Ampullary carcinoma: Treatment and prognosis

INTRODUCTION — Periampullary tumors are neoplasms that arise in the vicinity of the ampulla of Vater. Neoplasms that arise in this site can originate from the pancreas, duodenum, distal common bile duct (CBD), or the structures of the ampullary complex.

The ampulla of Vater is formed by the duodenal aspect of the sphincter of Oddi muscle, which surrounds the confluence of the distal CBD and main pancreatic duct as well as the papilla of Vater, a mucosal papillary mound at the distal insertion of these ducts on the medial wall of the duodenum (figure 1). Ampullary carcinomas are defined as those that arise within the ampullary complex, distal to the confluence of the distal common bile duct and the pancreatic duct (figure 2).

This topic review will cover the treatment and prognosis of ampullary carcinomas. The epidemiology, biologic behavior, clinical manifestations, and diagnosis and staging are covered separately. (See "Ampullary carcinoma: Epidemiology, clinical manifestations, diagnosis and staging".)

TREATMENT FOR LOCALIZED DISEASE — The only potentially curative treatment for ampullary carcinoma is surgical resection. Complete tumor resection with negative margins (R0 resection) is a prerequisite for cure.

It can be difficult to distinguish a primary ampullary carcinoma from other periampullary tumors preoperatively. However, true ampullary cancers have a better prognosis than periampullary malignancies of pancreatic or bile duct origin. Resectability rates are higher, and five-year survival rates are approximately 30 to 50 percent in patients with limited lymph node involvement. By contrast, less than 10 percent of patients with completely resected node-positive pancreatic cancer are alive at two years. Thus, an aggressive approach to diagnosis and treatment of periampullary tumors is needed to ensure that patients with these comparatively favorable cancers are treated optimally.

Current trends for locally advanced, borderline and categorically unresectable pancreatic cancer are to administer upfront (neoadjuvant) chemotherapy, with or without radiation therapy, prior to exploration. How initial treatment with chemoradiotherapy or with aggressive chemotherapy regimens such as FOLFIRINOX (leucovorin plus short-term infusional fluorouracil plus oxaliplatin and irinotecan) will affect the prognosis of cancers in the periampullary region that turn out to be ampullary and not pancreatic head cancers at the time of resection is not known. At least some data suggest potential benefit for neoadjuvant therapy in a subset of patients with ampullary cancer [1], but the selection of appropriate candidates is not established. (See "Initial chemotherapy and radiation for nonmetastatic, locally advanced, unresectable and borderline resectable, exocrine pancreatic cancer".)

Pancreaticoduodenectomy — Pancreaticoduodenectomy (Whipple operation) is considered the standard approach for ampullary cancer (figure 3). This can be done as a pylorus-preserving procedure or as a conventional pancreaticoduodenectomy, which includes an antrectomy (figure 4). Although some authors claim advantages of a pylorus-preserving procedure because of a shorter operative time and less intraoperative blood loss [2], this is based on European variations of the technique, which require creation of a Roux-en-Y to anastomose separately the gastric from the pancreatic and biliary anastomosis. Otherwise, there are no differences in long-term survival, and some series have shown higher incidence of delayed gastric emptying with the pylorus preservation. The impact of a pylorus-preserving procedure on long-term gastrointestinal function is less certain. Some studies suggest an improved nutritional state (as reflected by faster weight gain in the first postoperative year), but results have been inconsistent. (See "Pylorus-preserving pancreaticoduodenectomy" and "Overview of surgery in the treatment of exocrine pancreatic cancer and prognosis".)

Surgical outcomes from pancreaticoduodenectomy for ampullary cancer have improved over time. In contemporary single-institution series, rates of potentially curative resection have increased from approximately 80 to over 90 percent [3-7]. Long-term survival is possible after pancreaticoduodenectomy, even for patients with lymph node metastases or invasion beyond the duodenal wall (T3 disease (table 1)) [8]. (See 'Prognosis' below.)

Although pancreaticoduodenectomy has been associated with high perioperative morbidity and mortality rates in the past, contemporary series show that in experienced hands, perioperative (30-day) mortality rates are between 0 and 5 percent [3-7,9-12]. Perioperative morbidity rates are between 20 and 40 percent, with the most common problems being anastomotic (pancreatic) leak, delayed gastric emptying, and intra-abdominal infection [13,14]. (See 'Complications' below.)

These improved outcomes have been attributed to better operative technique and postoperative care. One of the most important reasons for this is the greater experience of a limited number of surgeons who perform the procedure regularly in high-volume institutions [15-17]. Good outcomes have been described even in patients older than 80 years in such centers. However, even within high-volume hospitals, operative mortality rates from pancreaticoduodenectomy vary considerably depending on the experience of the individual surgeon [18].

Preoperative biliary drainage — The most common presenting symptom of ampullary carcinoma is obstructive jaundice (80 percent) caused by obstruction or compression of the distal bile duct by the tumor. (See "Ampullary carcinoma: Epidemiology, clinical manifestations, diagnosis and staging".)

The role of preoperative biliary drainage in patients with periampullary tumors is controversial. Because obstructive jaundice can impair hepatic, renal, and immune function, it was hoped that preoperative relief of jaundice would correct these defects and decrease postoperative morbidity and mortality rates from pancreaticoduodenectomy. However, the available data from randomized trials of preoperative drainage versus no drainage are conflicting. Furthermore, three meta-analyses examining the benefit of preoperative biliary drainage for patients with obstructive jaundice have come to different conclusions, with one finding neither an adverse nor a favorable impact of preoperative stenting on the incidence of postoperative morbidity or mortality, another finding an overall adverse impact of stenting on the postoperative complication rate, and the third, significantly fewer postoperative complications in the stented group but no impact on postsurgical mortality. (See "Surgical resection of lesions of the head of the pancreas", section on 'Conventional versus modified pancreaticoduodenectomy' and "Overview of surgery in the treatment of exocrine pancreatic cancer and prognosis", section on 'Role of preoperative biliary drainage'.)

In the specific setting of ampullary cancer, the benefit of preoperative biliary drainage was addressed in a retrospective series of 82 patients undergoing potentially curative surgery at a single hospital in Singapore; 35 were drained preoperatively and 47 were not [19]. Preoperative drainage was associated with a significantly reduced incidence of postoperative wound infection (3 versus 26 percent), but there was no favorable impact on other postoperative complications or survival.

Uncertainty as to the benefit of preoperative drainage has led to differing approaches. Some surgeons routinely decompress jaundiced patients with an endoscopically placed stent prior to surgery. However, others reserve biliary decompression for selected patients in whom surgery will be delayed or those with debilitating pruritus or a clinical picture of cholangitis with fever and leukocytosis.

In practice, the majority of patients who present with obstructive jaundice will have been stented by a gastroenterologist before the diagnosis is established and it is known whether or not the patient is a surgical candidate; the surgeon usually has little influence on the decision. For those who are not stented, our preference is to proceed with endoscopic retrograde cholangiopancreatography (ERCP) and stenting only when there is high-grade jaundice (>15 mg/dL of bilirubin) and surgery will not take place within the following week.

The prognostic implication of obstructive jaundice at presentation on prognosis is discussed below. (See 'Obstructive jaundice' below.)

Complications — The most frequent treatment-related complication of pancreaticoduodenectomy is pancreatic fistula, and the rates are higher for patients with ampullary cancer than for pancreas cancer because the pancreatic parenchyma is typically normal in ampullary carcinoma. Other complications include delayed gastric emptying, hemorrhage, sepsis, bile leaks, and postoperative diabetes as a result of pancreatic resection. (See "Surgical resection of lesions of the head of the pancreas", section on 'Perioperative morbidity and mortality'.)

Local resection — Many patients with ampullary cancer are older adults and have significant comorbidities. This has generated interest in less aggressive surgical options, such as local resection or ampullectomy for selected patients.

Experience with this approach is limited to small published series, in which most of the patients were considered high-risk candidates for surgery [3,20-30]. Comparison of these studies with each other is limited by different eligibility criteria for ampullectomy and the fact that the extent of surgery (eg, "ampullectomy" versus "local resection") has not always been clearly specified.

In the aggregate, the available data indicate that local resection is associated with lower morbidity than pancreaticoduodenectomy, but at the expense of higher recurrence rates and inferior survival, at least in the setting of invasive disease [13,23,25,31-35].

The largest series included 37 patients who were planned for ampullectomy rather than pancreaticoduodenectomy at Memorial Sloan-Kettering Cancer Center because of significant comorbidity or a small ampullary lesion (median 1.5 cm, range 0.4 to 4.2 cm) that was thought to represent a benign adenoma after a preoperative biopsy [33]. Intraoperatively, eight cases were converted from ampullectomy to pancreaticoduodenectomy because of disease extent or a frozen-section finding of invasive tumor. Eight of the 29 patients treated by ampullectomy alone had invasive tumor on the final histology; seven recurred, and only one was alive at last follow-up. By contrast, of the 91 patients undergoing pancreaticoduodenectomy during the same time period for invasive adenocarcinoma, 68 were still alive at the end of the study period, 49 without recurrence.

Early, low-grade tumors — In contrast to pancreaticoduodenectomy, ampullectomy does not accomplish removal of the regional lymph nodes. Because of the low rate of nodal metastases (less than 4 percent in most series), some have suggested that local resection is a reasonable approach for well-differentiated small (<6 mm) tumors that do not penetrate through the ampullary musculature (ie, Tis, T1 (table 1) [8]) [30,36-39].

However, the majority of the patients reported in these series had ampullary adenomas, and few had carcinomas. A major concern is the inferior cancer-specific survival following local resection of small ampullary invasive carcinomas in the analysis from Sloan-Kettering described above [37]. Furthermore, in other series, 45 percent of patients with pathologically confirmed T1 invasive ampullary adenocarcinomas had lymph node metastases [40].

Thus, while local ampullary excision could be considered for high-risk patients with well-differentiated T1 ampullary cancers that are <6 mm based on endoscopic ultrasound (EUS), a more aggressive surgical approach is preferred by most surgeons for invasive tumors, including well-differentiated T1 lesions, as long as the patient is a reasonable candidate for pancreaticoduodenectomy.

In our view, local resection (ampullectomy with lymph node sampling) is a reasonable alternative to pancreaticoduodenectomy for selected patients with noninvasive (pTis) tumors (table 1). However, pancreaticoduodenectomy is preferred for any invasive adenocarcinoma, if the patient can tolerate this approach.

Recommendations for management of ampullary carcinoma are not included in published guidelines from either the National Comprehensive Cancer Network (NCCN) [41] or the European Society for Medical Oncology (ESMO) [42].

Minimally invasive nonsurgical therapies — Minimally invasive nonsurgical therapies for ampullary carcinoma include endoscopic snare resection, Nd:YAG laser ablation, and photodynamic therapy. While potentially curative in the setting of ampullary adenomas, all three modalities of nonsurgical treatment provide palliative rather than curative benefit for patients with ampullary carcinoma. These methods are appropriate only for patients who are not operative candidates and those who refuse surgery.

The literature on these techniques in the setting of ampullary carcinoma is limited to single-case reports and small series.

●Endoscopic snare resection (papillectomy) is an effective means of treating ampullary adenomas. (See "Ampullary adenomas: Management".)

Endoscopic papillectomy has been attempted in early stage (Tis, T1) well-differentiated ampullary cancers without angiolymphatic invasion [37]. However, acceptance of this method for definitive treatment is hampered by the inadequacy of biopsy in terms of correct pathologic characterization of the lesion (and therefore its risk of harboring lymph node metastases) and the inability to predict successful, margin-negative endoscopic resection [22,43-46].

Endoscopic debulking has been used mainly preoperatively to permit stent insertion and decompression of the biliary tree. The controversy surrounding the need for preoperative biliary decompression is discussed above. (See 'Preoperative biliary drainage' above.)

●Laser ablation offers the potential for control of local tumor growth in patients who are unfit for more aggressive therapy. In one such series of 12 patients with ampullary cancer, duodenal obstruction was relieved in one, and the longest survival was 36 months (median 21 months) [47].

●Compared with laser ablation, photodynamic therapy (PDT) eradicates local tumor with less surrounding tissue destruction. PDT uses a photosensitizing drug (a hematoporphyrin derivative, Photofrin), which is disproportionately retained by malignant tissue after intravenous administration. Irradiation with visible light via a light-transmitting catheter (light-guide) placed through an endoscope and targeted at the tumor energizes the photosensitizing molecule within the tumor and catalyzes oxygen free-radical generation leading to cell death [48]. The major side effect of PDT is prolonged cutaneous photosensitivity after the procedure, requiring the patient to avoid sunlight and wear appropriate clothing to protect the skin for several weeks after treatment.

In the largest series of PDT involving 10 patients with ampullary cancer, remission was achieved for 8 to 12 months in three patients who had small tumors confined to the ampulla [49]. Tumor bulk was greatly reduced in four additional patients, while little effect was observed in the remaining three.

PROGNOSIS — The outcome of resected ampullary cancer depends on the extent of local invasion, status of the surgical margins, and the presence or absence of nodal metastases [6,13,50-52]. Five-year survival rates following pancreaticoduodenectomy range from 64 to 80 percent for patients with node-negative disease, and from 17 to 50 percent for node-positive disease [3,5,9-13,15,20,31,32,53-58].

Survival curves from one retrospective single-institution series, stratified according to surgical stage (as defined by the 2017 American Joint Committee on Cancer [AJCC]/Union for International Cancer Control [UICC] staging system (table 1)) are illustrated in the figure (figure 5).

Outcomes tend to be somewhat worse in population-based analyses. As an example, five-year survival rates stratified by stage at presentation from a series of 1301 patients with ampullary cancer reported to the National Cancer Institute SEER (Surveillance, Epidemiology, and End Results) database between 1988 and 2003 are presented in the table (table 2) [52].

These survival data are better than similar presentations for pancreatic cancer, particularly for those with node-positive disease. It is unclear whether this represents a true difference in biology or earlier presentation of ampullary cancers due to earlier biliary obstruction. (See "Ampullary carcinoma: Epidemiology, clinical manifestations, diagnosis and staging", section on 'Epidemiology and biologic behavior'.)

Prognostic factors — The following sections will summarize the available data on predictors of recurrence.

Histology and histomolecular phenotype — In addition to depth of tumor invasion (the T stage (table 1)) and nodal status, high-grade histology and positive surgical margins are associated with a worse prognosis [3,7,11,13,20,56,57,59-67]. Whether there is a correlation between prognosis and morphologic type (ie, intestinal versus pancreaticobiliary) is unclear; the available data are conflicting, with some suggesting no difference, and others, a worse prognosis for the pancreaticobiliary type [67-71].

Contemporary data suggest that adenocarcinomas of the ampulla of Vater can be subdivided according to histologic subtype, immunohistochemical staining pattern, and molecular features into distinct subsets with differing biologic behavior:

●In a retrospective study of a cohort of 72 patients treated for ampullary adenocarcinoma in Sydney, Australia, those with a histomolecular pancreaticobiliary phenotype (CDX-negative, MUC1-positive) had a significantly worse outcome than did those with an intestinal phenotype (CDX-positive, MUC1-negative), median survival 16 versus 116 months [70]. When histomolecular phenotype was combined with the lymph node status, three subsets of ampullary adenocarcinomas emerged with significantly different survival outcomes:

•Patients with a node-negative, non-pancreaticobiliary histomolecular phenotype tumor had an excellent prognosis (five-year survival 88 percent).

•Patients with a node-positive, pancreaticobiliary phenotype had a poor prognosis (five-year survival 20 percent).

•The remaining patients (node-positive, non-pancreaticobiliary phenotype, node-negative pancreaticobiliary phenotype) had an intermediate prognosis (five-year survival 47 percent).

The results were comparable in two additional independent cohorts of 90 patients from Glasgow, Scotland, and 46 from Verona, Italy.

●In another study, 146 resected "ampullary carcinomas" were carefully evaluated using stringent pathologic criteria to exclude distal bile duct and pancreatic cancer cases [71]. The authors ended up with 97 "true" ampullary carcinomas, the majority of which (72 percent) were intestinal, and the remainder either pancreatobiliary (23 percent) or mixed (5 percent). The median overall survival for this cohort was remarkably good (101 months), and the five-year survival was significantly better for the intestinal phenotype compared with pancreatobiliary/mixed phenotype (73 versus 56 percent). Genotyping was performed in 75 cases, and the majority (67 percent) were wild-type for KRAS. Tumors with a KRAS G12D mutation had a worse median survival (62 months) compared with other KRAS mutations or the wild-type genotype (145 and 155 months, respectively).

●However, others have failed to find a significant overall survival difference between the intestinal and pancreaticobiliary subtypes of ampullary cancer [72].

Given the conflicting data, the most recent 2017 Tumor, Node, Metastases (TNM) staging classification of the AJCC/UICC did not include histomolecular phenotype as a component of its prognostic stage groups (table 1) [8].

Whether and how this information could be used to individualize treatment decisions, particularly with regard to adjuvant therapy, is unclear. The impact of adjuvant therapy on outcomes according to histomolecular phenotype could not be addressed in the Australian study since only a minority of patients (64 of 208) in all three cohorts received adjuvant chemotherapy, and it was not randomly assigned [70]. The same was true in the Massachusetts General Hospital study, where only 31 percent of patients received additional postsurgical treatment [71]. A retrospective analysis of a prospective randomized cooperative group study exploring the role of adjuvant therapy in periampullary cancers found no significant improvement in the pancreaticobiliary type compared with the intestinal type of ampullary cancers when adjuvant therapy was compared with no adjuvant therapy [73]. However, prospective study of treatment selection based on histomolecular phenotype is needed before conclusions can be drawn as to the utility and clinical significance of histomolecular phenotype [74]. At present, adjuvant therapy recommendations for patients with ampullary cancer follow guidelines established for pancreatic cancer rather than intestinal cancer. (See 'Adjuvant therapy' below.)

Nodal metastases — The presence of nodal involvement portends a worse prognosis. However, among patients with nodal metastases, the number of affected nodes, particularly compared with the total number of examined nodes, is also of prognostic importance [7,75-78].

The number of involved nodes divided by the total number of examined nodes is referred to as the lymph node ratio (LNR). One study evaluated the utility of using the LNR for predicting recurrence and survival in 90 patients who underwent resection of ampullary carcinoma [75]. The median number of resected nodes was 16, and patients with 16 or more examined nodes had a significantly lower recurrence rate and better five-year survival (81 versus 45 percent) than patients whose pathology material contained 16 or fewer nodes. Five-year survival was also predicted by the LNR:

●LNR = 0 – 75 percent

●LNR >0 to ≤0.2 – 49 percent

●LNR >0.2 to ≤0.4 – 38 percent

●LNR >0.4 – 0 percent

Given these data, in the 2017 revision of the TNM staging system, which went into effect in the United States on January 1, 2018, the number of involved regional lymph nodes influences N stage (table 1) [8]. The relevance of the new T and N stage designations and stage groupings on prognosis is outlined in the figure (figure 5) [79]. (See "Ampullary carcinoma: Epidemiology, clinical manifestations, diagnosis and staging", section on 'TNM staging system'.)

The implications of these data on the extent of lymph node dissection (ie, whether there is a role for extended lymphadenectomy) are unclear. A prospective, randomized study of 62 patients who underwent resection of ampullary carcinoma found no difference in the five-year survival in the group undergoing standard versus extended lymphadenectomy (56 versus 60 percent) [80]. The authors concluded that the added morbidity of extended lymphadenectomy could not be justified by the better oncologic outcomes. We agree with this conclusion, and do not routinely perform extended lymphadenectomy in these patients. The surgeon and pathologist should aim to dissect and analyze at least 12 lymph nodes.

Obstructive jaundice — Patients who present with obstructive jaundice tend to have a worse prognosis. In one study, 5- and 10-year survival rates were 70 and 49 percent, respectively, in patients who did not present with obstructive jaundice versus 34 and 29 percent, respectively, in patients who were jaundiced at presentation [81]. These data could be interpreted as demonstrating that early detection of ampullary carcinoma prior to the onset of obstructive jaundice is associated with a better oncologic outcome. The role of preoperative biliary drainage in patients who present with obstructive jaundice is addressed above. (See 'Preoperative biliary drainage' above.)

Intraoperative blood transfusion — In a systematic review, patients who required intraoperative transfusion of more than three units of red blood cell units had worse outcomes than those requiring less blood [82]. However, patients who require transfusion tend to be sicker as a group (or their tumors are more advanced, requiring more intraoperative dissection) than those who do not require transfusion. Therefore, the independent contribution of intraoperative blood transfusion to outcomes remains uncertain.

Tumor marker elevation — Elevated preoperative levels of the serum tumor markers CA 19-9 (carbohydrate antigen 19-9, also called cancer antigen 19-9) and carcinoembryonic antigen (CEA) are associated with a poorer prognosis relative to individuals with normal values [54,83]. In one study, the optimal cutoff levels of CA 19-9 to stratify risk for disease recurrence was >150 units/mL in non-jaundiced patients and >300 units/mL in the presence of cholestasis [83]. (See "Ampullary carcinoma: Epidemiology, clinical manifestations, diagnosis and staging", section on 'Serum tumor markers'.)

Prognostic nomograms — Prognostic nomograms for disease-specific survival have been developed for patients with non-metastatic ampullary cancer such as age, tumor size and depth of invasion, lymph node ratio, and histology [84]. Whether models such as these are better at predicting prognosis as compared with AJCC stage grouping (figure 5), is not clear.

Patterns of recurrence — In many series, recurrences are approximately evenly split between locoregional recurrence and distant spread [12,85], although others note a predominance of distant metastases [54,64]. The risk factors for locoregional and distant recurrence are slightly different. This was shown in a series of 127 patients who underwent pancreaticoduodenectomy with regional lymphadenectomy for ampullary carcinoma [12]. The median time to recurrence was 11.4 months, and the risk factors for locoregional recurrence were the presence of pancreatic invasion and tumor size. By contrast, lymph node metastasis was the sole risk factor for distant recurrence. Both pancreatic invasion and lymph node involvement were significant predictors of inferior survival.

The most common site of distant spread is the liver, but other sites include peritoneum, bone, supraclavicular lymph nodes and lung [12,54,85].

ADJUVANT THERAPY

Our recommended approach — The results of clinical trials are not definitive, and there is no consensus regarding the optimal management of patients after resection of an ampullary cancer. Recommendations for management of ampullary carcinoma are not included in published consensus-based guidelines from either the National Comprehensive Cancer Network (NCCN) [86] or the European Society for Medical Oncology (ESMO) [42].

Many clinicians do not recommend adjuvant chemoradiotherapy or chemotherapy for resected ampullary cancers, citing the more favorable prognosis of ampullary as compared with other biliary tract cancers [60,87] and the lack of data from randomized trials proving a survival advantage. However, we suggest that these patients be managed in a manner similar to the approach used for resected pancreatic cancer, even for those who have the intestinal histomolecular phenotype. (See 'Histology and histomolecular phenotype' above.)

Eligible patients should be encouraged to enroll in clinical trials evaluating the potential benefits of chemotherapy and/or chemoradiotherapy, as well as new therapies. If protocol therapy is not available or declined, the approach differs outside of and within the United States:

●Most European clinicians advocate chemotherapy alone, emphasizing the survival benefit of chemotherapy in the German CONKO-001 trial [88], the lack of a significant survival benefit with chemoradiotherapy in an early European Organisation for Research and Treatment of Cancer (EORTC) trial [89], and the detrimental impact of chemoradiotherapy on survival seen in the European Study Group for Pancreatic Cancer (ESPAC)-1 trial [90]. Similarly, the Japanese approach also includes chemotherapy alone.

If chemotherapy alone is chosen, for patients with an excellent performance status who are able to tolerate it, we suggest modified FOLFIRINOX (oxaliplatin plus irinotecan with leucovorin and short-term infusional fluorouracil [FU] (table 3)), rather than gemcitabine alone, based on results from the Canadian/Unicancer/PRODIGE group demonstrating improved survival with FOLFIRINOX compared with gemcitabine in the adjuvant setting for pancreatic cancer. We also prefer this regimen over gemcitabine plus capecitabine. (See "Treatment for potentially resectable exocrine pancreatic cancer", section on 'Modern combination regimens (FOLFIRINOX and gem-nabpaclitaxel)'.)

If FOLFIRINOX is deemed too toxic for the patient, then adjuvant therapy with gemcitabine plus capecitabine as per the ESPAC-4 trial (table 4) is a reasonable approach. (See "Treatment for potentially resectable exocrine pancreatic cancer", section on 'Other gemcitabine and FU-based approaches'.)

Therapy with gemcitabine alone (table 5) or, where available, S-1 alone is also a reasonable option, particularly for patients with a borderline performance status or a comorbidity profile that precludes combination chemotherapy. (See "Treatment for potentially resectable exocrine pancreatic cancer", section on 'Other gemcitabine and FU-based approaches'.)

When chemoradiotherapy is not given, we suggest six months of adjuvant chemotherapy.

●The American approach differs with regard to chemoradiotherapy. For most patients with resected ampullary cancer stage T2N0 or higher (table 1), we suggest the addition of concurrent infusional FU-based chemoradiotherapy to adjuvant chemotherapy. However, the benefit of adjuvant radiation therapy (RT) remains controversial, and this approach is not typically chosen outside of the United States. (See "Treatment for potentially resectable exocrine pancreatic cancer", section on 'Does chemoradiotherapy add benefit to chemotherapy?'.)

As with pancreatic cancer, the optimal way to sequence FU-based chemoradiotherapy and systemic chemotherapy is unclear. When chemoradiotherapy is planned, our preferred approach is four months of chemotherapy followed two to four weeks later by six weeks of concurrent RT with either infusional FU (250 mg/m² daily) or capecitabine (825 mg/m² twice daily, five days per week during RT).

Benefit from adjuvant therapy — Despite the high rate of potentially curative resections in contemporary series, more than one-half of patients die from recurrent disease, suggesting the need for effective adjuvant therapy. There are few published data to guide the use of adjuvant therapy in patients with resected ampullary cancer.

Chemoradiotherapy — Benefit from postoperative chemoradiotherapy in patients with completely resected disease has been suggested in several uncontrolled series [59,91-97]. In one of the largest reports from the Mayo Clinic, 29 of 125 patients who underwent definitive surgery for an ampullary cancer received adjuvant RT with concurrent FU, while the remainder received no adjuvant therapy [91]. In multivariate analysis, lymph node status emerged as the only significant predictor of outcome. Overall survival rates at one, three, and five years in the adjuvant therapy group were 91, 54, and 48 percent, compared with 66, 22, and 11 percent in the control group. Within the high-risk node-positive subgroup (n = 54), the 24 patients who received adjuvant therapy survived significantly longer than the 30 who did not receive it.

On the other hand, other retrospective comparisons [98,99], and the only phase III randomized trial that included a substantial number of patients with ampullary carcinoma have failed to show a benefit for postoperative chemoradiotherapy. In a trial from the EORTC, 218 patients with resected pancreatic or other periampullary cancers were randomly assigned to postoperative RT (40 Gy in split courses) plus concurrent FU (25 mg/kg per day by continuous infusion) or observation [89]. For the 104 periampullary cancers (which included cancers of the ampulla, distal common bile duct or duodenum), there was no difference in the two-year survival rate (67 versus 63 percent) or in the incidence of locoregional recurrence in the treated patients compared with controls. (See "Treatment for potentially resectable exocrine pancreatic cancer", section on 'Fluorouracil-based approaches'.)

Even the results of meta-analyses are discordant:

●A year 2016 meta-analysis of 10 retrospective reports totaling 3361 patients who either received adjuvant chemoradiotherapy or no adjuvant therapy after resection of an ampullary carcinoma concluded that despite the fact that more patients had locally advanced disease or nodal metastases in the treated group, adjuvant chemoradiotherapy significantly reduced the risk of death (hazard ratio [HR] 0.75, 95% CI 0.6-0.94) [100]. However, there was significant heterogeneity between studies, mainly attributed to the two series that reported the largest benefit from adjuvant RT [91,97].

●On the other hand, a later analysis of 14 studies of adjuvant therapy in periampullary adenocarcinoma (including six randomized trials, the remainder retrospective reports) concluded that there was no survival benefit for any adjuvant strategy (adjuvant chemoradiotherapy or chemotherapy) for treatment of periampullary cancers [101]. However, there were few T3 or T4 tumors in the analysis, and the data are most compelling for benefit of adjuvant therapy in this subgroup.

Chemotherapy alone — A benefit for adjuvant chemotherapy has been suggested in retrospective analyses [102,103], but has been difficult to prove in randomized trials:

●The benefit of adjuvant chemotherapy for resected ampullary adenocarcinomas was directly studied in the international ESPAC-3 trial, in which 428 patients with periampullary malignancies (297 ampullary, 96 bile duct, 35 other) were randomly assigned to one of three arms: observation alone, leucovorin-modulated FU (six courses of leucovorin 20 mg/m² IV bolus followed by FU 425 mg/m², daily for five days, once per month), or single-agent gemcitabine (1000 mg/m² weekly for three of every four weeks for six months) [73]. A complete (R0) resection was achieved in 84 percent of patients, and 59 percent had node-positive disease.

The use of adjuvant chemotherapy was associated with a potentially meaningful overall survival advantage but it was not statistically significant (median 43 versus 35 months, HR 0.86, 95% CI 0.66-1.11). Although of a greater magnitude, the difference in median survival between gemcitabine-treated and observed patients was still not statistically significant (median 46 versus 35 months, HR 0.77, 95% CI 0.57-1.05). However, in secondary multivariate analysis adjusting for predefined prognostic variables, there was a statistically significant survival benefit for any chemotherapy (HR for death 0.75, 95% CI 0.57-0.98) and for gemcitabine alone (HR 0.70, 95% CI 0.51-0.97). Furthermore, when the analysis was restricted to patients with ampullary cancer, those treated with gemcitabine had a median survival that was almost twice as long as those in the observation group (median 71 versus 41 months); the median survival in the FU/leucovorin group was 57.8 months. Post hoc analysis did not show differential treatment responsiveness based on histologic subtype.

There was no difference in overall survival between the chemotherapy arms, but grade 3 or 4 stomatitis (11 versus 0 percent) and diarrhea (14 versus 4 percent) were significantly more common with leucovorin-modulated FU, and there were more serious adverse effects in the FU group as well.

●The only other randomized trial that addressed the benefit of adjuvant chemotherapy was a multicenter randomized trial from Japan that compared surgery with and without postoperative chemotherapy (two courses of mitomycin plus infusional FU, followed by prolonged oral administration of FU until tumor progression) in 508 patients with pancreaticobiliary tract cancer (56 with ampullary cancer) [104]. A significant survival benefit for adjuvant chemotherapy was seen in the patients with gallbladder cancer (five-year survival 26 versus 14 percent), but not in those with ampullary cancer (five-year survival 28 versus 34 percent in the chemotherapy and control groups, respectively).

●As noted above, a meta-analysis of 14 studies of adjuvant therapy in periampullary adenocarcinoma (including the two randomized trials discussed above) concluded that there was no survival benefit for any adjuvant strategy (adjuvant chemoradiotherapy or chemotherapy) for treatment of periampullary cancers; however, the benefits of adjuvant chemotherapy were not analyzed separately from those of chemoradiotherapy [101]. Also, as noted above, there were few T3 or T4 tumors in the analysis, and the data are most compelling for benefit of adjuvant therapy in this subgroup.

POST-TREATMENT SURVEILLANCE — Post-treatment surveillance to detect recurrent or persistent disease is performed at regular intervals, although the optimal surveillance strategy is undefined [41]. National Comprehensive Cancer Network (NCCN) guidelines are not available. Many clinicians follow patients every six months for five years and annually thereafter. Follow-up visits usually include history and clinical examination, and serum tumor markers (typically CEA and CA 19-9). The utility of periodic computed tomography (CT) scan of the abdomen is unclear.

While the need for endoscopic surveillance is universally accepted, most published recommendations are similar to those reported after local resection of ampullary adenomas. A reasonable approach is surveillance endoscopy every six months for two years, then annually for an additional three to five years. (See "Ampullary adenomas: Management".)

CHEMOTHERAPY FOR ADVANCED DISEASE

Choice of therapy — Limited data exist to guide physicians in the choice of chemotherapy, largely because of the rarity of this disease. Much of the data on chemotherapy for advanced ampullary cancer are in combined series that include patients with small bowel, pancreatic, and ampullary adenocarcinomas, or more commonly, ampullary plus biliary tract cancers. However, particularly because of advances in treatment of advanced disease, the distinction as to a periampullary tumor is of intestinal, biliary, or pancreatic origin is particularly important for the selection of the treatment approach. (See "Treatment of small bowel neoplasms", section on 'Metastatic disease' and "Systemic therapy for advanced cholangiocarcinoma" and "Initial systemic chemotherapy for metastatic exocrine pancreatic cancer".)

For true ampullary cancers, there is no consensus on the best management for patients with metastatic disease. Patients with ampullary cancer were included in the ABC trial of gemcitabine with and without cisplatin, although they represented a small minority [105]. There was a significant survival advantage to the combination both in terms of progression-free and overall survival, compared with gemcitabine alone. Many consider this to represent the standard regimen for advanced ampullary as well as biliary tract cancers. (See "Systemic therapy for advanced cholangiocarcinoma" and "Treatment of advanced, unresectable gallbladder cancer", section on 'Palliative systemic chemotherapy'.)

However, others disagree, treating these patients more like those with advanced pancreatic cancer or small bowel adenocarcinoma [106]. (See "Initial systemic chemotherapy for metastatic exocrine pancreatic cancer" and "Treatment of small bowel neoplasms", section on 'Metastatic disease'.)

Importance of somatic and germline genomic testing — All patients who might be candidates for systemic targeted therapy should undergo next generation sequencing of tumor tissue, and be referred for germline genomic testing. DNA mismatch repair deficiency (dMMR)/ high levels of microsatellite instability (MSI-H), and other potentially actionable germline and somatic (tumoral) genetic alterations found in ampullary cancers more often than in pancreatic or biliary tract cancers [107-110]. These molecular alterations may identify patients for whom molecularly targeted therapy (eg, pembrolizumab for dMMR/MSI-H) may be an option. (See "Tissue-agnostic cancer therapy: DNA mismatch repair deficiency, tumor mutational burden, and response to immune checkpoint blockade in solid tumors".)

SUMMARY AND RECOMMENDATIONS

●Surgical versus nonsurgical treatment

•We recommend pancreaticoduodenectomy rather than local resection for most patients with invasive ampullary carcinomas (Grade 1B). (See 'Pancreaticoduodenectomy' above.)

•We suggest local ampullary excision rather than pancreaticoduodenectomy for patients with noninvasive ampullary tumors (pTis (table 1)) (Grade 2B).

Ampullectomy is also a reasonable approach for poor surgical candidates who have a well-differentiated T1 tumor (table 1) that is <6 mm in size (based on endoscopic ultrasound [EUS]). However, a more aggressive surgical approach is preferred for patients who are candidates for pancreaticoduodenectomy because of better outcomes. (See 'Local resection' above.)

•Nonsurgical treatment modalities (ie, endoscopic snare resection, laser ablation, photodynamic therapy) provide palliative rather than curative benefit for patients with ampullary carcinoma. These methods should be restricted to patients who are not operative candidates and those who refuse surgery. (See 'Minimally invasive nonsurgical therapies' above.)

●Adjuvant therapy – We offer adjuvant therapy to all patients with resected ampullary cancer stage T2N0 or higher (table 1). (See 'Adjuvant therapy' above.)

•For patients with an excellent performance status who are able to tolerate it, we suggest modified FOLFIRINOX (oxaliplatin plus irinotecan with leucovorin and short-term infusional fluorouracil [FU] (table 3)), rather than a gemcitabine-based regimen (Grade 2C). (See "Treatment for potentially resectable exocrine pancreatic cancer", section on 'Modern combination regimens (FOLFIRINOX and gem-nabpaclitaxel)'.)

For less fit patients, gemcitabine plus capecitabine (table 4) or gemcitabine plus nabpaclitaxel (table 6) are reasonable alternatives. (See "Treatment for potentially resectable exocrine pancreatic cancer", section on 'Other gemcitabine and FU-based approaches'.)

We reserve gemcitabine alone (table 5) or, where available, S-1 alone for patients with a borderline performance status or a comorbidity profile that precludes combination chemotherapy.

When chemoradiotherapy is not given, we suggest six months of adjuvant chemotherapy.

•We also suggest adding concurrent chemoradiotherapy to systemic chemotherapy for patients with resected ampullary cancers stage T2N0 or higher (table 1), rather than chemotherapy alone (Grade 2C). (See 'Our recommended approach' above.)

During the concurrent chemoradiotherapy portion, we prefer infusional FU, but oral capecitabine is an acceptable alternative.

The optimal way to sequence FU-based chemoradiotherapy and systemic chemotherapy is unclear. Our preferred approach is to administer all of the chemotherapy initially. For patients who have no evidence of metastatic disease, we recommend four months of systemic chemotherapy, followed by six weeks of chemoradiotherapy using concurrent infusional FU (250 mg/m² daily) beginning two to four weeks after finishing chemotherapy. This is the same approach we use for resected pancreatic cancer. (See 'Adjuvant therapy' above.)

●Patients with metastatic disease

•The optimal regimen for systemic therapy of true ampullary tumors has not been established. Based on the results of the ABC trial, the combination of gemcitabine plus cisplatin is a reasonable approach for patients who are able to tolerate it. Others use single-agent gemcitabine initially (table 7) followed by an oxaliplatin-based regimen, or vice versa, in a manner similar to treatment of pancreatic cancer. (See 'Chemotherapy for advanced disease' above.)

•All patients who might be candidates for systemic targeted therapy should undergo somatic and germline genomic testing for potentially actionable molecular alterations. DNA mismatch repair deficiency (dMMR)/ high levels of microsatellite instability (MSI-H), and other potentially actionable germline and somatic (tumoral) genetic alterations are more frequently found in ampullary cancers than in pancreatic or biliary neoplasms, and they may identify patients for whom molecularly targeted therapy (eg, pembrolizumab for dMMR/MSI-H) may be an option. (See 'Importance of somatic and germline genomic testing' above.)

ACKNOWLEDGMENT — The UpToDate editorial staff acknowledges David P Ryan, MD, who contributed to earlier versions of this topic review.