognostic factors for RFS and OS were assessed using univariate and multivariate Cox regression analyses. To assess the inter-observer agreement for each CT imaging feature, the intraclass correlation coefficient (ICC) was calculated. An ICC ˂0.5, 0.5–0.75, 0.75–0.9, and ˃0.90 indicated poor, moderate, good, and excellent agreement, respectively (21). All the statistical analyses were performed using SPSS Statistics for Windows version 27.0 (IBM Corp., Armonk, NY, USA) and MedCalc Statistical Software version 18.9.1 (MedCalc Software Ltd., Ostend, Belgium). P values <0.05 indicated statistical significance.

Results

Patient characteristics

Of the 125 patients, baseline CT resectability categories were resectable in 48 (38.4%) patients, borderline resectable in 53 (42.4%), and locally advanced in 24 (19.2%) (Table 1). The R0 resection rate was 68.0% (85/125). Radiologists’ sensitivity and specificity in determining R0 resectability on post-NAT CT was 58.8% (50/85) and 40.0% (16/40), respectively, when only resectable category on CT was treated as resectable disease (Supplementary Table E5). The sensitivity and specificity were 84.7% (72/85) and 15.0% (6/40), respectively, when both resectable and borderline resectable categories on CT were regarded as resectable disease. The mean SUV_max was 8.3 ± 4.8 (range = 1.0–20.0) on initial PET imaging and 3.1 ± 2.4 (range = 1.0–16.4) on post-NAT imaging. The mean ADC value was 1268.0 ± 214.6 mm²/s × 10⁻³ (range = 750.0–1833.0 mm²/s × 10⁻³) on initial MRI and 1535.6 ± 271.5 (range = 1030.0–2350.0 mm²/s × 10⁻³) on post-NAT imaging. Adjacent organ invasion was present on post-NAT CT imaging in 21/125 (16.8%) patients; the organs exhibiting invasion included the duodenum (n = 14), stomach (n = 3), left adrenal gland (n = 2), colon (n = 1), and both stomach and duodenum (n = 1).

Table 1.

Patient characteristics.

Characteristics	Values
Age (years)	62.7 ± 9.8 (36–83)
Sex
Male	59 (47.2)
Female	66 (52.8)
Tumor size (cm)
Initial	2.8 ± 0.7 (1.6–4.9)
Post-NAT	1.8 ± 0.6 (0.8–3.8)
Initial CT resectability category
Resectable	48 (38.4)
Borderline resectable	53 (42.4)
Locally advanced	24 (19.2)
Post-NAT CT resectability category
Resectable	74 (59.2)
Borderline resectable	32 (25.6)
Locally advanced	19 (15.2)
CA 19-9 (U/mL)
Initial	1412.7 ± 2841.7 (1.0–12000.0)
Post-neoadjuvant	76.2 ± 153.8 (1.0–987.0)
NAT
Chemotherapy only	61 (48.8)
Chemoradiation therapy	64 (51.2)
Types of surgery
Whipple's operation or PPPD	81 (64.8)
Distal pancreatectomy	35 (28.0)
Total pancreatectomy	9 (7.2)
Resection margin
Negative	85 (68.0)
Involved by tumor	40 (32.0)
Differentiation
N/A	7 (5.6)
Well differentiated	11 (8.8)
Moderately differentiated	100 (80.0)
Poorly differentiated	7 (5.6)
Pathologic T staging
T0	7 (5.6)
T1	44 (35.2)
T2	41 (32.8)
T3	28 (22.4)
T4	5 (4.0)
Pathologic N staging
N0	83 (66.4)
N1	38 (30.4)
N2	4 (3.2)
Tumor regression grade
0	7 (5.6)
1	37 (29.6)
2	57 (45.6)
3	24 (19.2)
Lymphatic invasion
Presence	24 (19.2)
Absence	101 (80.8)
Microscopic vascular invasion
Presence	30 (24.0)
Absence	95 (76.0)
Perineural invasion
Presence	84 (67.2)
Absence	41 (32.8)

Values are given as n (%) or mean ± SD (range).

CA 19-9, carbohydrate antigen 19-9; N/A, not applicable; NAT, neoadjuvant therapy; PPPD, pylorus-preserving pancreaticoduodenectomy; SD, standard deviation.

Predictors for TRG 0 or 1 and recurrence-free survival

Among the 125 patients, 7 (5.6%) had TRG 0 and 37 (29.6%) had TRG 1 on pathological analysis. For the prediction of TRG 0 or 1, univariate analyses revealed that post-CRT SUV_max (odds ratio [OR] = 0.78, 95% confidence interval [CI] = 0.64–0.96; P = 0.019), normalized post-CRT CA 19-9 level (< 37 IU) (OR = 3.63, 95% CI = 1.45–9.13; P = 0.006), CA 19-9 change (%) after CRT (OR = 2.30, 95% CI = 1.07–4.94; P = 0.032), tumor size (OR = 0.90, 95% CI = 0.84–0.97; P = 0.005), adjacent organ invasion on post-CRT CT images (OR = 0.26, 95% CI = 0.07–0.92; P = 0.038), and size change (%) after CRT (OR = 3.03, 95% CI = 1.41–6.50; P = 0.004) were significantly associated with TRG 0 or 1 (Table 2). Multivariate analyses revealed that a normalized post-CRT CA 19-9 level (<37 IU) (OR = 3.69, 95% CI = 1.25–10.90; P = 0.024) and adjacent organ invasion on post-CRT CT images (OR = 0.24, 95% CI = 0.06–0.95; P = 0.042) were independent predictors of TRG 0 or 1 (Fig. 2).

Fig. 2.

A 72-year-old female patient. (a) The axial image demonstrates a 3-cm pancreatic cancer in the uncinate process, encasing both superior mesenteric artery (arrow) and vein (arrowhead). There was no evidence of adjacent organ invasion by the tumor. (b) Initial PET image shows increased FDG uptake of the primary tumor. (c) The axial CT scan performed after chemoradiation shows a 2-cm pancreatic cancer without adjacent organ invasion. (d) Post-chemoradiation PET image demonstrates no significantly increased FDG uptake of the tumor. The initial carbohydrate antigen 19-9 level was 4479 U/mL, which reduced to 3 U/mL after neoadjuvant chemoradiation. The patient underwent Whipple's operation, and the pathologic analysis revealed pathologic no residual tumor with the College of American Pathologist grade 0. CT, computed tomography; FDG, fluorodeoxyglucose; PET, positron emission tomography.

Table 2.

Summary of important findings for prediction of TRG 0 or 1 in pancreatic cancer.

	TRG 0/1 (n = 44)	TRG 2/3 (n = 81)	Univariate analysis		Multivariate analysis
	TRG 0/1 (n = 44)	TRG 2/3 (n = 81)	OR (95% CI)	P value	OR (95% CI)	P value
Age (years)	63.5 ± 9.8	62.3 ± 9.8	1.01 (0.97–1.05)	0.532
Sex (M:F)	19:25	40:41	1.28 (0.61–2.69)	0.508
Initial SUV	8.7 ± 6.0	8.3 ± 4.7	1.02 (0.95–1.09)	0.666
Post-CRT SUV	2.3 ± 1.9	3.4 ± 2.6	0.78 (0.64–0.96)	0 . 019	0.98 (0.78–1.25)	0.875
SUV decrease >30% or metabolic CR after CRT	39/44 (88.6)	60/81 (64.8)	2.73 (0.95–7.84)	0.062
Initial ADC (mm²/s ×10⁻³)	1268.2 ± 231.0	1256.6 ± 248.7	1.00 (1.00–1.00)	0.996
Post-CRT ADC (mm²/s ×10⁻³)	1588.2 ± 309.0	1505.5 ± 245.1	1.00 (1.00–1.00)	0.153
ADC change (%) after CRT	25.3 ± 29.6	20.5 ± 25.6	1.95 (0.34–11.23)	0.455
Initial CA 19-9 level (<37 IU)	17/44 (38.6)	25/81 (30.9)	1.46 (0.68–3.17)	0.333
Post-CRT CA 19-9 level (<37 IU)	37/44 (84.1)	48/81 (59.3)	3.63 (1.45–9.13)	0.006	3.69 (1.25–10.90)	0.024
CA 19-9 change (%) after CRT	–181.1 ± 20.7	68.8 ± 40.4	2.30 (1.07–4.94)	0.032	1.97 (0.80–4.83)	0.139
CA 19-9 responder*	19/44 (43.2)	25/81 (30.9)	1.70 (0.80–3.64)	0.170
Post-CRT CT imaging features
Tumor size (cm)	1.6 ± 0.6	2.0 ± 0.6	0.90 (0.84–0.97)	0.005	0.95 (0.87–1.03)	0.219
Rim enhancement	0	3/81 (3.7)	N/A	0.998
Extrapancreatic perineural invasion	29/44 (65.9)	49/81 (60.5)	1.26 (0.59–2.72)	0.551
Adjacent organ invasion	3/44 (6.8)	18/81 (22.2)	0.26 (0.07–0.92)	0.038	0.24 (0.06–0.95)	0.042
CA encasement	4/44 (9.1)	3/81 (3.7)	1.00 (0.45–2.22)	0.995
CHA encasement	6/44 (13.6)	7/81 (8.6)	1.32 (0.74–2.34)	0.345
SMA encasement	3/44 (6.8)	1/81 (1.2)	1.60 (0.80–3.20)	0.186
SMV/MPV encasement	3/44 (6.8)	11/81 (64.8)	0.98 (0.54–1.78)	0.952
Regional LN metastasis	3/44 (6.8)	5/81 (6.2)	1.11 (0.25–4.89)	0.888
Initial CT imaging features
Tumor size (cm)	2.9 ± 0.7	2.8 ± 0.7	0.98 (0.92–1.03)	0.363
Rim enhancement	5/44 (11.4)	17/81 (21.0)	0.48 (0.16–1.41)	0.184
Extrapancreatic perineural invasion	34/44 (77.3)	53/81 (65.4)	2.19 (0.91–1.28)	0.082
Adjacent organ invasion	8/44 (18.2)	19/81 (23.5)	0.73 (0.29–1.82)	0.495
CA encasement	2/44 (4.5)	4/81 (4.9)	0.44 (0.05–3.65)	0.443
CHA encasement	5/44 (11.4)	6/81 (7.4)	1.00 (1.00–1.00)	0.998
SMA encasement	8/44 (18.2)	4/81 (4.9)	0.60 (0.19–1.96)	0.402
SMV/MPV encasement	35/44 (79.5)	59/81 (72.8)	1.09 (0.89–1.34)	0.398
Regional LN metastasis	6/44 (13.6)	16/81 (19.8)	0.99 (0.81–1.20)	0.891
Size change (%) after CRT	–40.6 ± 19.1	–31.1 ± 17.8	3.03 (1.41–6.50)	0 . 004	0.30 (0.01–8.37)	0.875

Values are given as n (%) or mean ± SD unless otherwise indicated.

*Defined as patients with serum CA 19-9 > 37 IU in the initial state and normalized after CRT.

ADC, apparent diffusion coefficient; CA, celiac axis; CA 19-9, carbohydrate antigen 19-9; CHA, common hepatic artery; CI, confidence interval; CR, complete response; CRT, chemoradiation therapy; IU, international unit; LN, lymph node; OR, odds ratio; SD, standard deviation; SMA, superior mesenteric artery; SMV/MPV, superior mesenteric/main portal vein; SUV, standardized uptake value; TRG, tumor regression grade.

During follow-up (mean duration = 33.6 ± 18.5 months; range = 4–83 months), 68/125 (54.4%) patients experienced tumor recurrence after surgery, among whom local recurrence was most common (n = 27, 21.6%) (see supplementary material). The estimated 1-, 3-, and 5-year RFS rates of 44 patients with TRG 0 or 1 were 77.1%, 64.7%, and 61.8%, respectively; these RFS rates were significantly higher than those of 81 patients with TRG 2 or 3, who had RFS rates of 67.1%, 38.1%, and 22.8%, respectively (P = 0.011) (Fig. 3). For the prediction of RFS, univariate analyses revealed that post-CRT SUV_max (OR = 1.10, 95% CI = 1.01–1.21; P = 0.048), initial CA 19-9 level (OR = 0.57, 95% CI = 0.33–1.00; P = 0.049) and normalized post-CRT CA 19-9 level (<37 IU) (OR = 0.42, 95% CI = 0.26–0.68; P <0.001) were significantly associated with RFS (Supplementary Table E6). According to multivariate analyses, normalized post-CRT CA 19-9 level (<37 IU) (hazard ratio [HR] = 0.51, 95% CI = 0.28–0.93; P = 0.028) was a significant predictor of RFS (Supplementary Table E6).

Fig. 3.

Kaplan–Meier estimation of recurrence-free survival in patients stratified according to tumor regression grade. The recurrence-free survival of 44 patients with TRG 0 or 1 were significantly higher than that of 81 patients with TRG 2 or 3 (P = 0.011). TRG, tumor regression grade.

Predictors of overall survival

In total, 65 (52.0%) patients died during the follow-up period. Adjacent organ invasion on both pre- and post-CRT CT images and size change (%) after CRT were significantly associated with OS according to univariate analysis (Table 3). Multivariate analysis revealed that size change (%) after CRT (HR = 0.24, 95% CI = 0.06–0.96; P = 0.044) was an independent predictor of OS (Fig. 4). Regarding inter-observer agreement (Table 4), tumor-vessel contact of the SMA and MPV/SMV on both initial and post-CRT CT images and size on post-CRT CT images showed good inter-observer agreement (ICC range = 0.774–0.856). Size measurement, rim enhancement, adjacent organ invasion, MPV/SMV reconstructability on initial CT images and tumor-vessel contact of the CA and CHA on both initial and post-CRT images showed moderate inter-observer agreement (ICC range = 0.540–0.721). Inter-observer agreement for extrapancreatic perineural invasion was poor for both initial and post-CRT images (ICC = 0.444 and 0.495, respectively).

Fig. 4.

A 62-year-old female patient. (a, b) The initial axial CT scan shows a 2 cm ill-defined pancreatic body cancer (arrows) with abutment of the celiac axis (arrowhead). Tumor resectability was assessed as borderline resectable status and the patient underwent neoadjuvant chemotherapy. (c) The axial CT scan performed after neoadjuvant chemotherapy demonstrates the size growth of the tumor (arrows) to 2.5 cm. The carbohydrate antigen 19-9 level was 18 mL at the initial state and 7 mL after neoadjuvant therapy. The patient underwent distal pancreatectomy and the pathologic analysis revealed moderately differentiated pancreatic ductal adenocarcinoma with pathologic staging ypT4N0. (d, e) The axial CT scan performed 3 months after the operation shows hepatic (arrow in panel d) and abdominal wall metastases (arrowhead in panel e). The patient died 6 months after the operation. CT, computed tomography.

Table 3.

Summary of important findings for prediction of overall survival in pancreatic cancer.

	Univariate analysis		Multivariate analysis
	HR (95% CI)	P value	HR (95% CI)	P value
Age	1.02 (1.00–1.05)	0.099
Sex	0.83 (0.51–1.35)	0.443
Initial SUV	1.02 (0.97–1.06)	0.479
Post-CRT SUV	1.05 (0.95–1.16)	0.308
SUV decrease > 30% or metabolic CR after CRT	0.98 (0.54–1.77)	0.939
Initial ADC	1.00 (1.00–1.00)	0.196
Post-CRT ADC	1.00 (1.00–1.00)	0.889
ADC change (%) after CRT	0.92 (0.29–2.95)	0.889
Initial CA 19-9 level < 37 IU	0.93 (0.55–1.58)	0.782
Normalized post-CRT CA 19-9 level < 37 IU	1.00 (1.00–1.00)	0.493
CA 19-9 change (%) after CRT	0.99 (0.98–1.00)	0.218
Post-CRT CT imaging features
Tumor size (cm)	1.03 (0.99–1.07)	0.164	1.48 (0.66–3.33)	0.340
Rim enhancement	0.73 (0.10–5.25)	0.753
Extrapancreatic perineural invasion	1.06 (0.64–1.76)	0.815
Adjacent organ invasion	1.98 (1.09–3.61)	0 . 025
CA encasement	0.45 (0.11–1.82)	0.260
CHA encasement	1.28 (0.58–2.81)	0.537
SMA encasement	1.54 (0.48–4.91)	0.469
SMV/MPV encasement	1.56 (0.78–3.17)	0.220
Regional LN metastasis	1.67 (0.72–3.97)	0.235
Initial CT imaging features
Tumor size (cm)	0.98 (0.95–1.02)	0.351	1.32 (0.63–2.76)	0.457
Rim enhancement	1.43 (0.77–2.62)	0.255
Extrapancreatic perineural invasion	1.13 (0.65–1.95)	0.660
Adjacent organ invasion	1.76 (1.02–3.03)	0.043
CA encasement	1.24 (0.75–2.06)	0.428
CHA encasement	0.92 (0.58–1.45)	0.707
SMA encasement	0.96 (0.63–1.46)	0.838
SMV/MPV encasement	1.56 (0.77–3.17)	0.244
Regional LN metastasis	1.67 (0.72–3.87)	0.235
Size change (%) after CRT	0.20 (0.05–0.80)	0.022	0.24 (0.06–0.96)	0.044

ADC, apparent diffusion coefficient; CA, celiac axis; CA 19-9, carbohydrate antigen 19-9; CHA, common hepatic artery; CI, confidence interval; CR, complete response; CRT, chemoradiation therapy; HR, hazard ratio; LN, lymph node; SMA, superior mesenteric artery; SMV/MPV, superior mesenteric/main portal vein.

Table 4.

Summary of interobserver agreement for imaging features between two reviewers.

Initial CT	ICC (95% CI)	Post-CRT CT	ICC (95% CI)
Size	0.606 (0.448–0.719)	Size	0.851 (0.650–0.810)
Rim enhancement	0.604 (0.436–0.722)	Rim enhancement	0.428 (0.186–0.598)
Extrapancreatic perineural invasion	0.444 (0.208–0.610)	Extrapancreatic perineural invasion	0.495 (0.282–0.645)
Adjacent organ invasion	0.559 (0.372–0.690)	Adjacent organ invasion	0.431 (0.278–0.563)
Tumor-vessel contact		Tumor-vessel contact
CA	0.724 (0.607–0.806)	CA	0.540 (0.404–0.653)
CHA	0.745 (0.637–0.820)	CHA	0.682 (0.548–0.777)
SMA	0.837 (0.768–0.886)	SMA	0.840 (0.773–0.888)
MPV/SMV	0.856 (0.795–0.899)	MPV/SMV	0.774(0.679–0.841)
MPV/SMV reconstructability	0.693 (0.562–0.784)	MPV/SMV reconstructability	1.000
Regional LN metastasis	0.383 (0.121–0.566)	Regional LN metastasis	0.604 (0.437–0.722)

CA, celiac axis; CHA, common hepatic artery; CI, confidence interval; CRT, chemoradiation therapy; ICC, intraclass correlation coefficient; LN, lymph node;MPV/SMV, main portal/superior mesenteric vein; SMA, superior mesenteric artery.

Discussion

In this study, 44/125 (35.2%) patients had TRG of 0 or 1. The normalized post-CRT CA 19-9 level and adjacent organ invasion on post-CRT CT images were significant predictors of TRG 0 or 1. The estimated 1-, 3-, and 5-year RFS rates of 44 patients with TRG 0 or 1 were 77.1%, 64.7%, and 61.8%, respectively; these RFS rates were significantly higher than those of 81 patients with TRG 2 or 3, who had RFS rates of 67.1%, 38.1%, and 22.8%, respectively. In addition, normalized post-CRT CA 19-9 level was associated with RFS, whereas change in size on CT imaging after NAT was an independent predictor of OS.

In our study, adjacent organ invasion was present on post-CRT CT images in 16.8% of patients, and among these patients, the duodenum was most commonly involved. Chang et al. (22) reported that patients with extrapancreatic perineural and/or duodenal invasion on preoperative CT scans had reduced survival after upfront pancreaticoduodenectomy to treat pancreatic cancer. In addition, a more recent study (23) provided new findings on the correlation between radiological duodenal invasion and early recurrence within 12 months after upfront pancreaticoduodenectomy and OS. In the aforementioned study (23), radiological duodenal invasion was associated with microscopic lymphovascular invasion, suggesting biologically aggressive tumor characteristics. Similarly, the association between radiological organ invasion and TRG in our study might be explained by tumor aggressiveness, although patient characteristics are different since our study included patients who underwent neoadjuvant therapy and previous studies included patients who received upfront surgery. With recent advances in NAT for patients with pancreatic cancer, efforts have been made to assess the degree of tumor regression after NAT and investigate the prognostic value of TRG. In our study, which used the CAP grading system, patients with TRG 0/1 had better RFS than those with TRG 2/3. However, the prognostic performance of the CAP grading system remains controversial. In a study by Ahn et al. (3), the CAP grading system was a significant predictor of OS. Similarly, in a retrospective study by Truty et al. (24), the CAP grading system was the most predictive factor for both RFS and OS. In contrast, another study (2) revealed significant overlaps in both OS and RFS according to the CAP grading system. These conflicting results might be due to the different sample sizes in each study, as well as the different stratifications of high- and low-grade regression. In addition to controversial results in terms of prognostic value, pathological TRG has limitations of inaccessibility before surgery and low inter-observer agreement (25). Thus, more robust, preoperative predictors of outcome are clinically in demand.

The role of CA 19-9 levels has been vigorously investigated in pancreatic cancer patients after NAT, and a previous study showed that a CA 19-9 responder group had better OS than a non-responder group (26). In a multicenter study (26), changes in CA19–9 levels during NAT indicated the optimal timing for subsequent surgery and had the ability to predict postoperative prognosis. In a meta-analysis (27), a post-NAT decrease in CA19-9 levels of more than 50% or normalization of CA19-9 levels was related to a more promising OS, suggesting that optimal CA19-9 response may be a suitable prognostic indicator for treatment decisions. However, there are limitations in the use of CA 19-9 levels in routine clinical practice due to the variability and inconsistency in the cutoff values for CA 19-9 response and the number of chemotherapy cycles among previous studies (28,29).

In patients with pancreatic cancer, radiological evaluation of tumor regression after NAT is difficult since the radiological response does not accurately reflect pathological tumor regression (9). Necrosis, inflammation, or fibrosis of the tumor, which is induced by NAT, interferes with the radiological evaluation of tumor regression. Therefore, CT has been reported to play a limited role in evaluating response and predicting prognosis. For example, a previous study (30) revealed that the change in size after NAT according to the RECIST criteria was not associated with OS in patients who underwent surgery for borderline resectable pancreatic cancer. However, according to a retrospective study (5), combining the CA 19-9 response and CT at the 8-week follow-up allowed more accurate stratification of survival among patients with non-metastatic pancreatic cancer. Furthermore, Yang et al. (31) showed that any degree of radiological improvement in tumor size was an independent predictor of OS for PDAC patients undergoing post-NAT curative-intent surgery, along with normalization of CA 19-9 levels, suggesting that it is a better criterion for predicting treatment response than the RECIST criteria. Similarly, our study results revealed that size change on post-NAT CT was significantly related to OS; thus, we cautiously suggest a predictive role of post-NAT CT for prognosis.

The present study has some limitations. Owing to its retrospective design, potential selection bias was unavoidable. Specifically, the study population consisted of patients who underwent curative-intent surgery after NAT, which unintendedly but inevitably excluded patients who exhibited poor responses to NAT.

In conclusion, a normalized post-CRT CA 19-9 level and adjacent organ invasion on post-CRT CT images were significant predictors of TRG. The normalized post-CRT CA 19-9 level was associated with RFS, whereas the change in size was an independent predictor of OS.

Supplemental Material

sj-docx-1-acr-10.1177_02841851261424497 - Supplemental material for Prediction of tumor regression grade and identification of prognostic factors using CT and biological features in patients with pancreatic cancer who underwent surgery after neoadjuvant therapy

Supplemental material, sj-docx-1-acr-10.1177_02841851261424497 for Prediction of tumor regression grade and identification of prognostic factors using CT and biological features in patients with pancreatic cancer who underwent surgery after neoadjuvant therapy by Jeongin Yoo, Sae-Jin Park, Haeryoung Kim, Kyung Bun Lee and Jung Hoon Kim in Acta Radiologica

Supplemental Material

sj-docx-2-acr-10.1177_02841851261424497 - Supplemental material for Prediction of tumor regression grade and identification of prognostic factors using CT and biological features in patients with pancreatic cancer who underwent surgery after neoadjuvant therapy

Supplemental material, sj-docx-2-acr-10.1177_02841851261424497 for Prediction of tumor regression grade and identification of prognostic factors using CT and biological features in patients with pancreatic cancer who underwent surgery after neoadjuvant therapy by Jeongin Yoo, Sae-Jin Park, Haeryoung Kim, Kyung Bun Lee and Jung Hoon Kim in Acta Radiologica

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by Seoul National University Hospital research fund (grant no. 04-2023-2280).

Supplementary material

Supplemental material for this article is available online.

ORCID iD

Jung Hoon Kim

References

Zins

Matos

Cassinotto

. Pancreatic adenocarcinoma staging in the era of preoperative chemotherapy and radiation therapy. Radiology 2018;287:374–390.

Matsuda

Ohkubo

Nakano-Narusawa

, et al. Objective assessment of tumor regression in post-neoadjuvant therapy resections for pancreatic ductal adenocarcinoma: comparison of multiple tumor regression grading systems. Sci Rep 2020;10:18278.

Ahn

Lee

J-c

Kim

, et al. Four-tier pathologic tumor regression grading system predicts the clinical outcome in patients who undergo surgical resection for locally advanced pancreatic cancer after neoadjuvant chemotherapy. Gut Liver 2022;16:129.

Cassinotto

Mouries

Lafourcade

J-P

, et al. Locally advanced pancreatic adenocarcinoma: reassessment of response with CT after neoadjuvant chemotherapy and radiation therapy. Radiology 2014;273:108–116.

Kim

S-s

Lee

, et al. Retrospective evaluation of treatment response in patients with nonmetastatic pancreatic cancer using CT and CA 19-9. Radiology 2022;303:548–556.

Yoo

Lee

Joo

, et al. Post-neoadjuvant treatment pancreatic cancer resectability and outcome prediction using CT, 18F-FDG PET/MRI and CA 19–9. Cancer Imaging 2023;23:49.

Kulkarni

Soloff

Tolat

, et al. White paper on pancreatic ductal adenocarcinoma from society of abdominal radiology's disease-focused panel for pancreatic ductal adenocarcinoma: part I, AJCC staging system, NCCN guidelines, and borderline resectable disease. Abdom Radiol (NY) 2020;45:716–728.

Conroy

Pfeiffer

Vilgrain

, et al. Pancreatic cancer: ESMO clinical practice guideline for diagnosis, treatment and follow-up. Ann Oncol 2023;34:987–1002.

Rhee

Park

M-S

. The role of imaging in current treatment strategies for pancreatic adenocarcinoma. Korean J Radiol 2021;22:23.

10.

Tsai

George

Wittmann

, et al. Importance of normalization of CA19-9 levels following neoadjuvant therapy in patients with localized pancreatic cancer. Ann Surg 2020;271:740–747.

11.

Heger

Sun

Hinz

, et al. Induction chemotherapy in pancreatic cancer: cA 19-9 may predict resectability and survival. HPB (Oxford) 2020;22:224–232.

12.

Lee

Kim

Park

, et al. Pancreatic ductal adenocarcinoma: rim enhancement at MR imaging predicts prognosis after curative resection. Radiology 2018;288:456–466.

13.

Renzulli

Cucchetti

Zucchini

, et al. Rim enhancement on imaging of pancreatic ductal adenocarcinoma: systematic review and meta-analysis of biological and prognostic values. Radiol Med 2025. DOI: 10.1007/s11547-025-02131-7.

14.

Joo

Lee

, et al. Preoperative MDCT assessment of resectability in borderline resectable pancreatic cancer: effect of neoadjuvant chemoradiation therapy. AJR Am J Roentgenol 2018;210:1059–1065.

15.

Patel

Olcott

Jeffrey

. Extrapancreatic perineural invasion in pancreatic adenocarcinoma. Abdominal Radiology 2018;43:323–331.

16.

Makino

Kitagawa

Ohta

, et al. Nerve plexus invasion in pancreatic cancer: spread patterns on histopathologic and embryological analyses. Pancreas 2008;37:358–365.

17.

Mochizuki

Gabata

Kozaka

, et al. MDCT Findings of extrapancreatic nerve plexus invasion by pancreas head carcinoma: correlation with en bloc pathological specimens and diagnostic accuracy. Eur Radiol 2010;20:1757–1767.

18.

NCCN Guidelines Version 2.2025 Pancreatic adenocarcinoma. Available at: nccn.org/professionals/physician_gls/pdf/pancreatic/pdf (2025, accessed December 3th 2025).

19.

Campbell

Smith

Whelan

, et al. Classification of R1 resections for pancreatic cancer: the prognostic relevance of tumour involvement within 1 mm of a resection margin. Histopathology 2009;55:277–283.

20.

Lawrence

Burgart

WVC

Jain

. Protocol for the Examination of Specimens from Patients with Tumors of the Endocrine Pancreas Version 4.1.0.0. June 2021 ed.. Northfield, IL: College of American Pathologists, 2021.

21.

Koo

. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med 2016;15:155–163.

22.

Chang

Jeffrey

Patel

, et al. Preoperative multidetector CT diagnosis of extrapancreatic perineural or duodenal invasion is associated with reduced postoperative survival after pancreaticoduodenectomy for pancreatic adenocarcinoma: preliminary experience and implications for patient care. Radiology 2016;281:816–825.

23.

Miyazawa

Hirono

Kawai

, et al. Radiographic duodenal invasion is associated with poor prognosis and early recurrence in patients with pancreatic ductal adenocarcinoma. Eur J Surg Oncol 2023;49:106960.

24.

Truty

Kendrick

Nagorney

, et al. Factors predicting response, perioperative outcomes, and survival following total neoadjuvant therapy for borderline/locally advanced pancreatic cancer. Ann Surg 2021;273:341–349.

25.

Kalimuthu

Serra

Dhani

, et al. Regression grading in neoadjuvant treated pancreatic cancer: an interobserver study. J Clin Pathol 2017;70:237–243.

26.

Aoki

Motoi

Murakami

, et al. Decreased serum carbohydrate antigen 19–9 levels after neoadjuvant therapy predict a better prognosis for patients with pancreatic adenocarcinoma: a multicenter case-control study of 240 patients. BMC Cancer 2019; 19: 252.

27.

Sadula

Ren

, et al. The prognostic value of CA19-9 response after neoadjuvant therapy in patients with pancreatic cancer: a systematic review and pooled analysis. Cancer Chemother Pharmacol 2020;86:731–740.

28.

van Veldhuisen

Vogel

Klompmaker

, et al. Added value of CA19-9 response in predicting resectability of locally advanced pancreatic cancer following induction chemotherapy. HBP (Oxford) 2018;20:605–611.

29.

Al Abbas

Zenati

Reiser

, et al. Serum CA19-9 response to neoadjuvant therapy predicts tumor size reduction and survival in pancreatic adenocarcinoma. Ann Surg Oncol 2020;27:2007–2014.

30.

Katz

Fleming

Bhosale

, et al. Response of borderline resectable pancreatic cancer to neoadjuvant therapy is not reflected by radiographic indicators. Cancer 2012;118:5749–5756.

31.

Yang

Park

M-S

Han

, et al. Risk stratification of pancreatic ductal adenocarcinoma patients undergoing curative-intent surgery after neoadjuvant therapy. Cancer Res Treat 2023;56:247–258.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB

0.02 MB