Pancreatic ductal adenocarcinoma (PDAC) still has one of the worst prognoses of all cancers and is the fourth leading cause of cancer-related deaths in Japan, Europe, and the United States (1-3). Recently, PDAC has been categorized as resectable (R), borderline resectable (BR), or unresectable (UR) (4). Multidisciplinary treatment, including pre- and post-operative chemotherapy, has become the standard treatment for R-PDAC (5). Even BR- or UR-PDAC may be eligible for surgical resection if preoperative treatment is effective (6-10). New chemotherapy regimens, such as gemcitabine plus nab-paclitaxel (GnP) and FOLFIRINOX have improved the prognosis of UR-PDAC as compared to previous regimens, such as gemcitabine alone (11,12). It has been suggested that, when positive effects are obtained using these new chemotherapy regimens, pancreatectomy may further improve the prognosis of BR- or UR-PDAC (6-10).

However, in clinical practice, not all patients with BR- or UR-PDAC have the opportunity to undergo surgery, and the optimal timing or condition of performing surgical resection remains unclear. In addition, the factors predictive of the effectiveness of chemotherapy, particularly those that can be determined before or early in chemotherapy, are unclear. If the effectiveness of chemotherapy could be predicted early, treatment methods could be changed and the prognosis of PDAC may be improved. Oncological, biological, immunological, inflammatory, and/or nutritional indicators that can be detected before surgery have been reported to be useful in predicting prognosis in patients with surgically resected PDAC (13-23). However, the relationship between these factors and prognosis has not been fully investigated in patients undergoing chemotherapy.

Therefore, we aimed to clarify early determinable prognostic factors, including some pre-chemotherapeutic measurable indicators, to facilitate prediction of treatment effects in PDAC patients receiving GnP.

Study design. Data from 145 patients who underwent GnP chemotherapy for PDAC between 2015 and 2019 at the Department of Surgery, Institute of Gastroenterology, Tokyo Women’s Medical University, Japan, were analyzed in this retrospective study. Patients with unknown examination status (n=12), serum carbohydrate antigen 19-9 (CA19-9) levels ≤37 U/ml (n=10), or pancreatectomy after GnP (n=10) were excluded. Ultimately, 113 patients who underwent GnP treatment for PDAC were analyzed retrospectively.

We evaluated the roles of clinicopathological factors in overall survival (OS). Pre-chemotherapeutic parameters included age, sex, Eastern Cooperative Oncology Group Performance Status score (24), resectability, number of metastatic sites, existence of liver metastasis, existence of pre-treatment, baseline body mass index, maximum tumor size, serum CA19-9 level, hemoglobin A1c (HbA1c) level, neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), lymphocyte-to-monocyte ratio (LMR), prognostic nutrition index (PNI), controlling nutritional status (CONUT) score and Glasgow prognostic score (GPS) (definitions of these indicators are shown in Table I). Intra- and post-chemotherapeutic parameters included the number of chemotherapy cycles, total amount of gemcitabine or nab-paclitaxel, occurrence of adverse events, serum CA19-9 level after two cycles of GnP, rate of change in CA19-9 after two cycles of GnP, response evaluation and post-treatment.

Multivariate analyses to predict the OS were performed with the above factors. The survival rates were further determined based only on the presence or absence of significant factors (as detected by multivariate analysis). This study was approved by the Ethics Committee of the Tokyo Women’s Medical University (approval number: 3952) and performed in accordance with the Declaration of Helsinki; the requirement for informed consent was waived owing to the retrospective nature of the analysis.

Miscellaneous definitions. Pre-chemotherapeutic laboratory and imaging data were acquired within 21 days of GnP initiation and after biliary decompression, with a total bilirubin level <2.0 mg/dl and without cholangitis. Resectability was determined using pre-chemotherapy imaging studies according to the National Comprehensive Cancer Network Clinical Practice Guidelines in Oncology of 2018 (25). In patients with R-, BR-, or UR-LA PDAC, histological evidence was obtained from the primary or metastatic lesion. In a recurrent case, the primary lesion was confirmed to be PDAC based on the resected specimen. The site of recurrence or metastasis was defined as one site each for the liver, lymph nodes, peritoneum, lung, residual pancreas and bone. Recurrent or metastatic cases were defined as the presence of a tumor on the above site by imaging studies. Isolated increases in tumor marker levels alone were not considered as recurrent cases. Pancreatectomy, chemotherapy, and/or chemoradiotherapy before GnP were defined as pre-treatment. Chemotherapy and/or chemoradiotherapy after GnP were defined as post-treatment.

Chemotherapy consisted of S-1 [oral fluorouracil prodrug tegafur, oteracil potassium, and gimeracil in a molar ratio of 1:1:0.4; Taiho Pharmaceutical, Tokyo, Japan (26,27), gemcitabine, and FOLFIRINOX. Patients who underwent pancreatectomy after GnP were excluded from the study.

In this study, we measured the rate of change in CA19-9 between pre-chemotherapy and after two cycles of GnP. Patients whose baseline CA19-9 was within normal limits (CA19-9 ≤37 U/ml) were excluded. As a rule, CA19-9 was measured at every cycle interval of GnP and tumor response was evaluated using dynamic computed tomography at two cycle intervals, according to Response Evaluation Criteria in Solid Tumors version 1.1 (28). Adverse events were evaluated according to the National Cancer Institute’s Common Terminology Criteria for Adverse Events version 5.0 (29). When it was difficult to continue GnP because the patients’ general condition deteriorated and/or it was judged that the effect of GnP was poor, patients underwent another treatment or the best supportive care after GnP. Ultimately, the decision to receive post-GnP treatment was based on the physicians’ or patients’ choices. OS was defined as the time from GnP treatment induction to death or the last follow-up date. The cutoff values for the baseline maximum tumor size, CA19-9, HbA1c, NLR, PLR, LMR, PNI, CONUT score, GPS score (definitions of these indicators are shown in Table I), CA19-9 after two cycles of GnP and the changing rate of CA19-9 after two cycles of GnP were determined using receiver operating characteristic (ROC) curve analysis and designated as the point at which the area under the ROC curve (AUC) was largest for predicting OS.

Statistical analysis. Univariate and multivariate analyses were performed to identify independent predictors of OS. Survival analyses were performed using the Kaplan–Meier method, log-rank test and Cox proportional hazards model. Factors showing statistical significance in univariate analysis were subjected to multivariate analysis. The Kruskal–Wallis test was used to compare the medians of three or more groups. Statistical significance was set at p<0.05. All analyses were performed using JMP 12.1.0, for Windows (SAS Institute Inc., Cary, NC, USA).

The median survival time (MST) was 1.2 years for all patients. The 1- and 3-year OS rates of all patients were 52% and 16%, respectively. The baseline characteristics of the patients are presented in Table II, Table III, Table IV, and Table V. There was no significant difference in the baseline CA19-9 levels according to tumor status (p=0.45) (Table VI). The median time to the highest decrease and re-increase in CA19-9 levels was 84 and 119 days, respectively. Based on ROC analysis, the cutoff values of baseline CA19-9, COUNT score, and changing rate of CA19-9 after two cycles of GnP were 747 U/ml (AUC: 0.66), 5 (AUC: 0.70), and 0.69 (AUC: 0.73).

Overall survival rate based on all risk factors. In multivariate analyses, baseline CA19-9 ≥747 U/ml [hazard ratio (HR)=1.9], baseline CONUT score ≥5 (HR=3.7), and changing rate of CA19-9 after two treatment cycles ≥0.69 (HR=3.7) were independent risk factors for a poor prognosis (Table VII). When examining outcomes according to these prognostic factors, the 1- and 3-year OS rates of patients with a risk score of 0 (n=37), 1 (n=45), 2 (n=27), and 3 (n=4) were 94% and 39% (MST: 2.0 years), 56% and 11% (MST: 1.2 years), 0% and 0% (MST: 0.55 years), 0% and 0% (MST: 0.17 years), respectively (p<0.001) (Figure 1). The HRs of the risk score groups 1, 2, and 3 were 2.8, 15.9, and 201.4 times that of risk score group 0, respectively (p<0.05).

Overall survival rate based on pre-chemotherapy-measurable risk factors. When examining outcomes according to the pre-chemotherapeutic obtainable parameters (baseline CA19-9 ≥747 U/ml and baseline CONUT score ≥5), the 1- and 3-year OS rates of patients with pre-chemotherapy-measurable risk scores (pre-risk scores) of 0 (n=63) were 79% and 27% (MST: 1.7 years); for those with pre-risk scores of 1-2, these were 20% and 4% (MST: 0.65 years), respectively (p<0.001) (Figure 2). The HR of the pre-risk score group 1-2 was 3.3 times that of the pre-risk score group 0 (p<0.001). We defined patients with pre-risk score 0 as pre-chemotherapy low-risk (pre-low-risk) group, and patients with pre-risk score 1-2 as pre-chemotherapy high-risk (pre-high-risk) group.

Overall survival rate based on pre-chemotherapeutic risk and effectiveness of GnP. We grouped the patients according to the effects of GnP. We defined patients with a changing rate of CA19-9 <0.69, after two courses of GnP, as the effective group, and patients with a changing rate of CA19-9 ≥0.69, after two courses of GnP, as the ineffective group. We analyzed the prognosis in these two categories (pre-risk and effectiveness). The 1- and 3-year OS rates of the pre-low-risk and effective group, pre-low-risk and ineffective group, pre-high-risk and effective group, and pre-high-risk and ineffective group were 94% and 39% (MST: 2.0 years), 58% and 12% (MST: 1.2 years), 48% and 10% (MST: 1.2 years), and 0% and 0% (MST: 0.52 years) (p<0.001), respectively (Figure 3). The HRs of the pre-low-risk and ineffective group, pre-high-risk and effective group, and pre-high-risk and ineffective group were 2.8, 2.9, and 20.3 times that of the pre-low-risk and effective groups, respectively (p<0.05) (Table VIII). There were no differences in the 1- and 3-year OS rates between patients in the pre-low-risk and ineffective and pre-high-risk and effective groups (p=0.89).

Time to most decrease and re-increase in CA19-9 levels. In patients in the pre-high-risk and ineffective group, pre-low-risk and ineffective group, pre-high-risk and effective group, and pre-low-risk and effective group, the median time to most decrease and re-increase in CA19-9 level was 61, 73, 86, and 117 days (p=0.012), and 41, 87, 150, and 233 days, respectively (p<0.001) (Table IX).

This study revealed that a baseline CA19-9 level ≥747 U/ml, baseline CONUT score ≥5, and changing rate of CA19-9 after two treatment cycles ≥0.69 were risk factors for shorter OS in patients with PDAC who were receiving GnP. In addition, we were able to predict prognosis in such patients more clearly by considering the former two pre-chemotherapeutic risk factors and the latter effectiveness factor separately. The prognosis for the pre-low-risk and ineffective groups and the pre-high-risk and effective groups were the same. We also clarified the maximum effect period and re-exacerbation period of GnP in each group. These results represent new findings that may be useful for decision-making regarding treatment strategies in patients with PDAC receiving GnP.

Although surgical resection plays an important role in PDAC treatment, it is difficult to achieve long-term survival by pancreatectomy alone. For a long time, surgical resection and adjuvant chemotherapy have been the standard treatments for PDAC (27,30), but the results of treatments have not been satisfactory. In recent years, preoperative treatment for PDAC was reported to reduce micrometastasis, which is difficult to identify on imaging studies, and to increase the curative resection rate of locally advanced PDAC (6-10). Therefore, preoperative treatment is considered necessary to prolong the prognosis of patients with PDAC. A randomized clinical trial reported that two courses of neoadjuvant chemotherapy using gemcitabine plus S-1 improved OS in patients with R-PDAC (5); thus, it has become the standard treatment in Japan (31). Even in patients with BR- or UR-PDAC, surgical resection after neoadjuvant chemotherapy leads to a better curative resection rate and prognosis (6-10). Satoi et al. reported that UR-PDAC patients who underwent pancreatectomy after preoperative treatment for more than 8 months had a better prognosis that those who underwent pancreatectomy after preoperative treatment under 8 months (32). This can be interpreted as a good prognosis when pancreatectomy was performed on patients who had long-lasting chemotherapy, but it is possible that the chance of surgical resection had been missed in the meantime. The optimal format and duration of preoperative treatment thus remain unclear.

GnP and FOLFIRINOX were originally used for UR-PDAC (11,12) and may also be used as preoperative treatment regimens. It has not yet been concluded which is the most effective regimen as neoadjuvant chemotherapy (33). Although FOLFIRINOX use is associated with a higher curative resection rate than GnP use, it has many adverse effects, and it has been reported that it may be difficult to complete neoadjuvant chemotherapy with this drug (6,7). Based on this background, GnP seems to be clinically used more often. Although surgical resection should be performed in the process of maintaining the effect of chemotherapy, the appropriate selection of patients who should undergo surgical resection after GnP and the optimal timing of surgery after GnP remain unclear.

In resected PDAC, oncological factors, such as a large tumor size and elevated serum CA19-9 levels, are associated with a poor prognosis (13,14). Furthermore, it has recently been reported that biological, oncological, immunological, inflammatory, and/or nutritional indicators, such as the NLR, PLR, LMR, PNI, CONUT and GPS, which can be measured before surgery, are good predictors of prognosis in patients with resected PDAC (15-23). The lymphocyte count is a representative index of immunity, which is also regarded as a nutritional parameter, together with serum albumin and total cholesterol levels. A decreased lymphocyte count has been associated with a poor prognosis in malignant tumors (15-17), and poor nutrition status also makes it difficult to continue chemotherapy and reduces its effectiveness. The CONUT score is one of those combined biomarkers and is calculated using serum albumin concentration, total lymphocyte count, and total cholesterol concentration to assess the undernourished status of patients (34) (Table I). It has been reported that the CONUT score more strongly predicts postoperative complications and prognosis compared to other indicators in patients with resected PDAC (19,35).

Few studies have evaluated the relationship between these indicators and prognosis in patients undergoing chemotherapy. The present study found that high CA19-9 and CONUT scores were important poor prognostic factors associated with OS among many of these indicators. The former refers to oncologically advanced tumors, while the latter to patients with a poor nutritional status. These are pre-treatment measurable factors that may also be useful when performing treatments other than GnP. This may imply that the pre-chemotherapy measurable group has a short prognosis, regardless of the treatment used. Baseline CA19-9 levels cannot be changed before treatment; however, nutritional intervention therapy can improve the CONUT score, and whether improving the CONUT score results in improved prognosis remains to be investigated.

Previous studies have reported that the rate of change in CA19-9 levels after two courses of GnP was an early predictive marker of GnP efficacy (36,37). A decrease in CA19-9 ≥50% or 60% after two courses of GnP led to significantly more prolonged OS than a lesser reduction in CA19-9 (36,37). While these results are informative, the cutoff value remains controversial. In addition, in previous reports, the number of patients and examination factors in multivariate analysis were small, and no multivariate analyses included other indicators. To our knowledge, no study prior to ours performed ROC analysis to determine the cutoff value for the changing rate of CA19-9 after two cycles of GnP, or performed multivariate analysis including multiple indicators; thus, the results of this study may be considered more reliable. These results indicated that prognosis may be improved in the pre-low-risk and ineffective groups by changing the treatment.

Although progression-free survival in patients with UR-PDAC receiving GnP was reported to be 5.5 months in the MPACT trial, the time to most decrease and re-increase of the CA19-9 level were not mentioned (11). In the present study, it was clarified that there was a correlation between the effect of GnP and the period of decrease in CA19-9. This information may be useful if surgical resection is planned during the GnP treatment period.

This study has some limitations. Firstly, although there was no statistically significant difference in OS between the presence and absence of pre- and post-GnP treatments, the pre- and post-GnP treatments were not identical in all patients. In particular, concerning post-GnP treatments, a selection bias might have occurred. These variations may have skewed the outcomes during different periods of the study. Secondly, because patients with baseline CA19-9 ≤37 U/ml were excluded from this study, the results may not be generalizable to such patients. Finally, this was a retrospective study performed at a single institution, and included a limited number of participants. The biases inherent to such settings should be taken into consideration.

In conclusion, we were able to predict the prognosis of PDAC patients who were treated with GnP, based on pre- and intra-chemotherapeutic factors. The CA19-9 level, the CONUT score, the rate of change in CA19-9 level over two treatment cycles, and the treatment validity period may be particularly useful for predicting prognosis in this setting. Future studies should aim to determine the prospective utility of these laboratory markers and construct the best therapeutic strategy for patients with PDAC who are being treated with GnP.

The Authors declare that they have no competing interests.

WI and RH contributed to conception and design. WI, RH, TF, TY, SU, YM, MS, and MY contributed to development of methodology and data acquisition. WI, RH, TF, and MY contributed to analysis and interpretation of data, writing, review, and/or revision of the manuscript. All Authors have read and approved the final manuscript.

This study was financially supported by NAKAYAMA Komei Research Fellowship Grant.