Introduction
Approximately 30% of patients with locally advanced pancreatic cancer (LAPC) die of local progression [
1], and only 11% of these patients could survive more than 3 years [
2]. Therefore, improving the rate of local control and preventing local progression are crucial issues to be addressed [
1,
3].
Intraoperative radiotherapy (IORT), a precise therapeutic approach that delivers high doses of radiation directly into the tumor while protecting adjacent organs, has demonstrated promising outcomes with improved local disease control rate and potential prognosis in LAPC [
4]. IORT is recommended as experts’ consensus to perform on LAPC patients to relieve symptoms and obtain extra benefits [
4,
5]. Around 50–80% of pancreatic ductal adenocarcinoma (PDAC) patients were assessed as stable disease (SD) based on response evaluation criteria in solid tumor version 1.1 (RECIST v.1.1) after treatment [
6]. Despite no significant change in tumor diameter in these patients, some of them could achieve major pathologic response and better prognosis [
7]. Some patients assessed as SD responded well and had longer progression-free survival (PFS), whereas others progressed rapidly [
8]. Consequently, it is crucial to determine an efficient approach to further stratify the risk of progression in individuals assessed as SD initially after IORT. Many efforts have been made to solve this problem, including clinicopathological features, molecular biomarkers, imaging features, and radiomics [
9‐
11]. Unfortunately, the potential of these methods is far from sufficiently investigated.
Multi-detector computed tomography is the recommended imaging technique in the evaluation of PDAC, which consists of a dual-phase contrast-enhanced protocol dedicated to the pancreas routinely [
12‐
14]. The extracellular volume (ECV) fraction, representing the sum of the extravascular extracellular space and the intravascular space, could be estimated by means of contrast-enhanced CT (CECT) [
15]. Recently, it has been revealed that ECV is associated with fibrosis and deposition of an extracellular stromal matrix of the pancreas [
15,
16]. Desmoplastic stroma exhibits an essential role in tumor oncogenesis, proliferation, progression, metastasis, and chemoresistance [
16,
17]. ECV fraction could be used to evaluate pancreatic fibrosis and predict tumor aggressiveness, treatment response, and prognosis in PDAC [
15,
18]. Based on the above, we hypothesize that ECV might become a potential imaging biomarker for stratifying the risk of progression in SD patients. However, until recently, ECV was estimated mainly via the delayed or equilibrium phase of CECT, which prolonged examination time and limited its clinical practicality. Dual-energy CT was also employed in other investigations, but this requires specialized equipment and scanning protocols. Nevertheless, it has not been determined whether the ECV, which is derived from the portal-venous phase (PVP) of conventional CT, could be a more acceptable non-invasive imaging biomarker for prognostic prediction in PDAC. Many studies have shown that clinical factors and CT imaging features, for example, CA19-9, necrosis, and peripancreatic tumor infiltration, were valuable in predicting the prognosis of PDAC [
9,
12].
A risk-scoring system could serve as a simple and efficient way to evaluate clinicopathological features or prognosis in PDAC [
19]. Therefore, the purpose of this study was to investigate the potential role of ECV derived from PVP of conventional CT in predicting progression and construct a progression risk-scoring system based on ECV and clinical-radiological features in LAPC patients assessed as SD initially after IORT.
Discussion
In the present study, the ECV fraction derived from PVP was proven to be an independent risk factor for progression in LAPC patients assessed with SD initially after IORT. Moreover, a risk-scoring system integrating ECV, CT radiological features, and CA19-9 response as a novel biomarker was constructed, and utilized to predict PFS in LAPC patients assessed with SD, which could further stratify the risk of progression in these patients with satisfactory prognostic predictive performance. Our scoring system could serve as a complement to the RECIST v.1.1 criteria to identify LAPC patients with SD who would develop progress at risk after IORT, allowing clinicians to adopt appropriate treatment strategies and improve the prognosis.
In this study, 103 patients included 64 males and 39 females, with a mean age of 58.52 ± 10.09 years, which is similar to the previous study [
25]. Demographic breakdown in our study aligns with known characteristics of LAPC patients.
Given the fact that PDAC is rich in stromal components, the treatment-induced desmoplastic stromal response, edema, and inflammatory response after radiotherapy can lead to no apparent change in lesion diameter or even pseudo-progression [
26]. As a result, evaluating and reflecting true therapeutic benefits, and stratifying the prognosis based on conventional morphological changes in SD patients are challenging. Various risk stratification methods have already been reported to overcome the drawbacks of RECIST in PDAC [
9‐
11,
27]. Yang et al proposed a radiomics signature to predict outcomes in LAPC patients with SD [
11]. Another study revealed that lower circulating tumor DNA before treatment was associated with SD [
27]. However, the clinical application of these methods was limited due to their complexity, poor repeatability, time-consuming nature, or high cost. Consequently, it is crucial to investigate a new simple, and effective approach for progression risk stratification in SD patients to facilitate precise therapeutic decisions. The four predictors identified in our risk-scoring system were easily accessible and routinely used in the clinic.
Our study showed the high value of ECV derived from PVP indicated a low risk of progression after IORT. This could be explained by the fact that PDAC contains a lot of tumor stroma and a poor blood supply [
7]. ECV has been shown to be associated with fibrosis, desmoplastic stroma, and tissue elasticity [
18]. A reduced ECV value indicates vascular deficiency and severe necrosis induced by hypoxia. Furthermore, tumor hypoxia contributes to the sensitivity of radiotherapy, which results in radiotherapy resistance and a poor prognosis [
28,
29]. In contrast, a high ECV reflects an enlarged extracellular space that might be composed of abundant micro-vessels at the histopathological level, potentially resulting in a low level of hypoxia and thus more sensitivity to radiotherapy. The expansion of extracellular space facilitates the penetration and distribution of chemotherapy drugs, allowing adjuvant chemotherapy to kill tumor cells more effectively. All the above reasons might contribute to the relatively good prognosis in patients with high ECV. In this study, we adopted ECV fraction derived from PVP rather than other delayed times (3, 5, or 10 min) [
18,
30], which effectively reduced the examination time and radiation doses. Moreover, there were also articles that calculated ECV late-arterial-phase [
31]. At present, there is no consensus on the delay time for calculating ECV. PVP, one of the conventional contrast-enhanced phases in pancreatic protocol CT, is routinely used in clinical practice and it might be a potential alternative to the equilibrium phase for calculating ECV [
14]. Meanwhile, no attempt has been made to generate ECV fraction by PVP previously. Although Noid calculated ECV fraction based on late-arterial-phase, they only analyzed the association between ECV and CA19-9 and did not directly investigate its correlation with treatment response [
31]. Our study confirmed that ECV in PVP was an independent predictor of prognosis.
Two CT radiological features, rim enhancement, and peripancreatic fat infiltration, were demonstrated as predictors for the progression of LAPC in our study, as previously reported [
21,
32]. It has been revealed that the hypo-attenuation areas in rim-enhancement were significantly associated with high histological grade, rapid proliferation, few residual acini, and severe necrosis [
21,
33], which might result in tumor hypoxia, consequently leading to resistance to radiotherapy. From this perspective, the presence of rim-enhancement might indicate insensitivity to radiotherapy [
28,
29]. Recently, some studies have found that peripancreatic fat infiltration, which reflects the extent of tumor invasion to surrounding tissues, is associated with a low R0 resection rate and poor prognosis [
12,
32]. These are consistent with the findings of our study.
Serum CA19-9 is a widely used tumor marker, treatment response indicator, and prognostic predictor in PDAC [
14,
34,
35]. CA19-9 positively correlated with tumor load, and a decline over 20%, 50%, or to normality after treatment was reported to be an indicator of favorable prognosis [
34,
36]. A recent clinical trial utilized a 50% reduction in CA19-9 levels after treatment as one of the criteria for evaluating response to treatment [
35], and Newhook et al had previously demonstrated that a decrease of CA19-9 levels over 50% or to normal implies improved overall survival (OS) [
34,
36]. What’s more, the normal range of serum CA19-9 was less than 37 U/mL according to the NCCN guideline for PDAC [
14]. However, as the literature reported, there exist CA19-9 non-secretors with very low CA19-9 levels (≤ 2 U/mL) or normal ranges (< 37 U/mL) [
37,
38]. To avoid difficult or inaccurate assessment of patients with normal CA19-9 levels who had a relatively small decrease in CA19-9 or remained within the normal range, we only included patients with CA19-9 > 37 U/mL. Therefore, our study adopted the criteria of a decrease of over 50% or to a normal range for CA19-9 responders, which confirmed a longer PFS time compared with non-responders.
Our study had several limitations. First, the sample size was relatively small, and the patients were from a single institution and a retrospective cohort. Thus, a prospective cohort from multi-centers is required to validate the results of this study. Second, multiple CT scanners were used due to the retrospective study. A standardized scanning parameter and contrast injection protocols were followed for adopted for all patients in this study, which might minimize the variations between devices and ensure comparable imaging results. Third, we only included LAPC patients with CA19-9 over 37 U/mL, so the scoring system is not applicable to all pancreatic cancer patients, such as Lewis negative PDAC. Fourth, only PFS, but not OS was analyzed in the present study. Fifth, the consistency or correlation between PVP-based ECV and equilibrium-based ECV was not investigated because of the retrospective nature of this study. Prospective research should be performed to evaluate whether PVP-based ECV could replace equilibrium-based ECV in the future. Last, CT-based ECVs are not suitable for patients with CECT contraindications, such as iodine allergy. Whether magnetic resonance imaging-based ECV could replace CT-based ECV remains to be investigated. Therefore, further research is needed to address the above concerns.
In conclusion, ECV derived from PVP can be used in predicting the progression risk in LAPC patients initially assessed as SD after IORT. The risk-scoring system integrating ECV, CT radiological features, and CA19-9 response could serve as an efficient and practical tool for prognosis stratification in LAPC patients with SD. It could assist RECIST v.1.1 to further identify SD patients who might be sensitive to and benefit from IORT accurately, aiding clinicians in choosing individual treatment strategies, preventing tumor progression, and improving the prognosis of patients with LAPC after IORT.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.