Development and Validation of Web Calculators to Predict Early Recurrence and Long-Term Survival in Patients with Duodenal Papilla Carcinoma after Pancreaticoduodenectomy

doi:10.21203/rs.3.rs-2453629/v1

Download PDF

Research Article

Development and Validation of Web Calculators to Predict Early Recurrence and Long-Term Survival in Patients with Duodenal Papilla Carcinoma after Pancreaticoduodenectomy

https://doi.org/10.21203/rs.3.rs-2453629/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: Duodenal papilla carcinoma(DPC) is prone to relapse even after radical pancreaticoduodenectomy (PD) (including robotic, laparoscopic and open approach).This study aimed to develop web calculators to predict early recurrence(ER) (within two years after surgery) and long-term survival in patients with DPC after PD.

Methods: Patients with DPC after radical PD were included. Univariate and multivariate logistic regression analyses were used to identify independent risk factors.Two web calculators were developed based on independent risk factors in the training cohort and then tested in the validation cohort.

Results: Of the 251 patients who met the inclusion criteria, 180 and 71 patients were enrolled in the training and validation cohorts, respectively. Multivariate logistic regression analysis revealed that tumor size [Odds Ratio (OR) 1.386; 95% confidence interval (CI) 1070-1.797; P= 0.014]; number of lymph node metastasis (OR 2.535; 95% CI 1.114-5.769; P= 0.027), perineural invasion (OR 3.078; 95% CI 1.147-8.257; P = 0.026), and tumor differentiation (OR 3.552; 95% CI 1.132-11.152; P = 0.030) were independent risk factors for ER. Nomogram based on the above four factors achieved good C-statistics of 0.759 and 0.729 in predicting ER in the training and the validation cohorts, respectively. Time-dependent ROC analysis (timeROC) and decision curve analysis (DCA) revealed that the nomogram provided superior diagnostic capacity and net benefit compared with single variable.

Conclusions:This study developed and validated two web calculators that can predict ER and long-term survival in patients with DPC with high degree of stability and accuracy.

Duodenal papilla carcinoma

Nomogram

Web calculator

Early recurrence

Long-term survival

Duodenal papilla carcinoma (DPC) is a malignant tumor occurring in the duodenal papilla region (including the intrapapillary bile duct and pancreatic duct) and accounts for about 60% of primary duodenal malignant tumors [1, 2]. Compared with pancreatic ductal adenocarcinoma (PDAC) and distal cholangiocarcinoma (DCC), DPC has a higher resection rate and a better long-term prognosis [3].

Radical pancreaticoduodenectomy (PD) represents the main surgical strategy for the treatment of DPC, with a 5-year survival rate of about 50% [3]. However, due to the difference in tumor histological grade, tumor size and surrounding tissue invasion, some patients with DPC are still prone to relapse after surgery, which seriously threatens the long-term survival of these patients [3, 4]. Multiple studies have shown that tumor size, lymph node metastasis, tumor differentiation, perineural invasion and TNM stage are independent prognostic factors affecting the long-term survival of patients with DPC [5–7]. However, there is no prediction model that combines these independent prognostic factors to predict early recurrence (ER) and overall survival (OS) of patients with DPC, so as to provide objective evaluation indicators and clinical references [8, 9].

Therefore, in the present study, we retrospectively analyzed clinical and follow-up data from patients with DPC at our center to screen for independent prognostic factors affecting ER and long-term survival. In addition, the web-based calculators for ER and long-term survival in DPC patients were developed following the predictive models to help clinicians screen high-risk patients with poor prognosis and timely adjust corresponding treatment and follow-up strategies to improve the long-term survival outcomes in these patients.

Study population

Clinical pathology results and follow-up data were included for patients with DPC who underwent radical PD at the Department of Liver Transplantation and Hepatobiliary Surgery of Shandong Provincial Hospital between January 2011 and October 2020. This study was approved by the Medical Ethics Committee of Shandong Provincial Hospital (No.2022 − 178), and all patients gave informed consent and signed written informed consent.

Inclusion and exclusion criteria

The pre-operative resectability of the DPC was evaluated under the multi-disciplinary team (MDT) strategy, and the diagnosis was confirmed based on the post-operative histopathology results. The inclusion criteria were as follows: (1) age ≥ 18 years; (2) DPC was confirmed by pathological report; (3) radical PD was performed and R0 resection was confirmed by postoperative pathology; (4) the American Society of Anesthesiologists (ASA) staging was I-III; (5) no neoadjuvant therapy (NAT) was received before operation. Exclusion criteria were as follows: (1) death due to complications or other causes other than DPC; (2) incomplete data or lost to follow-up; (3) accompanied by other malignant tumors; (4) the presence of distant metastasis.

Follow up after resection

Patients were followed up every three to six months after surgery through outpatient reviews or telephone interviews. The results of follow-up were recorded until the patient died or was lost to follow-up. Examination during follow-up included enhanced abdominal and pelvic CT and/or MRI scans, chest X-rays, and tumor marker screening, among others. Tumor recurrence was determined comprehensively based on imaging and serological findings. The primary endpoint of this study was early recurrence (ER), and the secondary endpoint was overall survival (OS). ER was defined as tumor recurrence within 2 years after radical resection (In the result section of this article, we use the minimum P-value method to reverse verify this definition [10, 11]). OS was calculated from the date of PD to either the date of death or the date of the last follow-up [12]. The final follow-up date for this study was October 31, 2022.

Study variables and definition

All patients underwent standard open or minimally invasive approach (robotic or laparoscopic). The specific surgical procedures and protocols have been reported in detail in our previous studies [13, 14]. Surgical modality was not a focus in this study, as a number of domestic and international studies have demonstrated that a minimally invasive or open approach is not a prognostic factor for long-term survival of patients after surgery [15–18].

Conventional demographic indicators included age, sex, body mass index (BMI), ASA health staging, and personal history. Serological variables include: blood routine, biochemical indicators and tumor markers, among others. Serological indicators were based on the results of the first post-admission examination. Histopathologic findings included tumor size (the largest tumor diameter recorded in the pathological reports), tumor differentiation, surrounding invasion, and lymph node metastasis. The Clavien-Dindo grading system assesses postoperative complications, and severe complications were defined as Clavien-Dindo grade III or higher [19]. R0 resection was defined as the absence of tumor cells at the microscopic margin [20].

Statistical analysis

Continuous variables conforming to normal distribution were expressed as mean ± standard deviation (SD), and differences between groups were compared by student’s t test; while continuous variables that did not conform to normal distribution were expressed as median (interquartile range, [IQR]), and differences between groups were compared by Mann-Whitney U test [21]. Categorical variables were expressed as frequency (%), and differences between groups were compared by Pearson chi-square test, continuity-corrected chi-square test or Fisher’s exact probability test. Refer to previous research [22], in the training cohort, univariate and multivariate logistic regression analysis were used to determine the independent prognostic factors for ER. The DPC-ER nomogram was developed according to the proportion of regression coefficients of independent prognostic factors. The calibration curve, receiver operating characteristic (ROC) curve, time-dependent ROC (timeROC) curve, and decision curve analysis (DCA) were used to evaluate the performances of different models across the two cohorts. The optimal cut-off value for the nomogram score was calculated using the maximum Youden index method. The Kaplan-Meier survival curve was used to describe the long-term survival of patients, and the log-rank test was used to compare the differences between groups. All tests were two-tailed, and P < 0.05 was defined as statistically significant. The statistical software we used in the present study includes R software (version 3.6.3, R Foundation for Statistical Computing, Vienna, Austria) and SPSS software (version 25.0, IBM, Armonk, New York, USA). The packages used in the R environment mainly include: “survival”,“survminer”,“rmda”,“nomogramFormula”,“timeROC”,“rms”,“pROC”, “DynNom” and “rsconnect”.

Clinical characteristics of patients

A total of 251 patients met the inclusion and exclusion criteria to be included in the study. All patients were randomized in a ratio of approximately 7:3, with 180 patients enrolled in the training cohort and 71 in the validation cohort (Supplementary Fig. 1). During the follow-up period, 81 (35.2%) patients experienced ER, including 56 (31.1%) patients in the training cohort and 25 (35.2%) patients in the validation cohort (P = 0.531). The median RFS of patients in the training cohort was 36.5 (95% confidence interval [CI]: 27.2–45.8) months, and the median OS was 50.2 (95%CI: 41.1–59.3) months. The median RFS in the validation cohort was 27.6 (95%CI: 22.9–32.3) months, and the median OS was 50.5 (95%CI: 25.0–76.0) months. There were no significant differences between the two cohorts in demographic, serological, tumor histopathology characteristics, or long-term survival outcomes. The detailed characteristics of patients with DPC in the training and validation cohorts were presented in Table 1.

Table 1

Clinicopathological characteristics of patients with DPC in the training and validation cohorts (n = 251)
Variables	Training cohort	Validation cohort	P value
Variables	(n = 180)	(n = 71)	P value
Age ≤ 60 years	101 (56.1)	34 (47.9)	0.239
Gender, Male	97 (53.9)	42 (59.2)	0.450
BMI, kg/m²	23.2 (21.1–25.7)	23.8 (21.5–25.9)	0.433
ASA grade ≤ II	154 (85.6)	56 (78.9)	0.197
Biliary infection	46 (25.6)	16 (22.5)	0.617
Biliary drainage	46 (25.6)	20 (28.2)	0.672
Jaundice history	99 (55.0)	37 (52.1)	0.679
Diabetes history	26 (14.4)	8 (11.3)	0.508
Preoperative WBC, 10⁹/L	6.1 (4.8–7.4)	5.9 (4.7–7.4)	0.539
Preoperative ALB, g/L	37.9 (35.4–40.5)	37.8 (35.1–40.8)	0.817
Preoperative TBIL, µmol/L	23.7 (11.9–97.4)	32.4 (9.7-115.5)	0.729
Preoperative glucose, mmol/L	5.4 (4.8-6.0)	5.1 (4.6–6.2)	0.346
Preoperative CA19-9, U/mL	39.9 (11.0-106.8)	29.6 (13.0-76.7)	0.782
Preoperative CA125, U/mL	11.2 (7.9–17.7)	11.7 (8.8–17.8)	0.360
Preoperative CEA, ng/mL	2.2 (1.5–3.3)	2.1 (1.2–3.5)	0.544
Preoperative FAR	0.107 (0.085–0.130)	0.097 (0.083–0.136)	0.624
Preoperative NLR	2.3 (1.7–3.6)	2.1 (1.5–3.3)	0.142
Preoperative PLR	162.0 (124.0-234.3)	151.2 (122.3-204.3)	0.357
Tumor size, cm	2.0 (1.5-3.0)	2.5 (1.8–3.5)	0.211
Without lymph node metastasis	135 (75.0)	53 (74.6)	0.681
Perineural invasion	24 (13.3)	15 (21.1)	0.125
Vascular invasion	7 (3.9)	1 (1.4)	0.543
Poorly differentiated	47 (26.1)	18 (25.4)	0.893
Clavien-Dindo grade ≥ III	23 (12.8)	13 (18.3)	0.260
Early recurrence	56 (31.1)	25 (35.2)	0.531
Adjuvant chemotherapy	87 (48.3)	38 (53.5)	0.459
Follow-up, months	37.5 (18.8–54.5)	33.8 (14.7–47.8)	0.051
Median RFS (95% CI), months	36.5 (27.2–45.8)	27.6 (22.9–32.3)	0.182
Median OS (95% CI), months	50.2 (41.1–59.3)	50.5 (25.0–76.0)	0.288

Patients with ER had a higher proportion of lymph node metastasis (39.5% vs. 18.2%, P = 0.001), peripheral nerve invasion (28.4% vs. 9.4%, P < 0.001) and poorly differentiated tumors (38.3% vs. 20.0, P = 0.001). Additionally, tumors were relatively larger in the ER group (2.5 cm vs. 2.0 cm, P = 0.003). The median RFS of patients in the ER group was 11.4 (95%CI: 10.1–12.7) months, and the median OS was 21.9 (95%CI: 18.2–25.6) months, both significantly shorter than those in the non-ER group (both P values < 0.001). Additional demographic, serological and histopathological results for both groups are detailed in Table 2.

Table 2

Clinicopathological characteristics and surgical outcomes between patients in DPC-ER group and Non-DPC-ER group
Variables	Non-DPC-ER group	DPC-ER group	P value
Variables	(n = 170)	(n = 81)	P value
Age ≤ 60 years	92 (54.1)	43 (53.1)	0.878
Gender, Male	97 (57.1)	42 (51.9)	0.438
BMI, kg/m²	23.8 (21.4–26.1)	22.7 (20.8–24.6)	0.023
ASA grade ≤ II	144 (84.7)	66 (81.5)	0.518
Biliary infection	41 (24.1)	21 (25.9)	0.756
Biliary drainage	43 (25.3)	23 (28.4)	0.602
Jaundice history	90 (52.9)	46 (56.8)	0.567
Diabetes history	23 (13.5)	11 (13.6)	0.991
Preoperative WBC, 10^9/L	6.2 (4.8–7.4)	5.9 (4.7–7.1)	0.519
Preoperative ALB, g/L	37.9 (35.6–40.4)	37.5 (34.5–40.8)	0.728
Preoperative TBIL, µmol/L	24.0 (11.9–99.8)	31.5 (9.7-104.8)	0.752
Preoperative glucose, mmol/L	5.3 (4.7-6.0)	5.2 (4.8–6.2)	0.733
Preoperative CA19-9, U/mL	31.8 (10.8–82.1)	46.9 (15.2-133.2)	0.048
Preoperative CA125, U/mL	11.0 (8.1–17.7)	12.1 (9.0-18.2)	0.267
Preoperative CEA, ng/mL	2.2 (1.4–3.4)	2.0 (1.3–3.1)	0.535
Preoperative FAR	0.106 (0.084–0.127)	0.108 (0.086–0.139)	0.363
Preoperative NLR	2.3 (1.6–3.6)	2.4 (1.6–3.4)	0.983
Preoperative PLR	154.8 (116.4-218.3)	162.3 (131.6–230.0)	0.258
Tumor size, cm	2.0 (1.5-3.0)	2.5 (1.8-4.0)	0.003
Lymph node metastasis	31 (18.2)	32 (39.5)	0.001
Perineural invasion	16 (9.4)	23 (28.4)	< 0.001
Vascular invasion	5 (2.9)	3 (3.7)	0.748
Poorly differentiated	34 (20.0)	31 (38.3)	0.001
Clavien-Dindo grade ≥ III	26 (15.3)	10 (12.3)	0.533
Adjuvant chemotherapy	85 (50.0)	40 (49.4)	0.927
Median RFS (95% CI), months	62.5 (NA-NA)	11.4 (10.1–12.7)	< 0.001
Median OS (95% CI), months	NA (NA-NA)	21.9 (18.2–25.6)	< 0.001

Validation of the definition of DPC early recurrence

As mentioned above, we used the minimum P-value method to conduct a minimum P-value analysis of postoperative OS in the potential ER group and the non-ER group following previous studies [10, 11]. The results indicated that the P-value of log-rank test was the smallest when the cut-off value was 24 months (P = 3.596×10^− 44) (Supplementary Table 1).

Identification of independent prognostic factors for ER and the development of a prognostic nomogram in the training cohort

Multivariate logistic regression analysis revealed that tumor size (odds ratio [OR]: 1.386; 95% CI: 1.070–1.797; P = 0.014), number of lymph node metastasis (OR: 2.535; 95% CI: 1.114–5.769; P = 0.027), perineural invasion (OR: 3.078; 95% CI: 1.147–8.257; P = 0.026), poorly tumor differentiation (OR: 3.552; 95% CI: 1.132–11.152; P = 0.030) were the independent prognostic factors for ER of DPC after radical resection (Table 3). Then, we constructed the prognostic nomogram for predicting ER of DPC based on the above four independent prognostic factors (Fig. 1A).

Table 3

Univariable and multivariable logistic analysis for ER in the training cohort
Characteristics	Univariable analysis			Multivariable analysis
Characteristics	B	OR (95% CI)	P value	B	OR (95% CI)	P value
Age, > 60 vs ≤ 60, years	0.044	1.045 (0.554–1.973)	0.891
Gender, male vs female	-0.435	0.647 (0.343–1.220)	0.178
BMI, per kg/m²	-0.058	0.943 (0.862–1.032)	0.205
ASA grade, ≥III vs ≤ II	0.187	1.205 (0.501–2.899)	0.677
Biliary infection, yes vs no	0.093	1.098 (0.535–2.250)	0.799
Biliary drainage, yes vs no	-0.182	0.834 (0.399–1.742)	0.629
Jaundice, yes vs no	0.126	1.134 (0.600-2.143)	0.698
Diabetes history, yes vs no	-0.236	0.789 (0.311–2.002)	0.619
Preoperative WBC, per 10^9/L	0.047	1.048 (0.900–1.220)	0.546
Preoperative ALB, per g/L	-0.025	0.975 (0.898–1.059)	0.552
Preoperative TBIL, per µmol/L	0.000	1.000 (0.996–1.004)	0.983
Preoperative glucose, per mmol/L	-0.056	0.945 (0.769–1.162)	0.593
Preoperative CA19-9, per U/mL	0.002	1.002 (1.000-1.003)	0.013	0.001	1.001 (0.999–1.002)	0.229
Preoperative CA125, per U/mL	0.009	1.009 (0.983–1.037)	0.495
Preoperative CEA, per ng/mL	-0.002	0.998 (0.981–1.015)	0.792
Preoperative FAR*100	0.054	1.056 (0.973–1.146)	0.195
Preoperative NLR	0.003	1.003 (0.943–1.067)	0.927
Preoperative PLR	0.001	1.001 (0.999–1.003)	0.393
Tumor size, per cm	0.326	1.385 (1.097–1.749)	0.006	0.327	1.386 (1.070–1.797)	0.014
Number of lymph node metastasis
0–3 vs 0	1.077	2.936 (1.400-6.157)	0.004	0.930	2.535 (1.114–5.769)	0.027
> 3 vs 0	1.822	6.182 (1.083–35.297)	0.040	1.754	5.780 (0.848–39.389)	0.073
Perineural invasion, yes vs no	1.335	3.800 (1.568–9.210)	0.003	1.124	3.078 (1.147–8.257)	0.026
Vascular invasion, yes vs no	0.530	1.698 (0.367–7.854)	0.498
Differentiation
Moderate vs Well	0.972	2.643 (0.999–6.995)	0.050	0.492	1.635 (0.572–4.677)	0.359
Poor vs Well	1.777	5.910 (2.099–16.636)	0.001	1.268	3.552 (1.132–11.152)	0.030
Clavien-Dindo grade, ≥III vs ≤ II	-0.281	0.755 (0.281–2.031)	0.578
Adjuvant chemotherapy, yes vs no	-0.111	0.895 (0.476–1.684)	0.731
Bold text hinted that these variables were statistically significant in univariate or multivariate analyses. ¶, Considering the collinearity between the tumor size and T stage, only the tumor size was included in the multivariable analysis.

Performance of the DPC-ER nomogram for predicting ER compared to a single variable in training and validation cohorts.

As shown in Fig. 1B-1C, calibration curves showed that the DPC-ER nomogram fitted well both in the training cohort and the validation cohort (the P values of Hosmer-Lemeshow test were 0.242 and 0.210 in the training cohort and the validation cohort, respectively). The C-statistics or the area under the ROC (AUC) curve of the DPC-ER nomogram for ER prediction was 0.759 (95% CI, 0.685–0.832) in the training cohort and 0.729 (95% CI, 0.601–0.856) in the validation cohort, which were both significantly superior to a single variable (all P < 0.05, Fig. 2A-2B, Supplementary Table 2).

Decision curve analysis (DCA) converts complex mathematical models into simple and easy-to-understand graphics for display, in order to intuitively judge the practicability and net benefits of different models [23]. DCA demonstrated that the DPC-ER nomogram provided superior net benefits when compared with a single variable (Fig. 2C-2D in the training and validation cohorts, respectively).

The discriminatory ability of the nomogram was further evaluated by dividing patients into two risk groups according to different nomogram scores (the low-risk group with a nomogram score ≤ 63; and the high-risk group with a nomogram score > 63) (Fig. 1A). The results revealed that the high-risk group had a higher ER incidence than the low-risk group in the two cohorts: high-risk group vs. low-risk group, 48.9% vs. 12.5% in the training cohort, P < 0.001; 47.2% vs. 22.9% in the validation cohort, P = 0.032 (Fig. 2E, Supplementary Table 3).

The long-term survival outcomes of patients in different risk groups

As shown in Fig. 3A-3D, there were significant differences in long-term survival outcomes between different risk groups. The 1-, 3-, 5-year OS rates of patients in the low-risk group were 100.0%, 87.6%, 66.0% in the training cohorts, and 100.0%, 68.6%, 40.0% in the validation cohort, which were all significantly higher than those in the high-risk group: 90.0%, 50.7%, 21.6% in the training cohort, and 93.9%, 46.4%, 30.9% in the validation cohort (P < 0.001) (Fig. 3A-3B, Supplementary Table 4). Similar trends in RFS rates were observed in both groups. The detailed results can be seen in Fig. 3C-3D and Supplementary Table 5.

Performance of the DPC-OS nomogram for predicting OS compared to a single variable in training and validation cohorts.

Subsequently, we developed the DPC-OS nomogram for predicting long-term survival outcomes of DPC patients based on the above four independent prognostic factors (Supplementary Fig. 2A). Time-dependent ROC curve (timeROC) analysis was used to further demonstrate the performance of DPC-OS nomogram and single variable in predicting long-term survival outcomes. The area under the timeROC (timeAUCs) of the DPC-OS nomogram for predicting OS within 5 years in the training and validation cohorts were 0.666–0.848 and 0.638–0.953, respectively, which were significantly superior to those of the other four variables (all P < 0.001). More details of the timeAUCs of the DPC-OS nomogram and other variables within 5 years are shown in Supplementary Table 6 and Supplementary Fig. 2B-2C.

Development of the web-based online calculators in predicting ER and OS

To facilitate clinical application, we further converted the DPC-ER nomogram and DPC-OS nomogram into web calculators (Supplementary Fig. 3- Supplementary Fiure 4). The web calculators can be accessed at http://114.115.144.103/dpc and https://abc123-456.shinyapps.io/AAC-OS/ to predict the ER and OS in patients with DPC after PD. The prediction probability can be easily determined by inputting clinical features and reading the output results generated by the webserver.

The duodenal papilla is located at the opening where the pancreatic-bile ducts merge into the duodenum [1, 2]. Tumors growing in this region can present with progressive and painless jaundice at an early stage, offering the possibility of early diagnosis [3, 24]. Compared to distal cholangiocarcinoma and pancreatic head adenocarcinoma, DPC has a higher resection rate and a better long-term prognosis, with a five-year survival rate of up to 50% [3]. PD is by far the preferred strategy for radical resection of DPC, however, even after radical resection, some patients are prone to relapse. In this study, we integrated four independent prognostic factors for DPC to develop and validate a series of nomograms to predict ER and long-term survival for DPC after PD. Both the DPC-ER and DPC-OS nomogram showed higher predictive accuracy and net benefit than single variable (including tumor size, perineural invasion, lymph node metastasis, and tumor differentiation status).

Early tumor recurrence is a common and fatal condition in various malignancies, including DPC, and often indicates a poor prognosis [2, 25, 26]. Although DPC has a better long-term prognosis compared with pancreatic head adenocarcinoma and distal cholangiocarcinoma, 30% of patients will relapse within 2 years after surgery [3]. Thus, screening for independent prognostic factors associated with tumor recurrence may help guide further treatment of these patients and improve their overall prognosis [12]. Previous studies have shown that tumor size, lymph node metastasis, tumor differentiation status, and TNM stage are important prognostic factors affecting the recurrence and long-term survival of DPC after resection [3]. Consistent with previous studies, in the present study, multivariate logistic regression analysis revealed that tumor size, peripheral nerve invasion, lymph node metastasis, and tumor differentiation status were also independent prognostic factors for ER of DPC.

Yoen et al. [27] analyzed preoperative imaging findings of ampullary or papilla carcinoma and found that tumor size was an independent prognostic factor affecting OS and disease-free survival (DFS). Park et al. [28] also indicated that tumor recurrence was significantly affected by tumor diameter. Larger tumor diameters often indicate that the tumor was detected later and has a broader invasion area [27, 28]. Several studies have shown that lymph node metastasis and perineural invasion are the major factors affecting the recurrence and long-term survival of DPC after surgery. de Castro et al. [29] demonstrated that the long-term prognosis of patients was negatively correlated with the number of lymph node metastases. Sakata et al. [30] also found that the number, not the location, of positive regional lymph nodes independently affects long-term survival after resection in patients with ampullary carcinoma. These findings suggest that standardized lymph node dissection may be important in improving long-term outcomes in these patients. Although the incidence of perineural invasion in DPC is lower, perineural invasion often indicates tumor progression and a poorer prognosis [31]. Junrungsee et al. [24] retrospectively analyzed the clinicopathological data of 72 patients with carcinoma of the ampulla of vater (CAV) treated by PD and found that tumor differentiation status was also an important prognostic factor affecting the long-term survival of these patients. Tumor differentiation status is also an important prognostic factor for ER and long-term survival in patients with DPC. Poor tumor differentiation often indicates strong invasion ability and early metastasis [24, 29, 31].

In the present study, we not only demonstrated the value of these four factors in predicting long-term survival outcomes for DPC, but for the first time integrated these four independent prognostic factors to develop the DPC-ER and DPC-OS models for predicting ER and OS for DPC following radical resection. The results indicated that the prediction models exhibited satisfactory prediction performance in both the training and validation cohorts. The DPC nomogram models were more accurate in predicting ER and long-term survival compared to a single variable. In addition, patients were further divided into different risk subgroups according to the DPC-ER nomogram score (high-risk group: nomogram score > 63; low-risk group: nomogram score ≤ 63). The results showed that patients in the high-risk group had significantly higher ER rates than those in the low-risk group, and that patients in the low-risk group had significantly better RFS and OS than those in the high-risk group. Thus, with the DPC predictive models, we can screen high-risk patients for tumor recurrence at an early stage and recommend them for closer post-operative follow-up and timely adjuvant therapy to prolong survival. In addition, we further transform the series nomogram into an online calculator so that it can be used in real-time to provide predictive capabilities in computers or mobile terminals, thus generating a high level of clinical utility [32].

Although the DPC prediction models constructed in this study can effectively screen ER patients and accurately predict their long-term survival after radical resection, several limitations need to be clarified. First, this study is a retrospective study conducted from a single center, and inherent biases are inevitable. In the future, multicenter and prospective studies are needed to validate the predictive power of the web-based calculators. Second, the sample size of this study is relatively small and we will further collect relevant patients in the future to provide more convincing validation results. Third, pathology classification of DPC was not performed in this study due to constraints from existing pathology diagnosis results. Finally, given the fact that the specific adjuvant therapy and outcomes were unknown, further exploration of the clinical efficacy of adjuvant therapy and different adjuvant therapies in patients in high-risk and low-risk subgroups is warranted in the future.

In conclusion, this study integrated four independent prognostic factors, including tumor size, lymph node metastasis, tumor differentiation status, and perineural invasion, to develop the DPC web calculators for predicting ER and long-term survival of DPC after radical resection. With its high predictive accuracy and net benefit, the web-based calculators can screen high-risk patients prone to relapses, which is expected to help clinicians make individualized clinical decisions and improve overall survival outcomes for these patients.

DPC

duodenal papilla carcinoma

PDAC

pancreatic ductal adenocarcinoma

DCC

distal cholangiocarcinoma

pancreaticoduodenectomy

MDT

multi-disciplinary team

BMI

body mass index

ASA

American Society of Anesthesiologists

early recurrence

overall survival

RFS

recurrence-free survival

standard deviation

IQR

interquartile range

odds ratio

confidence interval

ROC

receiver operating characteristic curve

AUC

area under the ROC curve

DCA

decision curve analysis

timeROC

time-dependent ROC analysis

CAV

carcinoma of the ampulla of vater.

Ethics approval and consent to participate: This study was approved by the Medical Ethics Committee of Shandong Provincial Hospital (No.2022-178). Written informed consent was obtained from all patients.

Consent for publication: Not applicable.

Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests: The authors declare that they have no competing interests.

Funding: This study was supported by National Natural Science Foundation of China (No. 81373172, 81770646).

Authors Contributions:

Guangsheng Yu: project development, data collection, manuscript writing;

Zile He: project development, data analysis, manuscript writing;

Shuai Xu: data analysis & collection, manuscript writing;

Junjie Kong: data collection, manuscript editing;

Jingyi He: data collection, manuscript editing;

Jun Liu: project development, manuscript editing, Financial support.

Acknowledgements: We would like to thank all the authors for their hard work on this article.

Myung-Hwan, Kim. Seo, and, Sun: Tumors of the major duodenal papilla. Gastrointest Endosc. 2001;54(5):609–20.
Lu Y, Li W, Liu G, Yang Y, Xiao E, Mu S, Guo Y, Li D, Yan G. Identification of critical pathways and potential therapeutic targets in poorly differentiated duodenal papilla adenocarcinoma. Cancer Cell Int. 2021;21(1):9.
Lian P-L, Chang Y, Xu X-C, Zhao Z, Wang X-Q, Xu K-S. Pancreaticoduodenectomy for duodenal papilla carcinoma: A single-centre 9-year retrospective study of 112 patients with long-term follow-up. World J Gastroentero. 2017;23(30):5579–88.
Zhao R, Chang Y, Wang X, Zhang P, Zhang C, Lian P. Pylorus-preserving pancreaticoduodenectomy versus standard pancreaticoduodenectomy in the treatment of duodenal papilla carcinoma. Oncol Lett. 2018;15(5):6368–76.
Li H-B, Zhao F-Q, Zhou J. Prognostic Nomogram for Disease-Specific Survival in Patients with Non-metastatic Ampullary Carcinoma After Surgery. Ann Surg Oncol. 2019;26(4):1079–85.
Moekotte AL, Malleo G, van Roessel S, Bonds M, Halimi A, Zarantonello L, Napoli N, Dreyer SB, Wellner UF, Bolm L, et al. Gemcitabine-based adjuvant chemotherapy in subtypes of ampullary adenocarcinoma: international propensity score-matched cohort study. Br J Surg. 2020;107(9):1171–82.
Moekotte AL, Lof S, Van Roessel S, Fontana M, Dreyer S, Shablak A, Casciani F, Mavroeidis VK, Robinson S, Khalil K, et al. Histopathologic Predictors of Survival and Recurrence in Resected Ampullary Adenocarcinoma: International Multicenter Cohort Study. Ann Surg. 2020;272(6):1086–93.
Moekotte A, van Roessel S, Malleo G, Rajak R, Ecker B, Fontana M, Han H, Rabie M, Roberts K, Khalil K, et al. Development and external validation of a prediction model for survival in patients with resected ampullary adenocarcinoma. Eur J Surg oncology: J Eur Soc Surg Oncol Br Association Surg Oncol. 2020;46(9):1717–26.
Huang X, Huang C, Chen W, Cai J, Gan T, Zhao Y, Liu Q, Liang L, Yin X. Development and validation of a nomogram for predicting overall survival of node-negative ampullary carcinoma. J Surg Oncol. 2020;121(3):518–23.
Sahara K, Tsilimigras D, Kikuchi Y, Ethun C, Maithel S, Abbott D, Poultsides G, Hatzaras I, Fields R, Weiss M, et al. Defining and Predicting Early Recurrence after Resection for Gallbladder Cancer. Ann Surg Oncol. 2021;28(1):417–25.
Groot VP, Gemenetzis G, Blair AB, Rivero-Soto RJ, Yu J, Javed AA, Burkhart RA, Rinkes IHMB, Molenaar IQ, Cameron JL, et al. Defining and Predicting Early Recurrence in 957 Patients With Resected Pancreatic Ductal Adenocarcinoma. Ann Surg. 2019;269(6):1154–62.
Xu S, Zhang X, Zhao G, Zou W, Zhao Z, Hu M, Gao Y, Tan X, Liu Q, Liu R. A novel online calculator to predict recurrence risk in patients with distal cholangiocarcinoma after radical pancreaticoduodenectomy.J Surg Oncol2021.
Wang M, Li D, Chen R, Huang X, Li J, Liu Y, Liu J, Cheng W, Chen X, Zhao W, et al. Laparoscopic versus open pancreatoduodenectomy for pancreatic or periampullary tumours: a multicentre, open-label, randomised controlled trial. Lancet Gastroenterol Hepatol. 2021;6(6):438–47.
Li G, Yuan L, Yu G, Xu Y, Liu J. A Modified Suture Technique in Hepaticojejunostomy. Med Sci Monitor: Int Med J Experimental Clin Res. 2020;26:e921896.
Liu Q, Zhao Z, Zhang X, Wang W, Han B, Chen X, Tan X, Xu S, Zhao G, Gao Y et al. Perioperative and Oncological Outcomes of Robotic Versus Open Pancreaticoduodenectomy in Low-Risk Surgical Candidates: A Multicenter Propensity Score-Matched Study. Annals of surgery 2021.
Jin J, Yin S-M, Weng Y, Chen M, Shi Y, Ying X, Gemenetzis G, Qin K, Zhang J, Deng X et al. Robotic versus open pancreaticoduodenectomy with vascular resection for pancreatic ductal adenocarcinoma: surgical and oncological outcomes from pilot experience.Langenbeck's archives of surgery2022.
Wang W, Liu Q, Zhao Z-M, Tan X-L, Wang Z-Z, Zhang K-D, Liu R. Comparison of robotic and open pancreaticoduodenectomy for primary nonampullary duodenal adenocarcinoma: a retrospective cohort study. Langenbeck's archives of surgery. 2022;407(1):167–73.
Zhang Z, Yin T, Qin T, Pan S, Wang M, Zhang H, Qin R. Comparison of laparoscopic versus open pancreaticoduodenectomy in patients with resectable pancreatic ductal adenocarcinoma: A propensity score-matching analysis of long-term survival. Pancreatology: official journal of the International Association of Pancreatology (IAP) [et al]. 2022;22(2):317–24.
Clavien PA, Barkun J, de Oliveira ML, Vauthey JN, Dindo D, Schulick RD, de Santibañes E, Pekolj J, Slankamenac K, Bassi C, et al. The Clavien-Dindo classification of surgical complications: five-year experience. Ann Surg. 2009;250(2):187–96.
de Castro SMM, Kuhlmann KFD, van Heek NT, Busch ORC, Offerhaus GJ, van Gulik TM, Obertop H, Gouma DJ. Recurrent disease after microscopically radical (R0) resection of periampullary adenocarcinoma in patients without adjuvant therapy.Journal of gastrointestinal surgery: official journal of the Society for Surgery of the Alimentary Tract2004, 8(7).
Xu S, Zhang X-P, Zhao G-D, Zou W-B, Zhao Z-M, Hu M-G, Gao Y-X, Tan X-L, Liu Q, Liu R. Robotic versus open pancreaticoduodenectomy for distal cholangiocarcinoma: a multicenter propensity score-matched study.Surgical Endoscopy2022.
Xu S, Zhang X, Zhao G, Zhao Z, Gao Y, Hu M, Tan X, Liu R. Development and Validation of an Online Calculator to Predict Early Recurrence and Long-term Survival in Patients with Distal Cholangiocarcinoma after Pancreaticoduodenectomy.J Hepato-Bil-Pan Sci2021.
Vickers AJ, Holland F. Decision curve analysis to evaluate the clinical benefit of prediction models. Spine J. 2021;21(10):1643–8.
Junrungsee S, Kittivarakul E, Ko-iam W, Lapisatepun W, Sandhu T, Chotirosniramit A. Prognostic Factors and Survival of Patients with Carcinoma of the Ampulla of Vater after Pancreaticoduodenectomy. Asian Pac J Cancer Prevention: APJCP. 2017;18(1):225–9.
Kim N, Han I, Ryu Y, Hwang D, Heo J, Choi D, Shin S. Predictive Nomogram for Early Recurrence after Pancreatectomy in Resectable Pancreatic Cancer: Risk Classification Using Preoperative Clinicopathologic Factors.Cancers2020, 12(1).
Sahara K, Tsilimigras D, Toyoda J, Miyake K, Ethun C, Maithel S, Abbott D, Poultsides G, Hatzaras I, Fields R, et al. Defining the Risk of Early Recurrence Following Curative-Intent Resection for Distal Cholangiocarcinoma. Annals of surgical oncology; 2021.
Yoen H, Kim JH, Hur BY, Ahn SJ, Jeon SK, Choi S-Y, Lee KB, Han JK. Prediction of tumor recurrence and poor survival of ampullary adenocarcinoma using preoperative clinical and CT findings. Eur Radiol. 2021;31(4):2433–43.
Park HM, Park S-J, Han S-S, Hong SK, Hong EK, Kim S-W. Very early recurrence following pancreaticoduodenectomy in patients with ampullary cancer. Medicine. 2019;98(44):e17711.
de Castro SMM, van Heek NT, Kuhlmann KFD, Busch ORC, Offerhaus GJA, van Gulik TM, Obertop H, Gouma DJ. Surgical management of neoplasms of the ampulla of Vater: local resection or pancreatoduodenectomy and prognostic factors for survival.Surgery2004, 136(5).
Sakata J, Shirai Y, Wakai T, Yokoyama N, Sakata E, Akazawa K, Hatakeyama K. Number of positive lymph nodes independently affects long-term survival after resection in patients with ampullary carcinoma. Eur J Surg Oncology: J Eur Soc Surg Oncol Br Association Surg Oncol. 2007;33(3):346–51.
Schneider M, Büchler MW. [Ampullary neoplasms: surgical management]. Der Chirurg; Zeitschrift Fur Alle Gebiete Der Operativen Medizen. 2021;92(9):796–802.
Zhang X-P, Gao Y-X, Xu S, Zhao G-D, Hu M-G, Tan X-L, Zhao Z-M, Liu R. A novel online calculator to predict early recurrence and long-term survival of patients with resectable pancreatic ductal adenocarcinoma after pancreaticoduodenectomy: A multicenter study. Int J Surg (London England). 2022;106:106891.

No competing interests reported.

SupplementaryMaterials.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Development and Validation of Web Calculators to Predict Early Recurrence and Long-Term Survival in Patients with Duodenal Papilla Carcinoma after Pancreaticoduodenectomy

Status:

Version 1

Abstract

Figures

Background

Methods

Study population

Inclusion and exclusion criteria

Follow up after resection

Study variables and definition

Statistical analysis

Results

Clinical characteristics of patients

Validation of the definition of DPC early recurrence

The long-term survival outcomes of patients in different risk groups

Development of the web-based online calculators in predicting ER and OS

Discussion

Conclusions

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1