Pancreatic head resection for carcinoma of the ampulla vateri – better long-term prognosis, but more postoperative complications

Our study was performed as a single center study at the University Medical Center Freiburg. Clinical data of 627 patients with pancreatoduodenectomies either due to pancreatic ductal adenocarcinoma (PDAC) or due to ampullary carcinoma (CAMP) in our institution between January 2002 and December 2021 were evaluated retrospectively, using a prospectively maintained pancreatic surgery database. Patients with total pancreatectomies were excluded due to reasons of homogenization. Details concerning patient collective are shown in Fig. 1.

Data collection at our clinic is performed continuously using a prospectively maintained pancreatic surgery SPSS database. Besides demographic data, preoperative BMI and ASA-score, variables include details on duration of surgery, blood transfusions and surgical techniques as well as duration of hospital stay and in-hospital-mortality. Follow-up studies with general practitioners or oncologists and cancer registries provide information on long-term survival. Primary outcome of our study was long-term overall survival, secondary outcomes were postoperative complications and 30-day postoperative mortality. Postoperative complications such as postpancreatectomy hemorrhage (PPH), pancreatic fistula (POPF) or delayed gastric emptying (DGE) were graded using current international definitions of the International Study Group on Pancreatic Surgery (ISGPS) [16‐19].

Statistical analysis was performed using SPSS (IBM SPSS Statistics for Windows, Version 28.0. IBM Corp., Armonk, NY, USA) and GraphPad Prism (GraphPad Software, Version 10, San Diego, CA, USA). After performing explorative analysis and descriptive statistics, statistical significance was examined by using chi-square tests and Fisher´s exact tests for categorical variables and ANOVA for continuous variables. Survival status was obtained from the comprehensive cancer center registry at our institution and/or from the computerized hospital information system. Overall survival was analyzed using the Kaplan–Meier method, differences in overall survival were assessed using log-rank tests and uni- and multivariable Cox regression models (forward selection method with likelihood ratio). Results with a p-value < 0.05 were considered statistically significant. Propensity score matching was performed to reduce bias for different surgical techniques. Multivariable logistic regression model was performed to generate the propensity score. The following factors were included in this model: laparoscopic resection, technique of reconstruction (pancreaticojejuno- vs. pancreatogastrostomy) and portal vein resection. After establishing the propensity score, 1:1 matching using the nearest-neighbour matching was performed with a caliper of 0.01 without replacement. Post hoc balance diagnostic was performed using mean standardized differences [20].

Data collection and analysis were performed in accordance with the Declaration of Helsinki and were approved by the local ethics committee (Ethics Committee of Albert-Ludwigs-University Freiburg, Germany, EK-No. 23-1424-S1-retro).

The total number of patients included in this study was 627 (109 (17.4 %) in the CAMP- and 518 (82.6 %) in the PDAC-group). In the unmatched cohort, we found no difference concerning sex (51.9% vs. 54.1% male patients, p = 0.676) or age of patients (66.3 vs. 66.5 years, p = 0.848). Mean body mass index (BMI) was 25.1 kg/m² in PDAC patients vs. 25.0 kg/m² in CAMP patients (p = 0.775). There was no significant difference in ASA (American Society of Anesthesiology) score, most patients in both groups had an ASA-score of 2 or 3 (93.3 % PDAC vs 95.4% CAMP, p = 0.680). There was no difference in relevant comorbidities concerning coronary heart disease, hypertension, lung disease, liver disease or renal insufficiency. However, there was a trend towards more preoperative diabetes mellitus in the pancreatic adenocarcinoma group (24.2% vs 15.7%, p = 0.057). Concerning preoperative bile duct stenting, we found no difference between both groups (PDAC 55.3% vs. CAMP 63.6%, p = 0.116). Significantly more patients in the PDAC group received neoadiuvant (7.7% vs. 0.0%, p = 0.005) or adiuvant chemotherapy (55.2 vs. 25.6%, p < 0.001), respectively. Operation time for CAMP was significantly shorter than in the PDAC-group (mean operative time 391 min vs. 432 min, p < 0.001). Venous resections were necessary in 42.3% of PDAC-patients but only in 8 patients (7.4%) in the CAMP-group (p < 0.001). Intraoperative assessment of the pancreatic texture revealed a soft pancreas in 70.6% of CAMP, but only in 38.1% of PDAC-cases (p < 0.001). Surgeons were free to choose a suitable reconstruction method according to the intraoperative situation. Reconstruction techniques in the CAMP group were pancreatogastrostomy in 48.6% (n = 53) of patients and pancreaticojejunostomy in 51.4% (n = 56). There was a significantly different distribution in the PDAC-group: pancreatogastrostomy was performed in 29.9% (n = 155) and pancreaticojejunostomy in 70.1% (n = 363) of patients. There was a trend towards more laparoscopically assisted resections in the CAMP group compared with the PDAC group (PDAC 21.6%, n = 112; CAMP 30.3%, n = 33; p = 0.051). Patient characteristics and intraoperative parameters of the unmatched cohort are summarized in Table 1.

In the unmatched cohort, tumor-free resection margins were achieved in 97.2% of cases in CAMP patients and in 75.8% of patients with PDAC (p < 0.001). Most PDAC tumors were of T3 state (71.3%) whereas in the CAMP group, there were nearly as many T2 as T3 tumors (34.3% and 39.8%). Remarkably, there were more T4 tumors in the CAMP group than in the PDAC group (12.0% vs. 2.3%). Concerning lymph node affection, there was also a significant difference with a higher N0-rate in the CAMP-group (46.8% vs. 29.3%, p < 0.001). Histopathological results of the unmatched cohort are shown in Table 2.

In the unmatched cohort, there was no significant difference in overall postoperative complication rate between CAMP and PDAC patients (64.2% vs. 56.1%, p = 0.119), but there were significantly more surgical complications in the CAMP-group than in the PDAC-group (57.9% vs. 40.9%, p = 0.001), mainly caused by a significantly higher rate of clinically relevant pancreatic fistula (CR-POPF) with 30.5% CR-POPF in CAMP patients compared to only 12.4% in PDAC patients (p < 0.001). Concerning delayed gastric emptying and postpancreatectomy hemorrhage, we found no difference between both groups (DGE 26.2% CAMP vs. 22.8% PDAC, p = 0.455; PPH 6.6% CAMP vs. 8.6% PDAC, p = 0.534). More patients with ampullary carcinomas needed a conservative therapy following pancreatoduodenectomy (73.1% vs. 58.6%, p = 0.006). Interestingly, the rate of acute kidney failure was significantly higher in PDAC patients (4.1% vs. 0.0%, p = 0.034). There was no difference concerning the need of surgical revisions between both groups (12.8% PDAC vs. 11.9% CAMP, p = 0.810). Postoperative 30-day-mortality was similar in both groups (4.1% vs. 3.7%, p = 0.849). The length of hospital stay was significantly longer in the CAMP group due to more surgical complications in the postoperative course (19 vs. 17 days, p = 0.012); however, there was no significant difference concerning the length of stay on the intensive care unit (ICU) (6 vs. 5 days, p = 0.562). Postoperative complications and survival of the unmatched cohort are summarized in Table 3.

Survival data were available for 516 PDAC- and 109 CAMP-patients. Median overall survival of CAMP patients (all T-, N- and R- states) was 53 months (95%-CI 19.2 - 86.8 months) compared to 21 months (95%-CI 18.9 – 23.1 months) in the PDAC group (p < 0.001) (Fig. 2A). Considering only tumors with R0 resection status, median survival in the CAMP group was 59 months (95%-CI 26.9 – 91.1 months; n = 106) vs. 23 months (95%-CI 20.4 – 25.6 months) in the PDAC group (n = 392; p < 0.001) (Fig. 3A), whereas survival after R1 resection was 14 months for CAMP patients (95%-CI 0.0 – 28.4 months; n = 3) and 13 months for PDAC patients (95%-CI 7.4 – 18.5 months; n = 116) (p = 0.285).

We performed a multivariable logistic regression model for development of the propensity score (details are shown in Supplementary Table 1). After 1:1 matching using the nearest-neighbour method, we identified 204 patients (102 PDAC patients and 102 CAMP patients) with comparable baseline and surgical characteristics (Table 4). Covariates which were used for development of the propensity score showed mean standardized differences <= 0.01 indicating adequate balance of the matched variables.

In the matched cohort, we found no significant differences concerning delayed gastric emptying (DGE B/C 19.0% PDAC vs. 24.0% CAMP, p = 0.389) and postpancreatectomy hemorrhage (PPH B/C 9.9% PDAC vs. 7.1% CAMP, p = 0.517) as well as concerning wound infections (19.8% vs. 18.0%, p = 0.744) or intraabdominal abscesses (12.9% vs. 11.9%, p = 0.831). Overall complications and 30-day-mortality were distributed equal as well between PDAC and CAMP patients (complications 56.4% vs. 64.7%, p = 0.228; mortality 4.0% vs. 3.9%, p = 0.989). There was just a trend concerning more postoperative surgical complications in CAMP patients (57.0% vs. 45.1 %, p = 0.091); however, patients with ampullary carcinomas still presented with a significantly higher rate of clinically relevant pancreatic fistula (CR-POPF 30.7% vs. 16.8% in PDAC patients, p < 0.001) and required significantly more conservative treatment following surgery (71.4% vs. 54.0%, p = 0.011). Details on histopathological results and postoperative complications in the matched cohort are summarized in Tables 5 and 6. Moreover, in the matched cohort, an improved overall survival of CAMP patients was consistent with 42 months median overall survival (95%-CI 17.1 - 66.9 months) in CAMP patients compared to 24 months (95%-CI 15.7 - 32.3 months) in PDAC patients (p = 0.003) (Fig. 2B). Considering only patients with R0 resections in the matched cohort, there was still a significantly better overall survival in patients with ampullary carcinomas (42 (95%-CI 15.1 – 68.9) vs. 26 (95%-CI 12.4 – 39.6) months, p = 0.006) (Fig. 3B). Moreover, by dividing the patients in two groups of either early or advanced primary tumors (T1/T2 vs. T3/T4), we found a significantly better overall survival for CAMP patients in the unmatched cohort (T1/T2 CAMP 128 (95%-CI 29.6 – 226.4) vs. PDAC 29 (95%-CI 21.3 – 36.7) months, p < 0.001; T3/T4 CAMP 27 (95%-CI 19.2 – 34.8) vs. PDAC 20 (95%-CI 18.0 – 22.0) months, p = 0.034) and still a trend towards a better overall survival for the early primary tumor stages of CAMP patients in the matched cohort (T1/T2 CAMP 128 (95%-CI 36.7 – 219.3) vs. PDAC 43 (95%-CI 10.9 – 75.1) months, p = 0.092; T3/T4 CAMP 20 (95%-CI 12.6 – 27.4) vs. PDAC 20 (95%-CI 12.1 – 27.9) months, p = 0.127). Survival curves for the different primary tumor stages in the unmatched and matched cohort are shown in Fig. 3C-F. Our analyses show that the prognosis of CAMP patients is altogether favorable in comparison to PDAC patients. In order to strengthen these data, we additionally performed Cox regression analyses in the unmatched cohort, highlighting the significant independent prognostic relevance of CAMP in comparison to PDAC. In these analyses, also the factors for propensity score matching were included. We performed uni- and multivariable Cox regression models using a forward selection method (forward variable selection, p(in) < 0.05, p(out) > 0.10, likelihood ratio). The results from this model can be found in Table 7.

As adiuvant treatment of most malignancies, including PDAC and ampullary carcinomas, has changed over the time, including more aggressive and effective chemotherapy regimens, we divided our patient cohort in two groups, depending on the decade of surgery (2002 – 2011 and 2012 – 2021), in order to evaluate a potential effect of these changes in adiuvant treatment on overall survival of patients. We found a consistently higher rate of adiuvant chemotherapeutical treatment in PDAC patients compared to CAMP patients for both decades: 61.6% PDAC vs. 17.9% CAMP from 2002 – 2011 (p < 0.001) and 52.1% PDAC vs. 31.4% CAMP from 2012 – 2021 (p = 0.002). In the second decade, we found an increasing adiuvant treatment in the group of CAMP patients compared to the first decade, but the rate of patients with adiuvant treatment remains significantly lower than in the PDAC group. However, in all of our analyses, CAMP patients present with a significantly better overall survival: 92 months (95%-CI 33.4 – 150.6 months) vs. 21 months (95%-CI 17.1 – 24.9 months; p < 0.001) from 2002 to 2011 and 33 months (95%-CI 20.3 – 45.7 months) for CAMP patients vs. 21 months (95%-CI 19.1 – 23.0 months) for PDAC patients (p = 0.010) from 2012 to 2021. Kaplan Meier curves for both decades are shown in Fig. 4A and B. Details on baseline characteristics and postoperative complications of both decades are summarized in Supplementary Tables 2-7.