Abstract
Purpose:
Despite advances in neoadjuvant therapy and surgery, a limited increase in the 5-year survival rate of patients with pancreatic ductal adenocarcinoma (PDAC) has occurred, implying the importance of postoperative management. This study aims to achieve risk stratification of patients with PDAC, providing guidance for clinicians to make better regimen choices and realizing precision treatment.
Experimental Design:
Five protein posttranslational modifications (PTM) were detected to find out the most prognostic marker using the IHC method. The predictive and prognostic performance of lysine pan-crotonylation (pan-Kcr) expression standing out from the initial selection was measured in the discovery cohort and validated in two independent cohorts. A risk stratification model was constructed based on pan-Kcr expression.
Results:
A total of 309 (93, 114, and 102 from Qingdao, Tianjin, and Jilin cohorts, respectively) patients with PDAC were included. Pan-Kcr expression was the only significantly deregulated PTM between cancer and its paracancerous tissues. The low pan-Kcr level was an independent predictive (HR, 0.470; P = 0.0090) and prognostic (HR, 0.312; P < 0.0001) marker for patients with PDAC. We constructed a risk-stratification model based on pan-Kcr expression, and high-risk patients suffered a significantly shorter disease-free survival and overall survival (median disease-free survival, 5 vs. 12.5; P < 0.0001; median overall survival, 7 vs. 24; P < 0.0001). The model was well validated in two independent cohorts.
Conclusions:
Pan-Kcr is an independent predictive and prognostic marker for resected patients with PDAC. The risk stratification could provide guidance for clinicians to make precision regimen choices. Also, our findings illustrated the feasibility of PTM-guided prognosis and biomarker-aided PDAC adjuvant therapy.
Translational Relevance.
Sensitive methods for stratifying recurrence risk, monitoring treatment efficacy, and detecting early recurrences could significantly influence treatment decisions and outcomes for patients with pancreatic cancer. Protein posttranslational modifications that can quickly and reversibly respond to environmental or intracellular stimuli are a valuable resource for biomarker candidates and therapeutic targets. We found that low lysine pan-crotonylation (pan-Kcr) expression is predictive for worse prognosis of patients with resected pancreatic ductal adenocarcinoma. Importantly, the pan-Kcr–based nomogram model could risk-stratify patients to predict the recurrence time. Accordingly, our study illustrated the feasibility of posttranslational modification–guided prognosis and biomarker-aided pancreatic ductal adenocarcinoma adjuvant therapy and underscored the need for further interventional studies assessing the clinical utility of pan-Kcr–based risk stratification. An emerging opportunity lies in the risk-stratified allocation of surveillance resources, which could enhance both the cost-effectiveness and overall clinical outcomes.
Introduction
Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal tumors with a 5-year survival rate of only 13% (1). Surgery remains the cornerstone and the only curative treatment for patients with resectable PDAC. However, despite the use of neoadjuvant therapy and improvements in surgical techniques, the 5-year survival rate of patients with regional PDAC has improved by only 3% in the last 5 years (1), implying that more attention should be paid to postoperative management.
Due to the specificity of early dissemination, all patients with resected PDAC are advised to receive postoperative adjuvant therapy according to the National Comprehensive Cancer Network (NCCN) guidelines (2). The CONKO-001 opens a milestone in adjuvant therapy for resected PDAC, which showed significant improvements in disease-free survival (DFS) and overall survival (OS) with the use of postoperative gemcitabine versus observation (3). Subsequently, the following regimens have been recommended for resected PDAC: gemcitabine plus capecitabine; bolus 5-fluorouracil plus leucovorin; capecitabine; gemcitabine plus albumin-bound paclitaxel; and modified FOLFIRINOX. For those with features that portend a high risk for local recurrence (e.g., positive resection margins), adjuvant radiotherapy is also recommended after resection (2). Additionally, S-1, an oral drug containing tegafur, is recommended for its higher efficacy, as compared with gemcitabine monotherapy, for its availability as an oral formulation, and its good tolerance (4). From the results of clinical trials, the intensity and drug combinations cannot be equated with efficacy. S-1 showed superiority to gemcitabine in OS for adjuvant chemotherapy (ACT; ref. 5). No significant difference was observed in the OS between fluorouracil plus folinate calcium or gemcitabine alone in the postoperative ACT, and severe adverse events were notably reduced in the gemcitabine-alone group compared with the fluorouracil plus folinate calcium group (6). Different regimens possess different efficacies and toxicities, and thus, how to risk-stratify patients for the optimal regimen is the current clinical bottleneck and may be the key for improving the DFS and OS of patients with PDAC.
Despite advancements in sequencing and multiomics, serum carbohydrate antigen 19-9 (CA 19-9) is presently the only FDA-approved biomarker widely used for preoperative prognostication of PDAC. However, the high false-positive rate due to other pathologic conditions and the 10% false-negative rate limit its use clinically (7, 8). Minimal residual disease assessment based on ctDNA is a promising biomarker to guide postoperative therapy of PDAC, but quantifying minute amounts of ctDNA is challenging and expensive. In addition, studies also explored RNA-based signatures to predict the sensitivity of chemotherapy (9, 10); however, RNA-based signatures reliant on high-throughput sequencing and the easy degradation feature of RNA lead to unstable results. In contrast, IHC staining not only can characterize the density and spatial distribution of certain markers (11) but holds advantages in stability, reproducibility, and simplicity. Exploring markers that can be detected using IHC methods would provide a broad application prospect.
Posttranslational modifications (PTM) are chemical modifications on specific amino acids that modulate conformation, activity, interaction, stability, and spatial distribution of most eukaryotic proteins (12). PTM can quickly and reversibly respond to environmental or intracellular stimuli and are a valuable resource for biomarker candidates and therapeutic targets. PTM such as phosphorylation (13), glycosylation (14), acetylation (15), methylation (16), and crotonylation (17) have been shown to play a role in the occurrence, progression, and metastasis of cancer. Given the hopeful role of PTM in prognostic and therapeutic potential, we previously detected the expression of protein methylation, acetylation, ubiquitination, and crotonylation using tissue samples from patients with resected PDAC to figure out whether PTM relate to disease recurrence. Patients were divided into two groups: those who experienced relapse or metastasis within 3 months and those with DFS longer than 2 years following radical surgery. Results demonstrated that lysine pan-crotonylation (pan-Kcr) was significantly differentially expressed between the two groups (Supplementary Fig. S1A–S1E), suggesting that it may be used as a marker for risk stratification of patients with resected PDAC.
Here, we analyzed the expression of pan-crotonylation in patients with resected PDAC using the IHC method, and the results demonstrated that low pan-crotonylation expression was an independent predictive and prognostic marker. Additionally, we constructed a risk-stratification model based on pan-crotonylation, and importantly, it was well validated in two independent cohorts. According to our risk-stratification model, patients with a high recurrence risk may be more suitable for intensive ACT and/or chemoradiation, whereas for those with a low recurrence risk, single gemcitabine or S-1 can be considered. Collectively, our findings illustrated the feasibility of PTM-guided prognosis and biomarker-aided PDAC adjuvant therapy.
Materials and Methods
Study population
This retrospective study consisted of patients with PDAC from three cohorts with formalin-fixed, paraffin-embedded tissue specimens. The discovery cohort, Qingdao cohort, included 93 patients with PDAC collected from the Affiliated Hospital of Qingdao University from January 2017 to December 2021. Two independent validation cohorts were included in this study. The Tianjin cohort consisted of 114 patients with PDAC from Tianjin Medical University Cancer Institute & Hospital between December 2006 and December 2010, and the Jilin cohort included 102 patients with PDAC from The First Affiliated Hospital of Jilin University between February 2018 and August 2022. All patients received radical surgery and had histologically confirmed PDAC. Patients with stage IV and other pathologic type were excluded. The disparity in patient numbers between the OS and DFS analyses is attributable to patients being lost to follow-up. The median follow-up durations were 15 (range, 2–71), 12 (range, 2–36), and 17 (range, 1–56) months for patients in the Qingdao, Tianjin, and Jilin cohorts, respectively. The median ages of the Qingdao, Tianjin, and Jilin cohorts were 60, 59, and 60 years, respectively. This study is conducted in accordance with the Declaration of Helsinki. Prior written informed consent from all patients and approval from the ethical review board of each hospital were obtained for experimentation with human subjects.
IHC
Tissue slides (4-μm thickness) were prepared from formalin-fixed, paraffin-embedded samples, deparaffinized, and rehydrated through graded ethanol. Antigen retrieval was performed using Tris-EDTA Antigen Repair Solution and microwave heating. The endogenous peroxidase activity was quenched by 0.3% H2O2, and the nonspecific binding was blocked by 5% goat serum. Slides were then incubated with the primary pan-crotonylation (AB_2877694, PTM Biolabs), pan-ubiquitination (AB_3676744, PTM Biolabs), pan-methylation (AB_3677442, Abnova), pan-phosphorylation (AB_2936786, Abmart), and secondary antibodies (AB_3675450, ZSGB-Bio), separately. IHC staining was visualized using 3,3′-diaminobenzidine (AB_2575228, Thermo Fisher Scientific). Immunoreactivity was semiquantitatively scored according to the estimate of staining intensity and extent of tumor cells as previously described (18). The final scores ≤4 were considered negative staining, whereas those >4 were considered positive staining. The surv_cutpoint method in R (SCR_001905, V.4.0.2) was used to determine the optimal cutoff value of pan-Kcr expression.
Nomogram model construction and validation
The nomogram model was constructed based on the Qingdao cohort and validated in Tianjin and Jilin cohorts. The multivariable Cox proportional hazards regression model was used to evaluate factors associated with DFS and OS. The discriminative ability of the nomogram was evaluated by the consistency index. The calibration curve was performed to compare predicted and actual survival, and the area under the ROC curve was used to compare the accuracy of the nomogram and tumor–node–metastasis (TNM) staging system for DFS and OS prediction. The nomogram model for predicting the OS rate and median survival time of patients with PDAC was constructed using “foreign” and “rms” packages in R.
Statistical analysis
Statistical analysis was conducted in SPSS software (SCR_002865, V.22.0), as appropriate. χ2 tests or Fisher exact test was used to determine differences in clinicopathologic variables. Cox proportional hazards regression was used to evaluate the HR for those positive risk factors. A comparison of quantitative data between two groups was performed using t tests. The Kaplan–Meier method, log-rank test, and univariate test were performed to analyze survival curves. A two-tailed P value of < 0.05 was defined as statistically significant.
Data availability
The data generated in this study are not publicly available, as they contain information that could compromise patient confidentiality, but they are available upon reasonable request to the corresponding author.
Results
Patient characteristics
A total of 309 patients with PDAC from three institutes were included in this study: 93 in the Qingdao cohort, 114 in the Tianjin cohort, and 102 in the Jilin cohort. The clinicopathologic information of the three cohorts is provided (Supplementary Tables S1–S3, respectively). The median ages of the Qingdao, Tianjin, and Jilin cohorts were 60 years (range, 28–81 years), 59 years (range, 31–79 years), and 60 years (range, 37–83 years), respectively. All patients were at stage I to III, and the proportion of patients across different stages was similar in the three cohorts.
Association between pan-Kcr expression and clinicopathologic features
The expression pattern of pan-Kcr in PDAC is unclear. Here, we performed IHC staining of pan-Kcr in PDAC and its paracancerous tissues (Fig. 1A). Pan-Kcr expression predominantly detected in cytoplasm was significantly lower in cancer compared with its paracancerous tissues (Fig. 1B). Low expression of pan-Kcr was significantly correlated with regional lymph node metastasis and higher TNM stage (Supplementary Table S4) in the Qingdao cohort and related to higher carcinoembryonic antigen levels in the Jilin cohort (Supplementary Table S5). No correlation was found between pan-Kcr and the clinicopathologic features in the Tianjin cohort (Supplementary Table S6). Interestingly, low expression of pan-Kcr was significantly associated with no smoking and no drinking history in the Qingdao cohort (Supplementary Table S4). These results uncovered the expression pattern of pan-Kcr in PDAC and suggested its prometastatic role.
Figure 1.
Expression and prognostic significance of pan-Kcr expression in patients with PDAC of the Qingdao cohort. Pan-Kcr expression in cancer and paracancerous pancreatic tissues of patients with pancreatic cancer (A and B). Kaplan–Meier estimates of DFS (C) and OS (D) for patients with high and low pan-Kcr expression based on the Qingdao training cohort.
Prognostic significance of pan-Kcr expression in patients with PDAC
The data of the Qingdao cohort demonstrated that low expression of pan-Kcr when compared with high pan-Kcr expression was associated with an increased risk of recurrence with a median DFS of 4 versus 9 months (P = 0.0014; Fig. 1C) and a significantly inferior median OS of 7.5 versus 24 months (P < 0.0001; Fig. 1D). The predictive and prognostic role of pan-Kcr was well validated in the two independent cohorts, Tianjin (median DFS: 4 vs. 10 months, P = 0.0220; median OS: 9 vs. 14.5 months, P = 0.0002; Fig. 2A and B) and Jilin cohorts (median DFS: 8 vs. 15 months, P = 0.0002; median OS: 13.5 vs. 26 months, P = 0.0003; Fig. 2C and D). Pooled analysis of discovery and validation cohorts confirmed that low pan-Kcr expression remained significantly associated with poorer DFS (P < 0.0001) and OS (P < 0.0001), reinforcing its robustness as a prognostic biomarker across heterogeneous populations (Supplementary Fig. S2A and S2B).
Figure 2.
Prognostic significance of pan-Kcr expression in patients with PDAC of two validation cohorts. Kaplan–Meier estimates of DFS and OS for patients with high and low pan-Kcr expression based on Tianjin (A and B) and Jilin (C and D) validation cohorts.
Importantly, on multivariate and univariate analyses, low pan-Kcr expression was the only variable that had a statistically significant association with disease recurrence and poor OS both in the discovery [DFS: HR, 0.465; 95% confidence interval (CI), 0.264–0.820, P = 0.0080; OS: HR, 0.305, 95% CI, 0.179–0.520, P < 0.0001; Table 1] and validation cohorts (Tianjin cohort DFS: HR, 0.547, 95% CI, 0.324–0.921, P = 0.0230; Tianjin cohort OS: HR, 0.419, 95% CI, 0.225–0.689, P = 0.0010; Jilin cohort DFS: HR, 0.352, 95% CI, 0.184–0.670, P = 0.0010; Jilin cohort OS: HR, 0.449, 95% CI, 0.252–0.800, P = 0.0070; Supplementary Tables S7 and S8).
Table 1.
Univariate and multivariate analyses of DFS and OS in the Qingdao cohort.
| Variables | DFS | OS | ||||||
|---|---|---|---|---|---|---|---|---|
| Univariate analysis | Multivariate analysis | Univariate analysis | Multivariate analysis | |||||
| HR (95% CI) | P | HR (95% CI) | P | HR (95% CI) | P | HR (95% CI) | P | |
| All patients (n = 93) | ||||||||
| Age (years) | ||||||||
| ≤60 | — | — | ||||||
| >60 | 1.001 (0.606–1.653) | 0.9980 | 1.216 (0.758–1.951) | 0.4170 | ||||
| Gender | ||||||||
| Male | — | — | ||||||
| Female | 0.884 (0.514–1.519) | 0.6550 | 1.132 (0.677–1.892) | 0.6360 | ||||
| Smoking | ||||||||
| No | — | — | ||||||
| Yes | 0.857 (0.516–1.423) | 0.5500 | 0.798 (0.492–1.294) | 0.3600 | ||||
| Drinking | ||||||||
| No | — | — | ||||||
| Yes | 0.884 (0.520–1.505) | 0.6500 | 0.768 (0.456–1.292) | 0.3200 | ||||
| Tumor size | ||||||||
| ≤4.0 cm | — | — | ||||||
| >4.0 cm | 0.769 (0.399–1.481) | 0.4330 | 0.754 (0.409–1.409) | 0.3760 | ||||
| Regional LN metastasis | ||||||||
| No | — | — | — | — | ||||
| Yes | 3.296 (1.901–5.713) | 0.0001 | 3.291 (1.864–5.811) | 0.0001 | 1.721 (1.067–2.778) | 0.0260 | 2.878 (1.390–5.961) | 0.0040 |
| TNM | ||||||||
| I | — | — | — | |||||
| II | 2.286 (1.329–3.932) | 0.0030 | 1.0260 (0.574–1.924) | 0.8650 | 2.302 (0.728–5.728) | 0.1760 | ||
| III | 3.504 (1.383–8.883) | 0.0080 | 2.419 (1.014–5.772) | 0.0470 | 4.320 (1.737–10.743) | 0.0020 | ||
| PNI | ||||||||
| No | — | — | ||||||
| Yes | 0.928 (0.531–1.620) | 0.7920 | 1.945 (0.993–3.812) | 0.0530 | ||||
| Pathologic grading | ||||||||
| Low | — | — | ||||||
| Medium–high | 0.546 (0.308–0.970) | 0.0390 | 0.936 (0.540–1.622) | 0.8140 | ||||
| CA 19-9 | ||||||||
| ≤39 u/mL | — | — | ||||||
| >39 u/mL | 0.828 (0.448–1.533) | 0.5490 | 0.861 (0.440–1.685) | 0.6620 | ||||
| CEA | ||||||||
| ≤5.0 ng/mL | — | — | ||||||
| >5.0 ng/mL | 0.852 (0.512–1.416) | 0.5360 | 0.915 (0.559–1.496) | 0.7230 | ||||
| Pan-Kcr expression | ||||||||
| Low | — | — | — | — | ||||
| High | 0.463 (0.271–0.792) | 0.0050 | 0.465 (0.264–0.820) | 0.0080 | 0.374 (0.231–0.606) | 0.0001 | 0.305 (0.179–0.520) | 0.0001 |
| NLR | ||||||||
| ≤3.09 | — | — | ||||||
| >3.09 | 2.063 (1.161–3.664) | 0.0130 | 1.261 (0.754–2.108) | 0.3770 | ||||
| PLR | ||||||||
| ≤139.63 | — | — | ||||||
| >139.63 | 1.199 (0.716–2.008) | 0.4900 | 0.783 (0.483–1.268) | 0.3200 | ||||
| MLR | ||||||||
| ≤0.45 | — | — | ||||||
| >0.45 | 1.611 (0.877–2.958) | 0.1240 | 1.200 (0.701–2.057) | 0.5060 | ||||
Abbreviations: CEA, carcinoembryonic antigen; LN, lymph node; MLR, monocyte-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; PNI, perineural invasion. Values with P < 0.05 are in bold.
Prognostic nomogram model establishment and application
In order to better predict the recurrence and OS of patients with PDAC, we established a nomogram model. Variables included in the nomogram were determined according to the results of multivariable Cox proportional hazards regression analysis in the Qingdao cohort. Pan-Kcr expression and lymph node metastasis, both independent predictive factors, were ultimately used to build the DFS prediction model (Fig. 3A), and pan-Kcr expression, lymph node metastasis, and TNM stage were included in the OS prognostic model (Fig. 3B). The point of each variable, total points, and the corresponding recurrence and survival rates are listed in Supplementary Table S2. Patients were stratified into high-risk (risk score ≥62.85 for DFS and risk score ≥132.93 for OS) and low-risk (risk score <62.85 for DFS and risk score <132.93 for OS) groups according to the risk scores and the optimal cutoff value determined using the surv_cutpoint method.
Figure 3.
Pan-Kcr expression–based nomogram model for predicting DFS and OS of the Qingdao cohort. Pan-Kcr expression based–nomogram model for DFS (A) and OS (B). Kaplan–Meier estimates of DFS (C and D) and OS (E and F) for patients with high and low risk based on the Qingdao training cohort. LN, lymph node.
In the Qingdao cohort, high-risk patients had a significantly poorer DFS (Fig. 3C and D) and OS (Fig. 3E and F) than low-risk patients (P < 0.0001 for DFS and P < 0.0001 for OS). At 2 years following surgery, only one patient in the high-risk group remained disease-free, and four patients were still alive (2-year DFS: 2.38% vs. 19.23% for the low-risk group; 2-year OS: 12.50% vs. 52.46% for the low-risk group). When the same cutoff value was applied to the Tianjin and Jilin validation cohorts, DFS and OS in the high-risk group were also significantly worse than those in the low-risk group [P = 0.0014 for DFS and P = 0.0043 for OS in the Tianjin cohort (Fig. 4A and B); P = 0.0490 for DFS and P = 0.0013 for OS in the Jilin cohort (Fig. 4C and D)]. Similarly, at 2 years following surgery, the majority of patients in the high-risk group suffered recurrence and death [2-year DFS: 0% vs. 7.14% for the low-risk group, 2-year OS: 3.70% vs. 9.09% for the low-risk group in the Tianjin cohort (Supplementary Fig. S3A and S3B); 2-year DFS: 26.98% vs. 63.64% for the low-risk group, 2-year OS: 46.67% vs. 68.75% for the low-risk group in the Jilin cohort (Supplementary Fig. S3C and S3D)]. All the above results suggested that the nomogram model, which serves as a risk stratification instrument, might be helpful for guiding the choice of chemotherapy regimens.
Figure 4.
Pan-Kcr expression–based nomogram model for predicting DFS and OS of two validation cohorts. Kaplan–Meier estimates of DFS and OS for patients with high and low risk based on Tianjin (A and B) and Jilin (C and D) cohorts.
Calibration and validation of the nomogram model
Calibration curves with 1,000 bootstrap resamples were constructed to assess the concordance between the actual probability and the nomogram-predicted DFS and OS in Tianjin (Fig. 5A and B) and Jilin cohorts (Fig. 5C and D). The calibration plots for postoperative 0.5-year, 1-year, and 2-year DFS indicated a high level of agreement between the predictions derived from the nomogram based on pan-Kcr expression and the actual observations, particularly for 1-year DFS. Similarly, the calibration plots for postoperative 0.5-year, 1-year, and 2-year OS indicated that the nomogram predictions were largely in line with the actual observations, especially for 2-year OS.
Figure 5.
Calibration curves for predicting DFS and OS rates. Calibration curves for predicting the half-year, 1-year, and 2-year DFS and OS of Tianjin (A and B) and Jilin (C and D) cohorts compared with that of Qingdao cohort. The plots along the 45-degree line indicate an appropriate calibration model, in which the predicted probabilities were identical to the actual outcome.
Comparison of the nomogram model with TNM stage
The TNM stage system is currently the accepted method to predict prognosis. We compared the nomogram model with TNM stage using decision curve analysis and ROC curve analysis. The decision curve analysis curves indicated that the nomogram for 0.5-year, 1-year, and 2-year DFS and OS had significantly better predictive clinical application than TNM stage in all three cohorts (Supplementary Fig. S4A–S4F). Similarly, ROC curves demonstrated that the nomogram showed stronger predictive power than TNM stage in 0.5-year, 1-year, and 2-year DFS and OS in all three cohorts (Supplementary Fig. S5A–S5F). To assess the independent predictive value of pan-Kcr expression, we also compared the model with lymph node metastasis to predict DFS and with lymph node metastasis plus TNM stage to predict OS. The model demonstrated a significant advantage over other models excluding pan-Kcr expression in predicting DFS and OS (Supplementary Fig. S6A–S6I).
Discussion
Pancreatic cancer has the lowest 5-year survival rate among all malignant tumors. Over the past few decades, due to the limited performance and inconsistent predictive values of clinicopathologic variables for survival, several molecular prognostic markers and subtypes based on integrated analysis of genomic and transcriptomic data have been identified (19–21). These studies have largely expanded our understanding of the molecular underpinnings of PDAC, except for CA 19-9, and currently, no highly predictive and well-validated biomarker is available clinically. Protein is the bridge between the genotype and phenotype; however, the dynamic decoration of PTM clutches the protein structural and functional alterations during the proteins' lifespan. Therefore, in this study, we examine the expression level of five PTM, namely, phosphorylation, ubiquitination, methylation, acetylation, and crotonylation, and ultimately, only pan-Kcr expression was significantly differentially expressed between PDAC cancer and its paracancerous tissues. The predictive and prognostic value of pan-Kcr expression was verified in our Qingdao cohort and two independent validation cohorts. Our results, derived from both our in-house and external cohorts, provide reliable insights into patient outcomes and improve the stratification of treatment approaches.
Despite numerous significant trials over the past 2 decades that have demonstrated the benefits of ACT for patients with PDAC, oncologists continue to face challenges in selecting the most appropriate therapeutic regimen for each individual (9, 22). This decision-making process requires careful evaluation of the tolerable risks in relation to the potential benefits. Particularly after the report of the noninferior effect of S-1 versus gemcitabine (4), there is an increasing demand for comparative studies between S-1 and other intensive chemotherapy regimens, which emphasizes the need for personalized medicine approaches that select the most effective treatment based on specific biomarkers.
With the development of multiomic sequencing, the molecular landscape of cancers continues to be unveiled. Some molecular profiling based on genomic, transcriptomic, and proteomic sequencing has been proven to assess phenotypic features (23–25) and has shown promise in predicting sensitivity of certain chemotherapy regimens (9, 22, 26). However, when it comes to the clinical application, CA 19-9 is still the only available biomarker to predict recurrence and prognosis of patients with PDAC. Here, we focused on protein PTM and discovered that lysine crotonylation (Kcr) expression is a powerful prognostic biomarker for patients with resected PDAC. Crotonylation was first recognized on lysine residues in histones that are prevalent in the promoter and enhancer regions of both human somatic and male germinal cells, suggesting that Kcr of histones could serve as a marker for gene expression (27). Several recent studies have elucidated that Kcr is implicated in cancer settings. It has been reported that 4.5% of cancer protein biomarkers were crotonylated according to the Early Detection Research Network database, and 5.9% of genes were identified to encode crotonylated proteins using P300 (28). Crotonylated proteins were found to be widely expressed in human tumor tissues. The global levels of Kcr were observed to decrease in liver, stomach, and kidney cancers and were elevated in lung, colon, esophageal, and thyroid cancers (29). In pancreatic cancer, only one study compared the pan-Kcr expression between a small sample size of cancer and its paracancerous tissues, which showed a higher expression level in cancer tissues (29). The varied expression patterns of Kcr suggest that it may play multiple roles in cancer progression, and global changes in Kcr levels may partly reflect its involvement in cancer advancement. More specific and critical factors related to crotonylation that work as biomarkers and regulate cancer progression are yet to be discovered.
Our study identified the elevated preoperative neutrophil-to-lymphocyte ratio (NLR) as a significant predictor of reduced DFS in univariate analysis and was also significantly related to pan-Kcr expression. These findings align with meta-analytic evidence linking NLR to poor prognosis (30), suggesting that NLR reflects a conserved inflammatory mechanism driving tumor progression, potentially through neutrophil-mediated immunosuppression. However, the relationship between pan-Kcr expression and NLR may suggest that neutrophilic inflammation may impair crotonylation-dependent tumor suppression via cytokine-mediated epigenetic dysregulation. Future studies should explore the regulation mechanism between PTM and inflammation.
There is no doubt that ACT can significantly enhance the survival rates of patients with PDAC who underwent radical resection, as demonstrated in well-designed randomized controlled trials (31, 32). The recent clinical trial has shown that the intensified regimen FOLFIRINOX demonstrates superior efficacy in patients with resected PDAC compared with gemcitabine alone (33). However, in real-world clinical practice, even in the United States, only about half of the patients received the FOLFIRINOX regimen, and an even smaller proportion completed the recommended six cycles of treatment (34). Although FOLFIRINOX was proven to have higher efficacy by clinical trials, the incidence of severe side effects (grades 3 and 4) reached nearly 70% (33), which implies that not all patients can benefit from it. This finding highlights the necessity for patient stratification. In this study, based on pan-Kcr expression, we constructed a nomogram model, which could risk-stratify patients with resected PDAC. High-risk patients exhibited significantly lower DFS and OS, indicating a likely need for more urgent intensified ACT such as FOLFIRINOX, and for those with low recurrence risk, gemcitabine or S-1 may be a preferable choice. It is important to note that the current study revealed the predictive and prognostic value of pan-Kcr expression in pancreatic cancer, and its directly predictive role in guiding regimen implementation remains to be tested in the future.
Studies have presented evidence that histone deacetylases (HDAC) work as the major histone decrotonylase, and HDAC inhibitors (HDACi) can significantly upregulate the expression of crotonylation (35–37). Recent clinical trials have demonstrated that HDACi exert an excellent anticancer effect. In microsatellite stable/proficient mismatch repair advanced colorectal cancer, the combination of an HDACi, a PD-1 antibody, and a VEGF antibody showed a significantly prolonged progression-free survival and OS (38). In relapsed or refractory peripheral T-cell lymphomas, a single HDACi, romidepsin, was associated with a 25% response rate and a 15% complete response rate (39). Low pan-Kcr expression is significantly associated with adverse clinical outcomes, and our ongoing mechanistic studies indicate that increasing crotonylation levels in pancreatic cancer cells markedly inhibits metastatic capacity in vitro and in vivo. We have initiated a phase II, investigator-initiated trial (ChiCTR2500095869) to evaluate the efficacy and safety of HDACi (acting as histone decrotonylases) in combination with a gemcitabine and albumin-bound paclitaxel regimen in advanced pancreatic cancer. Pretreatment tumor biopsies will undergo a pan-Kcr IHC assay to test the hypothesis that pan-Kcr–low patients exhibit enhanced sensitivity to HDACi-mediated epigenetic reprogramming.
In conclusion, our retrospective study unveiled the expression pattern of Kcr and established a risk model to predict recurrence and survival for patients with resected PDAC, which could provide guidance for clinicians to obtain a relatively accurate recurrence prediction and postoperative regimen decision. Our findings were rigorously validated in two independent cohorts. In addition to providing insights into patient outcomes, our study brings more implications for the treatment of pancreatic cancer.
Supplementary Material
Supplementary Figure S1. The expression of five protein post-translational modifications in pancreatic tissue of patients with DFS <3 months and DFS >24 months. The representative expression picture and their quantitative results of pan-crotonylation (A), pan-phosphorylation (B), pan-methylation (C), pan-ubiquitination (D), and pan-acetylation (E) in pancreatic tissue of patients with DFS <3 months and DFS >24 months. *P < 0.05, ns: nonsignificant.
Supplementary Figure S2. Prognostic significance of pan-Kcr expression in PDAC patients of all three cohorts. Kaplan-Meier estimates of DFS (A) and OS (B) for patients with high and low pan-Kcr expression based on all three cohorts.
Supplementary Figure S3. DFS and OS analysis according to the pan-Kcr based nomogram model. Bar chart of DFS (A) and OS (B) in low-risk and high-risk pancreatic cancer patients in the Tianjin cohort. Bar chart of DFS (C) and OS (D) in low-risk and high-risk pancreatic cancer patients in the Jilin cohort.
Supplementary Figure S4. Decision curve analysis in the training and validation cohorts. The nomograms and 8th edition AJCC-TNM staging system for predicting half-a-year, 1-year and 2-year DFS and OS of Qingdao (A, B), Tianjin (C, D), and Jilin (E, F) cohorts.
Supplementary Figure S5. ROC curve analysis of the nomogram model and the TNM staging system. ROC curve analysis of the nomograms and 8th edition AJCC-TNM staging system for predicting half-a-year, 1-year, 2-year DFS and OS in Qingdao (A, B), Tianjin (C, D) and Jilin (E, F) cohorts.
Supplementary Figure S6. ROC curve analysis of the full nomogram comparing LN metastasis for predicting DFS and OS, as well as LN metastasis + TNM stage for predicting OS. ROC curve analysis of the nomogram comparing lymph node (LN) metastasis for predicting half-year, 1-year, and 2-year disease-free survival (DFS) and overall survival (OS), as well as comparing LN metastasis + TNM stage for predicting half-year, 1-year, and 2-year OS in the Qingdao (A-C), Tianjin (D-F), and Jilin cohorts (G-I).
Supplementary Table S1. Clinicopathological Characteristics of Qingdao-cohort.
Supplementary Table S2. Clinicopathological Characteristics of Tianjin-cohort.
Supplementary Table S3. Clinicopathological Characteristics of Jilin-cohort.
Supplementary Table S4. Association of pan-Kcr expression with clinicopathological features in Qingdao-cohort.
Supplementary Table S5. Association of pan-Kcr expression with clinicopathological features in Jilin-cohort.
Supplementary Table S6. Association of pan-Kcr expression with clinicopathological features in Tianjin-cohort.
Supplementary Table S7. Univariate and multivariate analyses of DFS and OS in Tianjin cohort.
Supplementary Table S8. Univariate and multivariate analyses of DFS and OS in Jilin cohort.
Acknowledgments
This work was supported by the National Natural Science Foundation of China (82125026, 82330081, and 82203462), the Natural Science Foundation of Shandong Province (ZR2022QH016), and the “Intergovernmental International Cooperation on Science and Technology Innovation” Key Special Project of China (2024YFE0104100).
Footnotes
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
Authors’ Disclosures
No disclosures were reported.
Authors’ Contributions
L. Cao: Data curation, validation, project administration, writing–review and editing. H. Li: Data curation, formal analysis, writing–review and editing. F. Liu: Writing–review and editing. T. Zhao: Software, project administration. X. Shi: Investigation. H. Liu: Investigation. Y. Sun: Supervision. F. Wei: Resources. H. Ren: Resources, supervision, funding acquisition.
References
- 1. Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin 2024;74:12–49. [DOI] [PubMed] [Google Scholar]
- 2. Chiorean EG, Chiaro MD, Tempero MA, Malafa MP, Benson AB, Cardin DB, et al. Ampullary adenocarcinoma, version 1.2023, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw 2023;21:753–82. [DOI] [PubMed] [Google Scholar]
- 3. Oettle H, Neuhaus P, Hochhaus A, Hartmann JT, Gellert K, Ridwelski K, et al. Adjuvant chemotherapy with gemcitabine and long-term outcomes among patients with resected pancreatic cancer: the CONKO-001 randomized trial. JAMA 2013;310:1473–81. [DOI] [PubMed] [Google Scholar]
- 4. Uesaka K, Boku N, Fukutomi A, Okamura Y, Konishi M, Matsumoto I, et al. Adjuvant chemotherapy of S-1 versus gemcitabine for resected pancreatic cancer: a phase 3, open-label, randomised, non-inferiority trial (JASPAC 01). Lancet 2016;388:248–57. [DOI] [PubMed] [Google Scholar]
- 5. Maeda A, Boku N, Fukutomi A, Kondo S, Kinoshita T, Nagino M, et al. Randomized phase III trial of adjuvant chemotherapy with gemcitabine versus S-1 in patients with resected pancreatic cancer: Japan Adjuvant Study Group of Pancreatic Cancer (JASPAC-01). Jpn J Clin Oncol 2008;38:227–9. [DOI] [PubMed] [Google Scholar]
- 6. Neoptolemos JP, Moore MJ, Cox TF, Valle JW, Palmer DH, McDonald AC, et al. Effect of adjuvant chemotherapy with fluorouracil plus folinic acid or gemcitabine vs observation on survival in patients with resected periampullary adenocarcinoma: the ESPAC-3 periampullary cancer randomized trial. JAMA 2012;308:147–56. [DOI] [PubMed] [Google Scholar]
- 7. Chiorean EG, Von Hoff DD, Reni M, Arena FP, Infante JR, Bathini VG, et al. CA19-9 decrease at 8 weeks as a predictor of overall survival in a randomized phase III trial (MPACT) of weekly nab-paclitaxel plus gemcitabine versus gemcitabine alone in patients with metastatic pancreatic cancer. Ann Oncol 2016;27:654–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Osipov A, Nikolic O, Gertych A, Parker S, Hendifar A, Singh P, et al. The Molecular Twin artificial-intelligence platform integrates multi-omic data to predict outcomes for pancreatic adenocarcinoma patients. Nat Cancer 2024;5:299–314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Nicolle R, Gayet O, Duconseil P, Vanbrugghe C, Roques J, Bigonnet M, et al. A transcriptomic signature to predict adjuvant gemcitabine sensitivity in pancreatic adenocarcinoma. Ann Oncol 2021;32:250–60. [DOI] [PubMed] [Google Scholar]
- 10. Fraunhoffer N, Chanez B, Teyssedou C, Iovanna JL, Mitry E, Dusetti NJ, et al. ; Pdac Chemo Sensitivity Prediction Working Group . A transcriptomic-based tool to predict gemcitabine sensitivity in advanced pancreatic adenocarcinoma. Gastroenterology 2023;164:476–80.e4. [DOI] [PubMed] [Google Scholar]
- 11. Taube JM, Akturk G, Angelo M, Engle EL, Gnjatic S, Greenbaum S, et al. The Society for Immunotherapy of Cancer statement on best practices for multiplex immunohistochemistry (IHC) and immunofluorescence (IF) staining and validation. J Immunother Cancer 2020;8:e000155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Zhu G, Jin L, Sun W, Wang S, Liu N. Proteomics of post-translational modifications in colorectal cancer: discovery of new biomarkers. Biochim Biophys Acta Rev Cancer 2022;1877:188735. [DOI] [PubMed] [Google Scholar]
- 13. Uslu C, Kapan E, Lyakhovich A. Cancer resistance and metastasis are maintained through oxidative phosphorylation. Cancer Lett 2024;587:216705. [DOI] [PubMed] [Google Scholar]
- 14. Pinho SS, Reis CA. Glycosylation in cancer: mechanisms and clinical implications. Nat Rev Cancer 2015;15:540–55. [DOI] [PubMed] [Google Scholar]
- 15. Wang X, Li N, Zheng M, Yu Y, Zhang S. Acetylation and deacetylation of histone in adipocyte differentiation and the potential significance in cancer. Transl Oncol 2024;39:101815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Guccione E, Richard S. The regulation, functions and clinical relevance of arginine methylation. Nat Rev Mol Cell Biol 2019;20:642–57. [DOI] [PubMed] [Google Scholar]
- 17. Wang S, Mu G, Qiu B, Wang M, Yu Z, Wang W, et al. The function and related diseases of protein crotonylation. Int J Biol Sci 2021;17:3441–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Cao L, Wang F, Li S, Wang X, Huang D, Jiang R. PIM1 kinase promotes cell proliferation, metastasis and tumor growth of lung adenocarcinoma by potentiating the c-MET signaling pathway. Cancer Lett 2019;444:116–26. [DOI] [PubMed] [Google Scholar]
- 19. Collisson EA, Bailey P, Chang DK, Biankin AV. Molecular subtypes of pancreatic cancer. Nat Rev Gastroenterol Hepatol 2019;16:207–20. [DOI] [PubMed] [Google Scholar]
- 20. Bailey P, Chang DK, Nones K, Johns AL, Patch A-M, Gingras M-C, et al. Genomic analyses identify molecular subtypes of pancreatic cancer. Nature 2016;531:47–52. [DOI] [PubMed] [Google Scholar]
- 21. Barhli A, Cros J, Bartholin L, Neuzillet C. Prognostic stratification of resected pancreatic ductal adenocarcinoma: past, present, and future. Dig Liver Dis 2018;50:979–90. [DOI] [PubMed] [Google Scholar]
- 22. Nicolle R, Bachet J-B, Harlé A, Iovanna J, Hammel P, Rebours V, et al. Prediction of adjuvant gemcitabine sensitivity in resectable pancreatic adenocarcinoma using the GemPred RNA signature: an ancillary study of the PRODIGE-24/CCTG PA6 clinical trial. J Clin Oncol 2024;42:1067–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. O’Connell MJ, Lavery I, Yothers G, Paik S, Clark-Langone KM, Lopatin M, et al. Relationship between tumor gene expression and recurrence in four independent studies of patients with stage II/III colon cancer treated with surgery alone or surgery plus adjuvant fluorouracil plus leucovorin. J Clin Oncol 2010;28:3937–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Chen K, Ma Y, Zhong X, Lan J, Long D, Tian X, et al. Single-cell transcriptome profiling of primary tumors and paired organoids of pancreatobiliary cancer. Cancer Lett 2024;582:216586. [DOI] [PubMed] [Google Scholar]
- 25. Gu M, Liu Y, Xin P, Guo W, Zhao Z, Yang X, et al. Fundamental insights and molecular interactions in pancreatic cancer: pathways to therapeutic approaches. Cancer Lett 2024;588:216738. [DOI] [PubMed] [Google Scholar]
- 26. Jiang L, Qin J, Dai Y, Zhao S, Zhan Q, Cui P, et al. Prospective observational study on biomarkers of response in pancreatic ductal adenocarcinoma. Nat Med 2024;30:749–61. [DOI] [PubMed] [Google Scholar]
- 27. Tan M, Luo H, Lee S, Jin F, Yang JS, Montellier E, et al. Identification of 67 histone marks and histone lysine crotonylation as a new type of histone modification. Cell 2011;146:1016–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Huang H, Wang D-L, Zhao Y. Quantitative crotonylome analysis expands the roles of p300 in the regulation of lysine crotonylation pathway. Proteomics 2018;18:e1700230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Wan J, Liu H, Ming L. Lysine crotonylation is involved in hepatocellular carcinoma progression. Biomed Pharmacother 2019;111:976–82. [DOI] [PubMed] [Google Scholar]
- 30. Templeton AJ, McNamara MG, Šeruga B, Vera-Badillo FE, Aneja P, Ocaña A, et al. Prognostic role of neutrophil-to-lymphocyte ratio in solid tumors: a systematic review and meta-analysis. J Natl Cancer Inst 2014;106:dju124. [DOI] [PubMed] [Google Scholar]
- 31. Oettle H, Post S, Neuhaus P, Gellert K, Langrehr J, Ridwelski K, et al. Adjuvant chemotherapy with gemcitabine vs observation in patients undergoing curative-intent resection of pancreatic cancer: a randomized controlled trial. JAMA 2007;297:267–77. [DOI] [PubMed] [Google Scholar]
- 32. Neoptolemos JP, Dunn JA, Stocken DD, Almond J, Link K, Beger H, et al. Adjuvant chemoradiotherapy and chemotherapy in resectable pancreatic cancer: a randomised controlled trial. Lancet 2001;358:1576–85. [DOI] [PubMed] [Google Scholar]
- 33. Conroy T, Hammel P, Hebbar M, Ben Abdelghani M, Wei AC, Raoul J-L, et al. FOLFIRINOX or gemcitabine as adjuvant therapy for pancreatic cancer. N Engl J Med 2018;379:2395–406. [DOI] [PubMed] [Google Scholar]
- 34. Cui J, Fu Q, Chen X, Wang Y, Li Q, Wang F, et al. The real-world study of the clinical characteristics, diagnosis, and treatment of advanced pancreatic cancer in China. J Pancreatol 2023;7:1–9. [Google Scholar]
- 35. Wei W, Liu X, Chen J, Gao S, Lu L, Zhang H, et al. Class I histone deacetylases are major histone decrotonylases: evidence for critical and broad function of histone crotonylation in transcription. Cell Res 2017;27:898–915. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Fellows R, Denizot J, Stellato C, Cuomo A, Jain P, Stoyanova E, et al. Microbiota derived short chain fatty acids promote histone crotonylation in the colon through histone deacetylases. Nat Commun 2018;9:105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Tian D, Zeng X, Gong Y, Zheng Y, Zhang J, Wu Z. HDAC1 inhibits beige adipocyte-mediated thermogenesis through histone crotonylation of Pgc1a/Ucp1. Cell Signal 2023;111:110875. [DOI] [PubMed] [Google Scholar]
- 38. Wang F, Jin Y, Wang M, Luo H-Y, Fang W-J, Wang Y-N, et al. Combined anti-PD-1, HDAC inhibitor and anti-VEGF for MSS/pMMR colorectal cancer: a randomized phase 2 trial. Nat Med 2024;30:1035–43. [DOI] [PubMed] [Google Scholar]
- 39. Coiffier B, Pro B, Prince HM, Foss F, Sokol L, Greenwood M, et al. Results from a pivotal, open-label, phase II study of romidepsin in relapsed or refractory peripheral T-cell lymphoma after prior systemic therapy. J Clin Oncol 2012;30:631–6. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Figure S1. The expression of five protein post-translational modifications in pancreatic tissue of patients with DFS <3 months and DFS >24 months. The representative expression picture and their quantitative results of pan-crotonylation (A), pan-phosphorylation (B), pan-methylation (C), pan-ubiquitination (D), and pan-acetylation (E) in pancreatic tissue of patients with DFS <3 months and DFS >24 months. *P < 0.05, ns: nonsignificant.
Supplementary Figure S2. Prognostic significance of pan-Kcr expression in PDAC patients of all three cohorts. Kaplan-Meier estimates of DFS (A) and OS (B) for patients with high and low pan-Kcr expression based on all three cohorts.
Supplementary Figure S3. DFS and OS analysis according to the pan-Kcr based nomogram model. Bar chart of DFS (A) and OS (B) in low-risk and high-risk pancreatic cancer patients in the Tianjin cohort. Bar chart of DFS (C) and OS (D) in low-risk and high-risk pancreatic cancer patients in the Jilin cohort.
Supplementary Figure S4. Decision curve analysis in the training and validation cohorts. The nomograms and 8th edition AJCC-TNM staging system for predicting half-a-year, 1-year and 2-year DFS and OS of Qingdao (A, B), Tianjin (C, D), and Jilin (E, F) cohorts.
Supplementary Figure S5. ROC curve analysis of the nomogram model and the TNM staging system. ROC curve analysis of the nomograms and 8th edition AJCC-TNM staging system for predicting half-a-year, 1-year, 2-year DFS and OS in Qingdao (A, B), Tianjin (C, D) and Jilin (E, F) cohorts.
Supplementary Figure S6. ROC curve analysis of the full nomogram comparing LN metastasis for predicting DFS and OS, as well as LN metastasis + TNM stage for predicting OS. ROC curve analysis of the nomogram comparing lymph node (LN) metastasis for predicting half-year, 1-year, and 2-year disease-free survival (DFS) and overall survival (OS), as well as comparing LN metastasis + TNM stage for predicting half-year, 1-year, and 2-year OS in the Qingdao (A-C), Tianjin (D-F), and Jilin cohorts (G-I).
Supplementary Table S1. Clinicopathological Characteristics of Qingdao-cohort.
Supplementary Table S2. Clinicopathological Characteristics of Tianjin-cohort.
Supplementary Table S3. Clinicopathological Characteristics of Jilin-cohort.
Supplementary Table S4. Association of pan-Kcr expression with clinicopathological features in Qingdao-cohort.
Supplementary Table S5. Association of pan-Kcr expression with clinicopathological features in Jilin-cohort.
Supplementary Table S6. Association of pan-Kcr expression with clinicopathological features in Tianjin-cohort.
Supplementary Table S7. Univariate and multivariate analyses of DFS and OS in Tianjin cohort.
Supplementary Table S8. Univariate and multivariate analyses of DFS and OS in Jilin cohort.
Data Availability Statement
The data generated in this study are not publicly available, as they contain information that could compromise patient confidentiality, but they are available upon reasonable request to the corresponding author.