Introduction

Primary adenocarcinoma of the duodenum is a rare condition accounting between 0.3 and 0.5 % of all gastrointestinal malignancies [1]. Despite the duodenum representing <10 % of the total length of the small bowel, it is the site of between 25 and 45 % of small bowel cancers [2]. Cancer of the duodenum appears to be more frequent in the periampullary area, but has also been shown to arise in the third and fourth duodenal portions [3]. Since the symptoms are nonspecific and similar to other benign conditions, the diagnosis is often difficult and delayed. Nonetheless, duodenal carcinoma has a reported 5-year survival rate for resected tumors between 25 and 75 % [4–12], which is significantly better than that for cancer of the ampulla of Vater (46 %) and the head of pancreas (10 %) [13, 14]. In the last decade, different studies evaluated the correlations between clinical, pathological, and treatment variables in order to identify specific prognostic factors associated with survival [10–12, 15–20]. These studies showed that sex, age, tumor size and location, grade, stage, resection margins, weight loss, and white blood cell counts can influence survival significantly. However, due to a low incidence of the disease in the general population, debate on the prognostic factors in duodenal adenocarcinoma is still open. Therefore, questions have been raised especially on the prognostic significance of nodal status and of the absolute number and ratio of the involved lymph nodes [17], and a recent report established perineural invasion as another important independent prognostic factor [10]. The purposes of the present study were to evaluate the clinicopathologic features of patients with duodenal adenocarcinoma observed over a 9-year period at our institution and to analyze the main prognostic factors after potentially curative pancreaticoduodenectomy (PD).

Patients and methods

This study was approved by the Institutional Review Board. The clinical records of all patients diagnosed with primary duodenal adenocarcinoma and referred to the unit of General Surgery B, University of Verona, between January 2000 and December 2009 were reviewed. The diagnosis was confirmed by a histological examination of all resected specimens or by endoscopic biopsies. Patients with adenoma and malignant neoplasms other than adenocarcinoma were excluded. Demographic and clinical details and surgical and pathological data were collected from the patients’ notes. Pathological data included tumor location, size, grade, and stage. Tumors arising in D2 within 2 cm of the ampulla of Vater were defined as periampullary duodenal adenocarcinomas; all other tumors were classified as extra-ampullary. Pathological staging of duodenal adenocarcinoma was based on the American Joint Committee on Cancer (AJCC) system, 7th edition [21]. A tumor comprising two different degrees of differentiation was recorded as the category of poorer differentiation. Lymph node ratio (LNR) was calculated dividing the total number of lymph nodes harboring metastasis by the total number of resected nodes; cutoff points were chosen according to the existing literature [10]. Additional collected parameters included perineural invasion (defined as tumors cells within any layers of the nerve sheath or the perineural space) and the presence of lymph-vascular invasion (defined as microscopic lymphatic invasion, vascular invasion, or both). A curative resection was defined as a microscopically negative resection with no gross evidence of residual disease (R0 resection). A resection with curative intent was ultimately defined as a palliative intervention if the margin was microscopically (R1) or macroscopically (R2) positive. Palliative surgery also included biliary and gastrointestinal bypass. Postoperative morbidity and mortality were defined as the occurrence of complications and death within 30 days of the operation, respectively. Postoperative pancreatic fistula (PF) was defined according to the International Study Group on Pancreatic Fistula [22]. Follow-up information was obtained from our electronic database or through telephone interviews.

Statistical analysis

Data were analyzed using the SPSS software (version 19.0; SPSS, an IBM company, Chicago, IL, USA). Distribution of continuous variables was reported as median and range, and Mann–Whitney U test was used to compare medians. Categorical variables were presented as numbers and percentages; chi-squared test (with Yates continuity correction in 2 × 2 contingency tables) was used for statistical comparison. Fisher’s exact test was used when appropriate. All tests were two-sided. Overall survival time was calculated from the date of operation/pancreatic resection to the date of last follow-up/death. Cumulative event rates were calculated using the method of Kaplan and Meier. Univariate analyses were performed using the log-rank test to compare differences between categorical groups. A Cox proportional hazards model was developed using relevant clinicopathologic variables in a direct enter fashion (univariate inclusion criteria of p < 0.150) to determine the association of each with overall survival. Values of p were presented with hazard ratios and 95% confidence intervals. For univariate and multivariate analyses, the variable tumor size, resection margins, tumor grade, and stage were dichotomized as appropriate (≤3 versus >3 cm, R0 versus R1–R2, G1 versus G2–G3, and stages I–II versus stages III–IV). Statistical significance was determined by a p value of <0.05.

Results

Clinical and surgical findings

The study population consisted of 37 patients; their demographic characteristics are summarized in Table 1. Only one patient (2.5 %) developed duodenal adenocarcinoma following colectomy for familial adenomatous polyposis. Details of the 34 patients who underwent surgery are outlined in Table 2. Twenty-five patients (68 %) underwent potentially curative resection. These were 14 men and 13 women, with a median age of 54 years (range, 38–83 years). Abdominal pain (72 %) and jaundice (36 %) were the most common presenting symptoms in this subgroup. Fifteen patients (60 %) had tumors in the periampullary duodenum, while ten (40 %) patients had extra-ampullary tumors all located in the third portion. Eight patients (21.6 %) underwent palliative double bypass (biliary and gastric); their most frequent initial symptoms were weight loss (75 %) and abdominal pain with vomiting (62.5 %). One had a large tumor (80 mm) infiltrating peripancreatic vessels (T4N0M0, stage II), four were stage III, and three were metastatic (stage IV). Among patients with distant metastases, the liver and the lung were the most common sites (12 %). Four patients (10.8 %) received a non-surgical palliation, which consisted of chemotherapy with supportive treatment. One of them had previously undergone an exploratory laparotomy. Overall postoperative morbidity was 52.9 %, with a rate of 56 % in the curative resection group and 50 % in the bypass group. In the curative resection group, the most common postoperative complication was PF, which occurred in four patients (16 %). Three patients (12.0 %) had an abdominal collection; the other three patients (12.0 %) had a post-pancreatectomy hemorrhage. Pulmonary complications occurred in eight patients (32.0 %). In the bypass group, one or more complications developed in four patients (50 %), with chest infection appearing to be the most common complication (25 %). The median length of stay for the curative resection patients was 14 days (range, 9–57 days), while it was 11.5 days (range, 8–27 days) after double bypass. Postoperative mortality was nil. The frequency of adjuvant chemotherapy in resected patients did not differ between those with or without complications (50 versus 54 %, p = 1). Patients with complications were referred to the oncologist later [31.0 days (range, 20–82) versus 22.0 days (range, 14–61)], but this difference was not significant (p = 0.694).

Table 1 Demographic and clinical details of patients with duodenal adenocarcinoma
Full size table
Table 2 Surgical details of patients with duodenal adenocarcinoma who underwent surgery
Full size table

Pathological findings

This analysis was restricted to patients who underwent a potentially curative resection. Five patients (20 %) were stage I, 6 (24 %) were stage II, 11 (44 %) were stage III, and 3 (12 %) were stage IV. Among the three metastatic patients, one had a single lung metastasis, one had a single liver metastasis treated with concomitant wedge resection, and one had a proximal jejunal loop metastasis, which was excised with the PD specimen. Pathological data are summarized in Table 3. Seventeen of the curative resection were histopathologically confirmed as R0 (68 %), six were R1 (24 %), and only two resections (8 %) were reported as R2. Nodal metastases (pN1) were identified in 24 patients (64.8 %). The median number of lymph nodes retrieved was 19 (range, 5–63), while the median number of lymph nodes excised from the pN1 patient group was 22 (range, 10–61).

Table 3 Pathologic details of patients with duodenal adenocarcinoma who underwent potentially curative resection
Full size table

Survival analysis and prognostic factors

The overall median follow-up was 25 months (range, 5–180 months). Thirteen patients (35.1 %) had no evidence of disease, and these included three patients who are still alive for at least 5 years. Two patients (5.4 %) died of an unrelated cause. Eight patients (21.6 %) were alive with recurrent or stable disease, while 14 patients (37.8 %) died of recurrent disease; of these, only one patient survived more than 5 years. The median survival of the overall population (N = 37) was 70 months (95% CI = 41.7–98.2); the 5-year survival rate was 58.5 %. The median survival time for patients undergoing curative (R0) surgery was significantly longer (180 months, 95% CI = NA) than those undergoing palliative (R1/R2/bypass) surgery (35 months, 95% CI = 11.9–58.0), with 5-years survival rates of 76.6 and 35.4 %, respectively (p = 0.013, Fig. 1). Univariate and multivariate analyses were restricted to the 25 patients who underwent potentially curative resections (Table 4). Median follow-up in this subgroup was 27 months (range, 6–180 months), with a median survival of 70 months (95% CI = 14.0–125.9) and a 5-year survival rate of 71.1 %. Univariate analysis showed tumor grade (p = 0.05) and the occurrence of postoperative and of abdominal complications (p = 0.05 and p = 0.013, respectively) to be significantly associated with survival (Figs. 2, 3, and 4). In particular, the median overall survival was 180 months (95% CI = NA) in patients with G1 tumors and 70 months (95% CI = 50.4–89.5) in patients with G2–G3 tumors. The 5-year overall survival was 100 % for G1 tumors and 61.6 % for G2–G3 tumors. The median overall survival was 180 months (95% CI = NA) in patients with an uneventful postoperative course and 70 months (95% CI = 10.7–129.2) in patients with a complicated course. The 5-year overall survival was 100 % for an uneventful postoperative course and 53.3 % for a complicated postoperative course. The median overall survival was 180 months (95% CI = NA) in patients who did not develop abdominal complications and 52 months (95% CI = 22.5–129.4) in patients who developed abdominal complications. The 5-year overall survival was 100 % in the absence of abdominal complications and 60 % in patients who developed abdominal postoperative morbidity. The best regression model resulted in the same three variables being independently associated with survival. Hazard ratios were 1.345 (95% CI = 1.28–1.91, p = 0.03) for tumor grade, 1.781 (95% CI = 1.10–2.89, p = 0.037) for the development of postoperative complications, and 1.878 (95% CI = 1.21–3.08, p = 0.029) for the development of abdominal complications.

Fig. 1
figure 1

Kaplan–Meier survival curves of patients undergoing radical surgery (R0 pancreaticoduodenectomy, n = 17) versus palliative surgery (R1/R2 pancreaticoduodenectomy or double bypass, n = 16)

Full size image
Table 4 Variables with a potential influence on survival in patients undergoing resection for duodenal adenocarcinoma
Full size table
Fig. 2
figure 2

Kaplan–Meier survival curves comparing patients within the resection group by tumor grade (G1 versus G2–G3)

Full size image
Fig. 3
figure 3

Kaplan–Meier survival curves comparing patients within the resection group by postoperative recovery (uneventful versus complicated)

Full size image
Fig. 4
figure 4

Kaplan–Meier survival curves comparing patients within the resection group by abdominal complications (Yes versus No)

Full size image

Discussion

The debate regarding the prognostic indicators of primary duodenal adenocarcinoma is still open, likely because of the very low incidence of this malignancy and the small number of studies published in the last two decades [6–20, 23–28]. In the present paper, we retrospectively evaluated the clinical features of patients with primary duodenal adenocarcinoma observed over a 9-year period at our institution and addressed the prognostic relevance of different surgical and pathological factors after potentially curative PD.

At a median follow-up of 25 months, the overall survival of the study population, which included patients undergoing PD, double bypass, or best supportive care, was 70 months. The 5-year survival rate was 58.5 %. As expected, overall survival was significantly longer after R0 resections (180 months) when compared with palliative surgery (R1/R2) and bypass operations (35 months). The survival rate after curative surgery was substantially greater than in the study by Zhang et al. [11], who reported a median survival of 45 months, with a 5-year survival rate of 49.3 %. In the study by the Massachusetts General Hospital Group, the median survival of resected patients was 44 months (88 % of R0 resections), with a 5-year survival rate of 42 % [10], whereas in the study by the Mayo Clinic Group, the median survival was 38 months, with a 5-year survival rate of 43 % [20]. A number of factors have been considered to be of prognostic value after potentially curative resection, including sex, age, tumor size and location, grade, stage, resection margins status, LNR, perineural invasion, weight loss, and biochemical parameters (white blood cells, alanine aminotransferase) [6–20, 23–28]. However, not all these findings were consistent among different reports. The first factor which, in our study, was found to be related to survival on multivariate analysis was histological grade, with a hazard ratio of 1.345 and a p value of 0.03. In particular, the 5-year overall survival was 100 % in well-differentiated tumors (G1) and 61.6 % in moderately to poorly differentiated tumors (G2–G3). Data from other gastrointestinal malignancies generally support the relationship between the degree of differentiation and survival, but only a few other papers confirmed this concept in primary duodenal adenocarcinoma, both in the resection specimen and endoscopic biopsy [11, 23].

The most relevant finding of the present analysis was that postoperative complications impact significantly on survival. The 5-year overall survival was 100 % in patients with an uneventful postoperative course and 53.3 % in patients who experienced complications (hazard ratio on multivariate analysis of 1.781, p = 0.037). Abdominal complications being the most relevant after pancreaticoduodenectomy, they were analyzed separately and found to be likewise associated with survival (hazard ratio on multivariate analysis of 1.878, p = 0.029). The most frequent abdominal complication was PF, with an incidence of 16 %, followed by postoperative bleeding (8.8 %). According to the survival curves shown in Figs. 3 and 4, the four patients who died within the first 20 months had postoperative complications. In particular, two had a PF (together with pulmonary infection) and two a post-pancreatectomy hemorrhage. All these deaths were due to disease, and three patients out of four did not receive any adjuvant treatment.

The influence of postoperative morbidity on survival after tumor resection has been reported for patients with esophageal and colorectal cancer [29–33]. In this study, the specific mechanism through which postoperative complications impact on long-term results can only be speculated. Postoperative complications can induce a systemic inflammatory response which may affect tumor growth and influence the survival [34]. The degree of host inflammatory cell activity can be measured indirectly by markers related to inflammatory response. A high systemic level of C-reactive protein has been associated with poor survival in patients undergoing potentially curative resection for pancreatic and colorectal cancers; similarly, an elevated neutrophil/lymphocyte ratio has been shown to influence the disease recurrence rate in esophageal and colorectal cancers [35–37]. However, given the retrospective nature of this analysis, these markers of inflammatory response could not be evaluated. A recent study has also shown that a prolonged period of immunosuppression and angiogenic stimulation caused by intra- and postoperative blood transfusion was independently associated with earlier cancer recurrence and reduced survival after PD for pancreatic ductal adenocarcinoma [38]. However, we did not find any correlation between blood transfusions and survival.

Another potential cause of worse survival in patients with postoperative morbidity and prolonged hospital stay may be a delay in adjuvant therapy. Other papers have reported an improved median survival for patients with adjuvant therapy [39], but our series was unable to demonstrate a benefit. In particular, the frequency of adjuvant therapy was not different between patients with or without postoperative complications. Despite patients who did not experience complications being referred to the oncologist earlier, this did not seem to affect survival. These results, however, have to be interpreted with caution, and larger studies are necessary to clarify the role of adjuvant therapy for primary duodenal adenocarcinoma.

Controversy exists as to whether there is a prognostic difference in patients with periampullary versus extra-ampullary duodenal adenocarcinomas. In the present analysis, periampullary tumors occurred more frequently (67.6 %), similar to other reports [11, 20]. In the study by the Mayo Clinic Group [20], these two tumor variants were found to have similar survival after resection, and our results support this concept. In the above referenced study, survival for distal tumors was improved by curative PD without being compromised by segmental duodenal resection, but the number of lymph nodes sampled was significantly less with segmental resection. Segmental resection for tumors located at the third and fourth duodenal portions has also been proposed by other authors who found similar survival rates after these two procedures [24]. At our institution, PD is routinely employed even for cancers located in the distal duodenum because, in our opinion, it would satisfy the oncologic principles of adequate curative resection by a concomitant en bloc lymphadenectomy.

Contrary to different other reports [11, 14–17, 19], nodal status and LNR did not show a significant association with decreased survival. The median number of lymph node resected was 19, which seems adequate. Sarela et al. [17] demonstrated in this regard that after resection of duodenal adenocarcinoma, examination of 15 or more lymph nodes can improve diagnostic discrimination by the pN category. Furthermore, these findings may suggest that lymph node involvement should not preclude an aggressive surgical resection as a potential curative treatment.

Resection margin status, which is generally believed to be critical to survival in duodenal adenocarcinoma, did not result to be a significant prognostic factor in the resected population. A recent large single-institution study of 169 patients from the Massachusetts General Hospital reported perineural invasion (which was present in 40 % of resection specimens) to be the strongest independent predictor of recurrence and overall survival. LNR and tumor size failed to stratify prognosis [10]. However, the present study did not show a significant association between perineural invasion (present in 48 % of resection specimens) and overall survival. Similarly, we demonstrated no difference in survival based on tumor size and lymph-vascular invasion.

As with any retrospective study, this analysis has certain drawbacks. A lack of defined criteria for determining the operative or chemotherapeutic modalities was a major limitation. Moreover, given the small number of patients included in the study, factors which did not result to be associated with survival in this study may turn as relevant with a greater sample size. With this in mind, we remark that postoperative morbidity may be an additional parameter influencing long-term outcomes after potentially curative PD for primary duodenal adenocarcinoma. Despite substantial advancements in surgical techniques, this operation is still demanding and should be performed in high-volume centers, where an appropriate postoperative care can be delivered.