This combined series of patients with ampullary carcinoma represents the largest study to date that demonstrates an overall survival benefit in patients receiving adjuvant chemoradiation following surgical resection when controlling for adverse prognostic factors. After adjusting for institution, patient demographics such as age and gender, and disease characteristics such as tumor stage, tumor size, nodal involvement, and histology, patients treated with adjuvant CRT experienced enhanced survival (HR = 0.41, 95% CI: 0.25-0.67, p < 0.001). This series also confirms improved outcomes in patients with ampullary carcinoma when compared with pancreatic cancer, with a median survival of 39.9 months and two and five-year survival rates of 62.4% and 39.1% respectively.
Pancreaticoduodenectomy is the preferred surgical approach for carcinoma of the ampulla of Vater that is amenable to resection . However, similar to pancreatic cancer, the role of post-operative adjuvant therapy remains undefined. While prognosis for resectable ampullary carcinoma is considerably better than for pancreatic cancer, patients with node positive disease have poor survival and appear to benefit from adjuvant therapy [20, 21]. A number of reports, mostly consisting of single institution series, have established adverse prognostic factors, including extent of local invasion, status of surgical margins, presence of nodal metastasis, and histologic grade, all of which predict for overall survival as well as local and distant disease [5–9, 22–29]. In these cohorts, nodal involvement has been a particularly strong predictor of poor outcomes, with 5-year survival rates following PD ranging from 64-80% in patients with node-negative disease and 17-50% in patients with node-positive disease. A more recent population-based analysis of roughly 4,000 patients with ampullary carcinoma was conducted using the National Cancer Institute's Surveillance, Epidemiology, and End Results database . Outcomes were slightly worse than the aforementioned series from specialized cancer centers but were highly dependent on nodal metastasis (5 year survival: 47.6% vs. 21.0%).
High rates of relapse along with identification of adverse prognostic factors have led to exploration of adjuvant chemoradiation for "high risk" ampullary carcinoma, although the literature in this area remains sparse. Willett et al. first reported a trend towards improved local control with no improvement in overall survival when adjuvant 5-FU based chemoradiation was given to a small cohort of patients with high risk features, defined as invasion of the pancreas, nodal metastasis, positive margins, or poor histology . A subsequent review by Mehta et al. reported a favorable 3-year actuarial survival rate of 44% using adjuvant 5-FU based chemoradiation in patients with large tumor size, nodal involvement, positive surgical margins, poor histology, or neurovascular invasion . Similarly, Lee et al. achieved superior disease-free survival in patients with advanced tumor stage (T3/T4) or positive nodes receiving adjuvant chemoradiation . On multivariate analysis, adjuvant therapy was also a significantly favorable factor for the entire cohort (HR: 0.16, p = 0.030). However, less than twenty patients received adjuvant therapy in each of these studies, making it difficult to derive convincing conclusions.
More recently, three retrospective studies from institutions that treat high volumes of periampullary malignancies reviewed their experience with adjuvant CRT for ampullary carcinoma. Krishnan et al. examined 96 patients, 54 of whom had received adjuvant CRT consisting of either preoperative radiation to a median dose of 45 Gy or postoperative radiation to a median dose of 50.4 Gy, with concurrent 5-FU or capecitabine . Patients with advanced T-stage (T3/T4) who were treated with CRT showed a borderline significant increase in survival (mOS: 35.2 vs. 16.5 months, p = 0.06). Similarly, a JHH review of 111 patients identified a trend towards improved survival with adjuvant CRT among those patients with nodal metastasis (mOS: 30.0 vs. 21.6 months, p = 0.092) . Postoperative therapy in this study consisted of a median radiation dose of 50.4 to the tumor bed and regional nodes with concurrent 5-FU or capecitabine. Furthermore, a statistically significant difference in survival among patients with lymph node involvement treated with adjuvant CRT was found in a previous study from the Mayo Clinic (mOS: 3.4 vs. 1.6 years, p = 0.02) . Note, however, that in none of these studies was adjuvant CRT associated with increased survival on multivariate analysis.
In the present series, adjuvant therapy was not found on univariate analysis to be associated with increased survival when compared to surgery (mOS: 39.9 vs. 40.1 months, p-0.839), as summarized in Table 2 and illustrated in Figure 1. This lack of survival benefit is likely a result of the imbalance in adverse prognostic factors between treatment groups. In this series, nodal metastasis (p < 0.001), advanced T stage (p = 0.002), and poorly differentiated histology (p = 0.011) were all significantly associated with decreased survival. While margin status was not a predictor of survival (p = 0.493), margin status is widely considered to be a poor prognostic factor, and its lack of association with survival in this study may be attributable to the small number of patients with close or positive margins (n = 3). Regardless, the cohort that received CRT had a significantly higher proportion of patients with advanced T-stage (p = 0.002), pathologic lymph node involvement (p < 0.001), and positive surgical margins (p = 0.019), and a borderline statistically significant trend towards poorer histologic grade (p = 0.053). The fact that survival was comparable between treatment groups despite these discrepancies suggests the potential benefit of adjuvant CRT, particularly amongst high risk populations.
Moreover, when baseline demographic and treatment-related characteristics were adjusted for on multivariate analysis, a significant association between adjuvant therapy and improved survival appeared. Indeed, this study represents the second reported survival benefit from adjuvant CRT found on multivariate analysis, albeit with a much larger cohort than was analyzed in the aforementioned study by Lee et al . Interestingly, when patients were stratified by baseline demographic and disease-related characteristics, no subgroup showed a significant survival benefit from adjuvant CRT except for patients with nodal metastasis, who experienced a large difference in median survival (mOS: 32.1 vs. 17.5 months, p = 0.004). As suggested in multiple previous studies, node-positive patients were found to carry a very poor prognosis on both univariate analysis (p < 0.001) and multivariate analysis (p < 0.001), with a median survival of only 18.4 months. The fact that node-positive patients who were not treated with adjuvant CRT showed a dismal 5-year survival rate of only 5.9% indicates that this group may be particularly suited for post-operative therapy. Moreover, while the effect of adjuvant therapy in node negative patients did not reach statistical significance, the absolute difference in survival (mOS: 103.2 vs. 61.6 months) is noteworthy and reminiscent of the CONKO-001 trial in which node negative pancreatic cancer patients experienced superior survival with adjuvant chemotherapy .
While good local control was achieved in this study, nearly a third of patients suffered from distant relapse, and roughly 90% of recurrences were attributable in part to metastatic disease. Consistent with the literature, the most common sites of metastasis were the liver and peritoneum. Overall, progression of disease led to more than half of the deaths in the cohort, with nearly one third of patients who died harboring disease in the liver. The prevalence of metastatic disease suggests the need for more effective systemic therapy, particularly in high risk patients. Unfortunately, there is even less information regarding appropriate type and duration of chemotherapeutic agents when incorporated with radiation for ampullary cancer. Furthermore, the role of adjuvant chemotherapy alone is an area that has been largely understudied, a remnant of borrowed U.S. practice patterns supporting adjuvant CRT for resected pancreatic cancer. A Japanese study of adjuvant mitomycin C and 5-FU for pancreaticobiliary carcinomas found no overall or disease-free survival benefit in a subset of 24 patients with ampullary carcinoma when compared to surgery alone . More recent and robust results from the European Study Group for Pancreatic Cancer (ESPAC) - 3(v2) trial also showed no difference in survival in 304 patients with resected ampullary cancer who were randomized to 5-FU/folinic acid, gemcitabine, or observation . Combination chemotherapy may provide better results, as a randomized control trial comparing gemcitabine and cisplatin versus gemcitabine alone in 410 patients with locally advanced or metastatic biliary or ampullary cancers did show superior survival with the combination regimen (mOS: 11.7 vs. 8.1 months, p < 0.001), although it should be acknowledged that only 5% of tumors in the study had an ampullary origin . Of note, no study has directly compared adjuvant chemotherapy alone with adjuvant chemoradiation.
Given the retrospective nature of this study and wide time period over which this study spanned, our findings are limited by the variability in treatment regimens and the potentially unequal distribution of confounding factors in patient selection between treatment groups. Certainly, patients in our study were subject to different operative methods, radiation plans, and chemotherapeutic agents, which we were unable to control for due to incomplete information or insufficient power. Additionally, this study has shown that patients selected for adjuvant CRT at the Johns Hopkins Hospital and the Mayo Clinic possessed more adverse prognostic factors than those treated with surgery alone. While several high risk characteristics were adjusted for in our analysis, other variables that were not taken into account include performance status and weight loss, both of which may be correlated with disease outcomes. Since patients who received adjuvant CRT were significantly younger, it is easy to imagine that healthier patients were more likely offered adjuvant treatment. The retrospective nature of our study may have also compromised our ability to accurately capture certain information such as the toxicity data, which was lower when compared to prior experience . Furthermore, variations in institutional protocols regarding treatment delivery can be a source of bias in studies analyzing data from multiple sites, but it should be noted that while the distribution of treatment type did in fact vary by institution, there was no association between institution and survival on univariate or multivariate analysis, and institution did not affect outcomes when stratified by treatment type. Another limitation was the number of patients excluded for either missing data (i.e. stage or nodal status) or because they were lost to follow-up. It is probable that follow-up was not consistent among treatment groups, with patients receiving adjuvant therapy likely showing better follow-up. The number of patients lost to follow along with the number of patients alive at time of analysis resulted in a low number of documented recurrences. Ideally, we would have been able to examine the association between adjuvant therapy and patterns of recurrence, but the low number of recurrences prevented the possibility of meaningful analysis. Nevertheless, this study combines the experience of two high volume institutions to allow for the largest series to date that has examined the role of adjuvant therapy following surgery for ampullary cancer.