Survival after curative pancreaticoduodenectomy for ampullary adenocarcinoma in a South American population: A retrospective cohort study

Abstract

BACKGROUND

Ampullary adenocarcinoma (AAC) is a rare neoplasm that accounts for only 0.2% of all gastrointestinal cancers. Its incidence rate is lower than 6 cases per million people. Different prognostic factors have been described for AAC and are associated with a wide range of survival rates. However, these studies have been exclusively conducted in patients originating from Asian, European, and North American countries.

AIM

To evaluate the histopathologic predictors of overall survival (OS) in South American patients with AAC treated with curative pancreaticoduodenectomy (PD).

METHODS

We analyzed retrospective data from 83 AAC patients who underwent curative (R0) PD at the National Cancer Institute of Peru between January 2010 and October 2020 to identify histopathologic predictors of OS.

RESULTS

Sixty-nine percent of patients had developed intestinal-type AAC (69%), 23% had pancreatobiliary-type AAC, and 8% had other subtypes. Forty-one percent of patients were classified as Stage I, according to the AJCC 8^th Edition. Recurrence occurred primarily in the liver (n = 8), peritoneum (n = 4), and lung (n = 4). Statistical analyses indicated that T3 tumour stage [hazard ratio (HR) of 6.4, 95% confidence interval (CI) of 2.5-16.3, P < 0.001], lymph node metastasis (HR: 4.5, 95%CI: 1.8-11.3, P = 0.001), and pancreatobiliary type (HR: 2.7, 95%CI: 1.2-6.2, P = 0.025) were independent predictors of OS.

CONCLUSION

Extended tumour stage (T3), pancreatobiliary type, and positive lymph node metastasis represent independent predictors of a lower OS rate in South American AAC patients who underwent curative PD.

Key Words: Gastrointestinal neoplasms; Adenocarcinoma; Ampulla; Pancreaticoduodenectomy; Survival; South America

Core Tip: The pancreatobiliary type of ampullary adenocarcinoma, lymph node metastasis and T3 tumour stage (AJCC 8^th Ed) are risk factors for lower overall survival in a South American population.

INTRODUCTION

Ampullary adenocarcinoma (AAC) is a rare neoplasm that represents 0.2% of all gastrointestinal cancers[1,2]. AAC has better prognosis and resection rates than pancreatic ductal adenocarcinoma (PDAC)[3,4]. This may be partly explained by the early symptom of jaundice caused by its location in the ampulla of Vater[5,6]. Nevertheless, three different epithelia (duodenal, biliary, and pancreatic) are present in the ampullary region[7], and their derived malignancies display different clinical behaviours[8]. Kimura and colleagues classified AAC into two histologic subtypes: Pancreatobiliary (PB) and intestinal (INT)[9]. Other features, such as preoperative CA 19-9[7], imaging[10], molecular phenotype[11,12], genetic mutations[13-15], and the diagnosis and classification of AAC[16], have been correlated with overall survival (OS). Consequently, the anatomic paradigm has shifted towards the interaction between genetic and epigenetic factors that determine OS and relapse-free survival (RFS)[14,17]. This may explain the wide range of outcomes reported in different centres (5-year OS: 30%-70%)[2].

However, most of these studies have been conducted in European, Asian, and North American countries. To the best of our knowledge, only one study has evaluated the impact of the lymph node ratio in predicting OS among AAC patients in Latin America[18]. Therefore, we evaluated the histopathologic predictors in AAC patients who underwent curative pancreaticoduodenectomy (PD) at the National Cancer Institute of Peru.

MATERIALS AND METHODS

Study design and patient selection

We conducted a retrospective cohort study in patients diagnosed with AAC who underwent curative (R0) PD between January 2010 and October 2020 at our tertiary centre. We specifically analysed histopathologic factors that influenced the patients’ overall survival. Our institutional review board approved this study (Protocol Number 21-17), according to the Declaration of Helsinki[19].

Histopathology

Double reads in a blinded manner by pathologists specializing in hepatobiliary cancers were applied to ensure the diagnosis of AAC and classification into INT intestinal (INT)- and pancreatobiliary (PB)-type according to Kimura et al[9,20].

Morphologically, INT-type tumours are reminiscent of colorectal adenocarcinoma, with solid nests, tall columnar cells, and elongated pseudostratified nuclei[21]. A significant proportion of INT-type is related to intestinal adenomas, which correlates with the adenoma-carcinoma sequence[22]. Conversely, PB-type adenocarcinomas are similar to extrahepatic bile duct and pancreatic duct adenocarcinomas. The glandular units have more pleomorphism than the intestinal type, with no evident nuclear pseudostratification, and they are separated by stroma[21]. Additionally, a mixed subtype has been described as having more than 25% of each INT and PB differentiation or with hybrid features, such as intestinal architecture with pancreatobiliary cytology[23,24]. Immunohistochemistry has led to a better classification of this mixed subtype; nevertheless, a standard definition has not been established[24,25]. In the present study, the following antibodies were used to determine the dominant type: MUC1 (#6151, BioSB, California, United States), MUC2 (#6158, BioSB, California, United States), CDX2 (MAD-000645QD-12, Vitro S.A., Spain), CK20 (MAD-0005105QD-12, Vitro S.A., Spain), and MUC5AC (MAD-000434QD-12, Vitro S.A., Spain). In cases of no definite conclusion, the tumour was classified as tubular into “other subtypes”.

Resection was classified as R0 when the 1-mm width of the surgical margin was free of neoplastic cells[26]. Tumour and nodal staging were categorized according to the AJCC 8^th Edition.

PD

PD was considered the treatment of choice because it was demonstrated to be a more radical approach to achieve satisfactory lymph node clearance and tumour-free surgical margins[27]. Patients were eligible for surgery after a comprehensive evaluation. The clinical parameters included performance and nutritional status, anatomy, and tumour extension (evaluated with contrast-enhanced computed tomography scan or magnetic resonance imaging). CA19-9 Levels were monitored within one month before surgery. We also assessed the vascular structures of the mesenteric and celiac axes along the diameter of the pancreatic duct.

Our surgical approach has been described previously[28]. In brief, the procedure was carried out using level 2 mesopancreas resection[29], and the pancreatic stump was managed using Blumgart, duct-to-mucosa, or modified dunking (at the discretion of the surgeon). In all cases, two Blake drains were placed around the pancreaticojejunostomy. Prophylactic octreotide was not used. External stents were applied in patients with a high risk of postoperative pancreatic fistula[30].

Adjuvant therapy

Patients with adjuvant therapy (AT) were interpreted as those who received chemotherapy (two or more courses), radiotherapy (with or without a sensitizing chemotherapy drug), or a combination of both. The AT regimen was left at the discretion of treating physicians, according to the best evidence available and/or institutional protocol.

Patient follow-up

Follow-ups and patient check-ups were performed on postoperative days 15, 30, and 90. computed tomography (CT) scans and CA 19-9 tests were scheduled every 4 mo after the index procedure during the first year, every 6 mo during the second year, and annually from the third year onward. The National Database for Civil Status (RENIEC) was solicited to determine the fate of patients. OS (months) was monitored from the date of surgery to the date of death or last follow-up, and patients with no events were censored. Any event (recurrence or death) was recorded during the follow-up. The cut-off for the last follow-up was 60 mo.

Statistical analysis

Continuous variables were reported as medians (interquartile ranges), and categorical variables were reported as counts (percentages). For the univariate analysis, the log-rank test was used, and the histopathologically relevant variables were integrated into a Cox regression model. Statistical analyses were performed with an alpha significance level of 0.05 using IBM SPSS v.25 (IBM Corp., Armonk, NY, United States) and R software (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Study population

From 2010 to 2020, 297 PDs were performed at the National Cancer Institute of Peru. Patients with R1/R2 resection, unavailable slides for revision, incomplete medical records, or synchronic neoplasms were excluded from the study. All patients included in the study underwent R0 resection. After a thorough revision of the medical files, 83 patients were included in the present study. Clinical, laboratory, and operative patient characteristics are presented in Table 1. The median age of the patient cohort was 59 years [interquartile range (IQR), 49-67], with a predominance of women (ratio = 1.3). The mean follow-up time was 39 mo. Twenty-five patients (30%) died during the follow-up period.

Table 1 Clinical, laboratory and operative patient characteristics (n = 83).

Clinical, laboratory and operative patient characteristics (n = 83)
Age (yr), median (IQR)	59 (49–67)
Sex, male/female, n (%)	36 (43)/47 (57)
Perioperative transfusion, n (%)	21 (25)
Haemoglobin in g/L, median (IQR)	115 (108–127)
Platelet count in 109/L, median (IQR)	285 (243–372)
International Normalized Ratio, median (IQR)	1.06 (1.01–1.15)
Serum glucose in mmol/L, median (IQR)	5.1 (4.8–5.7)
Serum creatinine in mmol/L, median (IQR)	53 (47–65)
Serum albumin in g/L, median (IQR)	38.1 (32–41.1)
Serum total bilirubin in µmol/L, median (IQR)	23.9 (12.9–60)
Serum CA 19-9 in IU/mL, median (IQR)	26.3 (10–91.4)
Pancreaticoduodenectomy
Pylorus-preserving PD, n (%)	69 (83)
Whipple procedure, n (%)	14 (17)

IQR: Interquartile range; PD: Pancreaticoduodenectomy.

Histopathologic characteristics

Sixty-nine percent of patients had developed INT-type AAC (69%), 23% PB-type AAC, and 8% other subtypes (including five patients with the tubular subtype and two patients with the tubular subtype with signet ring cells). Approximately 40% of cases demonstrated pancreatic invasion (T3 tumour stage), and 40% of patients had lymph node metastasis. Thirty-four (41%), 20 (24%), and 29 (35%) patients had stage I, II, and III disease, respectively. The histopathological characteristics of the cohort are shown in Table 2.

Table 2 Histopathologic characteristics (n = 83).

Histopathologic characteristics (n = 83)
Tumour size in mm, median (IQR)	27 (17–40)
Subtype, n (%)
Intestinal	57 (69)
Pancreatobiliary	19 (23)
Others	7 (8)
Tumour status, n (%)
T1	7 (8)
T2	44 (53)
T3	32 (39)
Number of lymph nodes assessed, median (IQR)	17 (12–24)
Lymph node status, n (%)
N0	50 (60)
N1	22 (26)
N2	11 (14)
Differentiation, n (%)
Well differentiated	25 (30)
Moderately differentiated	53 (64)
Poorly differentiated	5 (6)
Lymphovascular invasion	30 (36)
Perineural invasion	26 (31)

IQR: Interquartile range.

Use of AT

Twenty-four patients received AT (15 patients underwent chemotherapy, two patients underwent radiotherapy, and seven patients were subjected to both treatments). The most frequently employed chemotherapy regimen included gemcitabine, which was administered to 20 patients (24%). When chemoradiotherapy was applied, a dose of 4500 cGy in 25 sessions was administered using capecitabine as a sensitizing agent.

The evaluation of AT on OS was impaired by the heterogeneity of the AT regimen and the number of patients. Therefore, we decided not to include the AT variable in the survival analysis.

Patterns of recurrence

Recurrent distant metastases were diagnosed during the postoperative period in the liver (n = 12), peritoneum (n = 8), and lung (n = 7). Additionally, lymph node recurrences around the superior mesenteric artery and the retroperitoneal space were primarily observed in one and two patients, respectively (Table 3).

Table 3 Recurrence patterns after pancreaticoduodenectomy (n = 19).

Organs involved
Distant metastasis, n (%)	(A) First organ	(B) Second organ	(C) Third organ	A + B + C	%
Liver	8	3	1	12	32
Peritoneum	4	3	1	8	22
Lung	4	2	1	7	19
Supraclavicular lymph node	1			1	3
Bone		1		1	3
Suprarenal gland			1	1	3
Sub-table total				30	81
Lymph nodal recurrence, n (%)
Celiac trunk		1		1	3
Hepatic hilum		1		1	3
Mesenteric lymph nodes	1	1		2	5
Retroperitoneal lymph nodes	2	1		3	8
Sub-table total				7	19
Total				37	100

Overall survival and prognostic factors

The 5-year OS rate in the cohort was 62% (Figure 1). Applying the Cox regression model, three predictive factors were identified, i.e., T staging, lymph node metastasis, and PB type. Time and outliers had no impact on these independent factors, according to the modelling Supplementary Figures (Table 4).

Figure 1 Survival probability of patients with adenocarcinoma of the ampulla of Vater undergoing pancreaticoduodenectomy.

Table 4 Cox regression model analysis for predictors of overall survival.

Variables	Hazard ratio	95%CI		P value
		Lower	Upper
Age in yr				0.355
Tumour size in mm	1.03	1	1.06	0.059
Histopathologic subtype
Intestinal/other types
Pancreatobiliary type	2.7	1.2	6.2	0.025
T classification
T1-T2
T3	6.4	2.5	16.3	< 0.001
Lymph node metastasis
No
Yes	4.5	1.8	11.3	0.001
Differentiation grade				0.54
Well differentiated
Moderately differentiated				0.268
Poorly differentiated				0.755
Perineural invasion				0.517
Lymphovascular invasion				0.26

CI: Confidence interval.

Impact of the T tumour classification

Univariate analysis showed lower OS in patients with T3 classification (P < 0.001). The 5-year OS rates were 80% in T1/T2 patients and 30% in T3 patients, with a median OS of 30% in the latter group. According to the multivariate analysis, T3 patients had an HR of 6.4 (95%CI: 2.5-16.3, P < 0.001) (Figure 1).

Effect of lymph node invasion

Patients with lymph node metastases (N+) had a lower survival rate than those with no lymph node invasion (N0) (P = 0.001). The 5-year OS rates in the N+ and N0 groups were 38% and 80%, respectively. The median OS was 46 mo in the N+ group. The HR was 4.5 (95%CI: 1.8-11.3, P = 0.001) (Figure 1).

Influence of the histopathologic subtype

PB-type patients had a lower OS than patients with INT or other subtypes (P = 0.004). The 5-year OS rate for PB-type patients was 38%, whereas patients with INT or other subtypes had a 5-year OS rate of 70%. The median OS was 46 mo in PB-type patients, whereas the OS in the intestinal/other group was not reached during the follow-up period. The HR was 2.7 (95%CI: 1.2-6.2, P = 0.025) in PB-type patients (Figure 1).

DISCUSSION

To the best of our knowledge, the present study represents the first retrospective histopathologic work on AAC performed in a tertiary centre in South America, in which PD and the multimodal approach are standard. Our findings indicate that T3 tumour classification (pancreatic invasion), positive lymph node metastasis, and PB type are independent prognostic factors of OS in AAC patients treated with PD (R0).

Various factors have previously been described to be associated with AAC patient outcomes. In a meta-analysis, Zhou and colleagues identified age (> 65 years old), tumour size (> 20 mm), poor differentiation, PB-type, pT3-T4 stage diseases, lymph node metastasis, perineural invasion, lymphovascular invasion, pancreatic invasion, and positive surgical margins as independent factors associated with lower survival[32]. However, Koprowski and colleagues claimed that histotypes were not correlated with OS and concluded that disease stage was the primary determinant of patient outcomes[33]. In this study, the authors report 32% locoregional recurrence, despite the median number of retrieved lymph nodes and the low number of patients with R1 resection. Moreover, Quero and collaborators recently corroborated this finding about no difference between INT- and PB-types, but higher overall and recurrence-free survivals with excision of the mesopancreas[34].

Since AT allocation is based on tumour and nodal stages, we decided to consider these variables in the Cox model. We further stratified the patient cohort according to histopathologic subtypes (i.e., INT, PB, and "others"). Of note, we did not observe the mixed subtype in our cohort from South America, contrasting with the studies published in other regions of the world[16,31].

Our model supports the predictive impact of the histology of AAC on survival in a patient cohort from South America. In our hands, PB type, pT3 stage, and lymph node metastases were associated with lower OS; other variables scrutinized were not significantly associated with OS. The low rate of locoregional recurrence reported in our cohort could be partly explained by the application of level 2 mesopancreas resection, in accordance with the data by Quero and collaborators[34].

AAC has been documented to have a better prognosis than PDAC. However, the present study suggests that there are detrimental factors associated with subgroups of AAC patients, with OS rates comparable to PDAC (Figure 2). In this regard, our data suggest that a better outcome would be primarily explained by the biology of the tumour and secondarily by its location. Hence, assessing the impact of AT in high-risk patients is of utmost relevance. In the ESPAC-3 study, which included 428 patients with periampullary adenocarcinoma, the use of chemotherapy (5-fluorouracil /leucovorin or gemcitabine) demonstrated a benefit in OS (HR 0.75) but no greater effectiveness based on the histological type[35]. Additionally, a multicentre retrospective analysis did not report any benefit of adjuvant chemotherapy in AAC patients, including those with high-risk criteria (N+ or advanced stages T3 and T4)[36]. Other studies have provided more contrasting results on the impact of adjuvant chemotherapy on OS[31,37-39]. Regarding adjuvant radiotherapy, benefits have essentially been analysed among PDAC patients, preventing definite conclusions in AAC patients[40-42]. A recent meta-analysis showed that AT, especially chemoradiotherapy, was associated with increased OS among patients with PB-type or high-risk factors[43].

Figure 2 Comparison of survival probability between the intestinal/other (A) and pancreaticobiliary (B) types in patients with pT3 and pN⁺ adenocarcinoma of the ampulla of Vater.

There is a lack of specific guidelines for AAC, except one that comprises the management of biliary tract and ampullary carcinomas[44]. The authors recommend AT in patients with high-risk features (pancreatic invasion, lymph node metastasis, and perineural invasion) but did not specify any regimen. The predictive ability of mutation driver mutations (e.g., TP53, KRAS, and ELF3) in AAC histotypes has not been studied in great detail[45]. The characterization of AAC patient subgroups, based on their molecular alterations, would provide information on the choice of AT after radical surgery.

There are some limitations to recognize in the present study. Our primary AAC patient population displayed a high perioperative mortality rate (10 patients were excluded from this study), which we addressed and analysed previously[28]. We consider this a very important drawback, in addition to the retrospective design of the study. Another weakness was the heterogeneity in the multimodal management of the patients, which is reflected in international practices[31,39,46]. Therefore, we decided not to evaluate the impact of AT, as few patients would have been included in each group. Accordingly, further prospective studies are required because of the limited evidence available to date.

CONCLUSION

PB type, T3 tumour stage, and positive lymph node metastasis are independent predictors of lower survival in South American patients with ampullary adenocarcinoma treated by curative pancreaticoduodenectomy. Further evaluation of adjuvant and multimodal treatments is warranted, especially in patients with these high-risk factors.

ARTICLE HIGHLIGHTS

Research background

Ampullary adenocarcinoma (AAC) is a rare neoplasm that has not been studied previously in South American countries.

Research motivation

AAC might have different patterns of recurrence and overall survival than what has been reported in centres from Europe, Asia or North America.

Research objectives

To identify risk factors and their impact on overall survival in patients who underwent pancreaticoduodenectomy (PD) for AAC.

Research methods

We conducted a retrospective cohort study and analysed histopathologic predictors of survival in a Cox regression model.

Research results

Nearly two-thirds of patients had the intestinal-type AAC and around 25% had the Pancreatobiliary (PB)-type AAC. However, overall survival (OS) was lower for the latter subtype. Independently of the T3 and N+ tumour stage.

Research conclusions

Patients with PB-type AAC, T3 and N+ tumour stage are at higher risk of lower survival after curative PD.

Research perspectives

Identification of high-risk patients would guide the clinicians for the use of AT. Further studies are warranted.

ACKNOWLEDGEMENTS

We would like to thank Pr. Lorenzo-Bermejo J for his valuable support in the review of the statistical methods.