Abstract
Background/Aim: This study investigated the feasibility of an integrated scoring system of preoperative prognostic parameters using data from before/after neoadjuvant therapy in patients with borderline resectable pancreatic cancer (BRPC). Patients and Methods: We constructed and analyzed a prognostic scoring system using factors that were previously reported to be significant prognostic indicators or predictors of histological response. Results: We analyzed 28 consecutive patients with BRPC who underwent neoadjuvant therapy and subsequent surgical resection. Overall survival (OS) and recurrence free survival (RFS) were greater in patients with high scores (n=11) than in patients with low scores (n=17; log-rank test p=0.03/0.028). Pathological N0 status (p<0.05) and tumor cell destruction rate >50% (p<0.05) were found at a higher incidence among patients with high scores. Conclusion: OS and RFS can be predicted with an integrated scoring system that uses prognostic indices before/after neoadjuvant therapy for BRPC.
Neoadjuvant therapy is an essential established strategy against various cancers (1-3). Efficacy of neoadjuvant therapy, meanwhile, can differ depending on the type of cancer and region. Such differences may be attributed to the lack of universal indicators for the assessment of the significance of neoadjuvant therapy. Pancreatic ductal adenocarcinoma (PDAC) ranks as the third leading cause of cancer-related deaths and is estimated to become the second leading cause in United States by 2030 (4). In PDAC, neoadjuvant therapy has been considered important for surgical strategy (5). Until now, the most easily implemented response determination method has been the evaluation of tumor diameter by contrast-enhanced CT imagery. However, the response of borderline resectable pancreatic cancer (BRPC) to neoadjuvant therapy is not always reflected by radiographic indicators, according to Katz et al. (6). Pancreatic cancer is typically a tumor accompanied by desmoplastic change, so it is thought that the tumor diameter does not change even if the cancer largely disappears.
Recently, various investigators have reported on the potential of tumor markers in laboratory tests or as parameters in imaging studies before and after neoadjuvant therapy to predict survival time or early recurrence. Sugiura et al. have reported a preoperative CA19-9 value ≥100 U/ml as a significant predictor of early recurrence and a poor prognosis after resection for PDAC (7). Meanwhile, Tsai et al. have reported that normalization of CA19-9 following neoadjuvant therapy, rather than the magnitude of change, was the strongest prognostic marker for long-term survival (8, 9). Yamamoto et al. introduced the preoperative pretreatment value of maximum standardized uptake value (SUV-max) ≥6.0 of 2-deoxy-2-[F-18] fluoro-D-glucose (FDG) positron emission tomography (FDG PET/CT) examination as systemic prognostic factor (10), and so did Akita et al. about a regression index >50% in SUV-max as local prognostic factor (11).
We have previously reported evaluation of apparent diffusion coefficient (ADC) value of whole tumor by diffusion-weighted magnetic resonance imaging (DW-MRI) before and after neoadjuvant therapy as an effective predictor of histological treatment effect (12, 13). When the residual cancer cells are histologically minimal, the prognosis of cancer is favorable (14). The weakness of the prediction of prognosis by a single parameter, however, is that it is affected by population bias. The current study, therefore, investigated the feasibility of an integrated scoring system of preoperative prognostic parameters using data from before and after neoadjuvant therapy. This prospective study uses patients with PDAC. We selected prognostic parameters that were previously reported as significant prognostic indicators or predictors of histological response (13). We constructed and analyzed the prognostic scoring system using the pre- and post-treatment values of CA19-9, SUV-max of FDG PET/CT examination, and whole tumor ADC values on DW-MRI. We compared overall survival (OS) and recurrence free survival (RFS) between patients with high scores and those with low scores. This trial is registered at the UMIN Clinical Trials Registry (UMIN000035672, 000028030) and at ClinicalTrials.gov (NCT02777463).
Patients and Methods
Patients. This study was approved by the Wakayama Medical University Hospital (WMUH) Institutional Review Board (No. 2484, 2092). It is a prospective study defining its protocol treatment as completion of neoadjuvant chemotherapy following pancreatectomy. Patients and methods have been described previously, including a CONSORT diagram (13). During the study period, 31 patients were enrolled. Of them, two did not undergo surgery for disease progression according to restaging imaging studies and one did not complete neoadjuvant therapy because of repeated cholecystitis. Finally, 28 consecutive patients were included in the study, who underwent surgery 3-8 weeks after the final administration of chemotherapy and were all recommended to receive postoperative adjuvant chemotherapy by systemic administration of oral S-1. Resectability status was defined according to National Comprehensive Cancer Network (NCCN) criteria, version 2.2015 (15). During the study period, patients with BRPC received neoadjuvant nab-paclitaxel plus gemcitabine therapy (13). As we have previously reported (12), routine imaging studies included staging and restaging of CT scans, enhanced magnetic resonance imaging (MRI), DW-MRI, and 2-deoxy-2-[F-18] fluoro-D-glucose (FDG) positron emission tomography (FDG PET/CT) prior to and after neoadjuvant therapy.
Neoadjuvant therapy. During the study period all enrolled patients were scheduled for 2 cycles of neoadjuvant nab-paclitaxel plus gemcitabine therapy. One cycle of regimen for BRPC comprised a 30 min intravenous infusion of nab-paclitaxel at a dose of 125 mg/m2, followed by a 30 min intravenous infusion of gemcitabine at a dose of 1000 mg/m2, on days 1, 8 and 15. All patients were then given a week of rest before a second identical cycle (16, 17).
Follow-up. After the registered patients discharged, they were observed as outpatients once per month for six months and then every three months thereafter. Laboratory study including serum CA19-9 levels and other general blood tests were performed at every visit. Dynamic computed tomography was performed every 1-3 months. Patients with elevated tumor markers or lesions suspicious for malignancy underwent FDG PET/CT examination or enhanced magnetic resonance imaging.
Endpoints. The primary endpoint of this prospective study was to confirm the feasibility of a prognostic scoring system with prognostic indices of pancreatic carcinoma. Secondary endpoints included analysis of overall survival and recurrence free survival between patients with high scores and those with low scores.
Definition of prognostic indices and scoring points. We determined six favorable parameters. Three systemically-weighted factors were determined as: value of CA19-9 <100 U/ml prior to neoadjuvant therapy (7), normal post-treatment value of CA 19-9 (8, 9), SUV-max value of FDG PET/CT <6.0 prior to neoadjuvant therapy (10). Three locally-weighted factors were determined as: reduction index of post/pre-treatment SUV-max value >50.0% (11), pre-treatment whole tumor ADC value >1.20×10−3 mm2/s (12), and post-treatment whole tumor ADC value >1.40×10−3 mm2/s (12, 13). Regarding the value of CA 19-9, a patient with Lewisa-b- type was included in this study. The collected data was assessed as favorable or non-favorable; favorable factors were counted as one point each and finally were evaluated as an integrated total point value between zero and six. An integrated point score ≥4 was categorized as a high score, and <4 as a low score (Table I) (7-13).
Diffusion MRI. Diffusion MRI procedures were the same as we have been previously reported (12, 13). Patients underwent MRI within three weeks before the start of neoadjuvant therapy and three weeks after the last administration. The ADC map was constructed using diffusion weighted images at b=0, 50, and 1000 s/mm2. Mean ADC values for each tumor were automatically calculated on the image of ADC map. The same MRI system/scanner, (Intera Achieva 3.0T, Philips Medical Systems, Andover, MA, USA) was used for all patients analyzed in this study. Regions of interest (ROI) were determined by a single experienced MR radiologist as 3-4 of the largest cross-sectional areas representing the whole tumor, as previously reported (12, 13).
18F-fluorodeoxyglucose (FDG) positron emission tomography (PET)/CT. Patients underwent PET/CT and MRI on the same day. PET studies and calculation of tumor SUVmax were performed as previously reported (13). The same PET/CT scanner (SET-3000BCT/L; Shimadzu, Kyoto, Japan) was used for all patients.
Ethical approval. The protocol for this research project has been approved by the Ethics Committee of the institution and it conforms to the provisions of the Declaration of Helsinki. Informed consent was obtained from all individual participants included in the study.
Statistical analysis. Descriptive statistics were calculated to examine the demographic characteristics of the study population. Survival times between the date of registration and death from any cause (without discrimination between deaths resulting from PDAC and from other causes) was taken as the outcome. Survival curves were calculated by the Kaplan-Meier method and comparison of survival curves was based on the log-rank test. Univariate analyses (Chi-square test) were primarily used for selecting variables based on a p-value <0.05. The significant variables and clinically effective factors were subjected to forward logistic regression analysis to determine the net effect for each predictor while controlling the effects of the other factors. The defined prognostic indices were compared between surviving and non-surviving patients, and the presence/absence of metastasis or recurrence between patients with high and low scores by Chi-squared test for categorical variables. All analyses were performed using the statistical software package SPSS II (version 18.0; SPSS, Tokyo, Japan) and JMP ver. 13.
Prognostic scoring system with prognostic indexes of pancreatic carcinoma.
Results
Patient characteristics. Between June 2016 and November 2018, 28 consecutive patients with borderline resectable pancreatic cancer (BRPC) underwent neoadjuvant therapy and subsequent surgical resection. Table II shows the characteristics of the 28 analyzed patients with BRPC. There was one patient with Lewisa-b- type. No patients required metal biliary stent insertion during the period of neoadjuvant chemotherapy. There was no mortality of the registered patients.
Prognostic indices. Figure 1 shows the actual integration score of each prognostic index by individual registered patients. The mean of the actual integration score was 2.9±1.5 (0-5), and the median score was 3 (Figure 2).
Validity of cutoff value for high score. ROC curve was calculated to determine the cutoff value of the actual integration score so that the two groups could be discriminated. The area under the curve for high scores discriminating the two groups was 0.718 (95% confidential interval, 0.538-0.898). Among candidate cutoff values, the final cutoff value was selected using Youden's index. Finally, we adopted 4 as the cutoff value in this study, and total score ≥4 was considered as high score (sensitivity 52.4%, specificity 100%).
Survival. Median follow-up time was 16 months (range=7-36 months). The estimated 1-year, 2-year, and 3-year survival rates were 89%, 72%, and 58% respectively, and the estimated median survival time (MST) was 16 months in all patients, including four cases of mortality. Estimated median recurrence-free survival time (RFS) was 16 months (range=5-36 months). Patients presented recurrence in local sites (n=6), in the liver (n=10), in the lungs (n=1), in the lymph node (n=1), and in the peritoneum (n=5) with some overlap. Table III shows univariate and multivariate analysis of prognostic indices for survival using the six prognostic indices and integrated total score. OS and RFS were greater in patients with high scores (n=11) than in patients with low scores (n=17; log-rank test p=0.03/0.028; Figures 3 and 4).
Demographics and characteristics of patients with BRPC.
The actual integration score of each prognostic index by individual registered patients.
Assessment for independence of prognostic factors. We performed multivariate analysis of prognostic indices and clinicopathological factors for overall survival to investigate the independence of clinically effective factors with the integrated total score. However, integrated total score was not an independent prognostic factor as other clinically effective factors (Table III).
Clinicopathological features of the high score group. The relationship between high/low scores and the significant clinicopathological features is shown in Table IV. Pathological N0 status (p=0.019) and histological response of more than Evans grade IIb (tumor cell destruction rate >50%) (p=0.019) were found in significantly higher incidence in patients with high scores.
Discussion
The present study was considered to be reliable prognostic analysis with a multimodality scoring system using the total and cutoff score of prognostic indices. OS and RFS of patients with high scores were significantly better than those in patients with low scores. To our knowledge, this is the only model that utilizes such preoperative predictors and the first to describe a multimodality scoring system for predicting prognosis in patients who undergo neoadjuvant therapy for PDAC. This concept may apply to various other kinds of cancer that require preoperative therapy.
The mean of the actual integration score was 2.9±1.46 (0-5), and the median score was 3. Y-axis represents the integrated total score, and x-axis represents number of cases.
Originally, it is difficult to predict the overall survival rate by assessing preoperative treatment alone, because the overall survival rate depends on many other variables, such as race, surgical techniques, and regimen of chemotherapy (18). In this context, it is possible to preoperatively predict the long-term results by treatment effect if the method of evaluation of treatment response is appropriate. A universal treatment response assessment method with good accuracy is therefore required.
It is difficult to determine the response to treatment before excision. Currently, the golden standard for determining the response to preoperative treatment is considered to be the histologic death of the tumor cells in the resected specimen. Chua et al. have conducted a systematic review of 17 clinical studies (977 cases) of preoperative chemoradiotherapy for resectable pancreatic cancer between 2000-2010 (19, 20). Pathological complete response was, therefore, obtained at a relatively low rate of 5-15%, partial response was reported as 33-60%, and the minimal response as 38-42%. Since, both, Shimosato and Evans classifications require evaluation by an experienced pathologist, there is a possibility that the institutions are limited, and there is a tendency toward subjective evaluation because it cannot be quantified (21). The most easily implemented response assessment method is the evaluation of tumor diameter by imaging in contrast-enhanced CT. However, according to Evans et al., no cases have shown more than 50% tumor reduction by NACRT (21). Pancreatic cancer is originally a tumor accompanied by desmoplastic change, so it is thought that the cancer diameter does not significantly change even if, to a large extent, the cancer disappears. FDG PET/CT shows uptake of FDG into cells, so it is expected that reduction of tumor cells will reduce FDG accumulation. In esophageal cancer and malignant lymphoma, there are many reports that SUV in PET examinations are useful for response evaluation. Akita et al. have reported the usefulness of PET in pancreatic cancer, and stated that groups with a ≥50% or more decrease in SUV-max before and after NACRT were significantly related to Evans grade III and IV, and the long-term results were also significantly better. The reliability may be lowered, however, if the control of diabetes mellitus is complicated as a peculiarity of pancreatic cancer. If pancreatitis is complicated, FDG may be accumulated in the whole pancreas and it may not be suitable for evaluation. Detection of trace markers by blood tests may compensate for the shortcomings of diagnostic imaging. Regarding tumor markers, CA19-9 has been used, but there may be unreliability owing to biliary obstruction and the possibility that it will not be elevated depending on the race of patients. Each of the above prognostic indices have individual weaknesses in independently predicting prognosis.
Comparison of overall survival curves according to the integrated scoring index. Overall survival rate was significantly higher in patients with high index ≥4 (solid line, n=11) than those with low index <4 (dotted line, n=17) (log-rank test).
Comparison of recurrence-free survival curves according to the integrated scoring index. Recurrence free survival rate was significantly higher in patients with high index ≥4 (solid line, n=11) compared to those with low index <4 (dotted line, n=17) (log-rank test).
Univariate and multivariate analysis of prognostic indexes for overall survival.
In a recent study, Dreyer et al. aimed to define preoperative clinical and molecular characteristics that would allow better patient selection for operative resection for PDAC without neoadjuvant chemotherapy or radiotherapy (22). Among their patients with poor preoperative nomogram scores using existing risk factors, approximately 50% died within a year of resection. In the present study, we used predictors prior to and after neoadjuvant therapy and demonstrated that high scores were negatively associated with lymph node metastasis and the patients with higher than grade IIb cancer had better histological response to neoadjuvant therapy. Similarly, Perri et al. have also revealed radiographic and serologic change in two predictors of pathologic major response to preoperative therapy for pancreatic cancer (23).
High score group and clinicopathological features in pancreatic cancer.
Our results may be limited by selection bias of prognostic indices, the small number of patients, short follow-up period, and the total score ≥4 as high score, which revealed low sensitivity. Differences between populations in resectability categories that produced each prognostic index may be also considered to be a bias in the present study. The real question in clinical situations is whether this is just a prognostic indicator, or if there will be a definitive role of this value in changing therapy. How this analysis will be used to improve therapy or to change the planned approach requires validation.
However, the strength of the multimodality scoring system is that, when applied to different populations, the weakness of each parameter will be compensated by its combination with other parameters and differences in the predictive ability due to population bias for each population will be corrected. The predictive role of the scoring system may be useful in algorithmic treatment approach to neoadjuvant therapy for BRPC as well as resectable pancreatic cancer (24-26). Recently, new surrogate markers obtained by liquid biopsy have attracted attention (27, 28), and it is expected that such novel markers will be gradually developed and used for decision-making regarding preoperative treatment within an integrated scoring system and treatment response assessment on clinical setting during the neoadjuvant therapy. These stronger markers are expected to dramatically improve sensitivity, specificity, and accuracy of the scoring system. This may more precisely predict site/time of recurrence and prognosis after neoadjuvant therapy. A future collaborative study with a larger patient population is therefore needed to confirm the feasibility of this integrated scoring system in predicting prognosis in patients with PDAC.
In conclusion, prediction of OS and RFS with the integrated scoring system using preoperative prognostic indices before/after neoadjuvant therapy for PDAC in this study was feasible. Patients with high scores had favorable survival compared with those with low scores. The poorer prognosis in patients with low scores may be related to more advanced tumors.
Acknowledgements
The Authors would like to thank Benjamin Phillis at the Clinical Study Support Center, Wakayama Medical University Hospital, for proofreading and editing the manuscript.
Footnotes
Authors' Contributions
Study concept and design: KO; Acquisition of data: MM, YK, RK, MU, SH; Analysis and interpretation of data: KT (Statistical analysis); Drafting of manuscript: KO, MK, SH; Critical revision of manuscript: HY.
Conflicts of Interest
All Authors declare no conflicts of interest in relation to this study.
- Received May 10, 2020.
- Revision received June 2, 2020.
- Accepted June 3, 2020.
- Copyright© 2020, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.