Association of Histopathologic Phenotype of Periampullary Adenocarcinomas With Survival

Key Points

Question  Should periampullary adenocarcinomas be classified by histopathologic phenotype (intestinal vs pancreaticobiliary) rather than anatomic location?

Findings  In this analysis of medical records of 510 patients who underwent pancreatoduodenectomy for periampullary adenocarcinoma, histopathologic phenotype was independently associated with overall survival, while anatomic location was not. Those with pancreaticobiliary-type cancers of the duodenum, ampulla, or distal common bile duct had pathologic features and survival similar to those with pancreatic ductal adenocarcinoma.

Meaning  Classifying periampullary adenocarcinomas by histopathologic phenotype rather than anatomic location more accurately estimates prognosis and may better guide adjuvant treatment.

Importance  Patients with periampullary adenocarcinomas have widely variable survival. These cancers are traditionally categorized by their anatomic location of origin, namely, the duodenum, ampulla, distal common bile duct (CBD), or head of the pancreas. However, they can be alternatively subdivided histopathologically into intestinal or pancreaticobiliary (PB) types, which may more accurately estimate prognosis.

Objectives  To identify factors associated with survival in patients with periampullary adenocarcinomas and to compare survival between those having intestinal-type or PB-type cancers originating from the duodenum, ampulla, or distal CBD with those having pancreatic ductal adenocarcinoma (PDAC).

Design, Setting, and Participants  This study was a retrospective analysis of medical records in a prospectively maintained database. Three pathologists separately evaluated histopathologic phenotypes at a university-based tertiary referral center. Study participants were all patients (N = 510) who underwent pancreatoduodenectomy for adenocarcinoma between January 1995 and December 2014.

Main Outcome and Measure  Overall survival.

Results  This study identified 510 patients (mean [SD] age, 66.1 [10.9] years; 245 female [48%]) who underwent pancreatoduodenectomy for adenocarcinomas: 13 duodenal, 110 ampullary, 43 distal CBD, and 344 PDAC. The median overall survival was 61.2 (interquartile range [IQR], 22.0-111.0), 70.4 (IQR, 26.7-147.7), 40.6 (IQR, 15.2-59.6), and 31.4 (IQR, 17.3-86.3) months for patients with cancers of the duodenum, ampulla, distal CBD, or pancreas, respectively (P = .01), indicating a significant difference between the 4 tumor anatomic locations. Most duodenal (61.5% [8 of 13]) and ampullary (51.8% [57 of 110]) cancers were intestinal type, and most distal CBD tumors were PB type (86.0% [37 of 43]). Those with intestinal-type duodenal, ampullary, or distal CBD adenocarcinomas had longer median overall survival than those with PB type (71.7 vs 33.3 months, P = .02) or PDAC (31.4 months, P = .003). There was no survival difference between PB-type cancers and PDAC (33.3 vs 31.4 months, P = .66). On multivariable analysis, histologic grade (hazard ratio [HR], 1.98; 95% CI, 1.56-2.52; P < .001), histopathologic phenotype (HR, 1.75; 95% CI, 1.16-2.64; P = .008), and nodal status (HR, 1.45; 95% CI, 1.12-1.87; P = .05) were significantly associated with survival, while anatomic location was not.

Conclusions and Relevance  Histopathologic phenotype is a better prognosticator of survival in patients with periampullary adenocarcinomas than tumor anatomic location. Those with PB-type duodenal, ampullary, or distal CBD adenocarcinomas have survival similar to those with PDAC.

The periampullary region is a complex anatomic area consisting of the ampulla of Vater, distal common bile duct (CBD), second portion of the duodenum, and head of the pancreas. Adenocarcinomas arising in this area are now classified by their anatomic location of origin according to the scheme developed by the World Health Organization¹ and the American Joint Committee on Cancer staging.² However, considerable debate has recently ensued concerning how best to categorize these cancers.^3,4

There are several reasons for this deliberation. First, patient survival after resection of cancers of the duodenum, ampulla, and distal CBD greatly varies, even when comparing tumors arising from the same anatomic location.^5-8 Response to adjuvant therapy is less predictable.^9-11 In addition, determining the precise anatomic location of origin can be challenging¹² and is based on identifying the center of the tumor grossly or microscopically in an area where the anatomy can vary and is frequently distorted by the tumor.¹³

Given the difficulty in categorizing these cancers by their origin and the variability in outcomes seen in tumors arising from the same anatomic location, an alternative classification scheme has been developed that instead dichotomizes the adenocarcinomas by histologic differentiation, namely, intestinal-type or pancreaticobiliary (PB) type.^14,15 Intestinal-type adenocarcinomas frequently resemble colon cancer and are characterized by well-formed elongated glands lined by columnar epithelium with oval, basally oriented nuclei. On immunohistochemical staining, they stain strongly for caudal-type homeobox 2 (CDX2), but stains for mucin 1 (MUC1) are negative to weakly positive. Conversely, PB-type adenocarcinomas are characterized by glands of rounded or ductular morphologic structure lined by cuboidal or low columnar epithelium with frequently enlarged round pleomorphic nuclei; most stain strongly for MUC1, but they rarely have positive CDX2 staining (eFigure in the Supplement). Histopathologic (HP) phenotype is often easier to determine pathologically because it can frequently be determined based on hematoxylin-eosin (H&E) staining alone and may be a better indicator of tumor behavior and biological features than anatomic location.¹⁶ Indeed, this categorization is now increasingly accepted for ampullary adenocarcinomas, with multiple studies^17-19 consistently showing poorer outcomes for PB-type vs intestinal-type cancers. Furthermore, 2 reviews^20,21 recommend considering HP phenotype when selecting adjuvant treatment and planning clinical trials for ampullary cancers.

Although the role of HP phenotype is well studied for ampullary cancers, studies^22-24 addressing its prognostic value for other adenocarcinomas arising in the periampullary region have had conflicting results. In the present study, we aimed to review the frequency at the University of California, Los Angeles (UCLA), of intestinal-type and PB-type duodenal, ampullary, and distal CBD adenocarcinomas and to compare pathologic features and patient survival between the 2 HP phenotypes. In addition, unlike other cancers in this region, outcomes for patients with pancreatic ductal adenocarcinoma (PDAC) are more consistent, and well-defined guidelines exist for neoadjuvant and adjuvant treatment.²⁵ Therefore, we also sought to determine how other periampullary adenocarcinomas compare with PDAC with regard to biological features and patient outcomes.

Medical records of all patients who underwent pancreatoduodenectomy for adenocarcinoma between January 1995 and December 2014 at UCLA, were reviewed from a prospectively maintained database. Archival H&E-stained glass slides of all pathologic specimens were reviewed by 3 pathologists (E.A.S., J.K., and D.W.D.) separately to determine HP phenotypes. The classification of intestinal type, PB type, or mixed type was made based on previously established criteria.^14,15 In instances where there was not uniform consensus regarding HP phenotype, H&E-stained slides were rereviewed by all 3 pathologists in conference to reach consensus.

Other information obtained from pathology reports included tumor size, histologic grade, TNM stage, presence or absence of perineural invasion and lymphovascular invasion, and margin status. Patients were excluded from the study if the cancer anatomic location of origin could not be determined or if they had indeterminate or mixed-type tumors. In addition, given the high proportion of intestinal and gastric subtypes seen in pancreatic cancers arising within intraductal papillary mucinous neoplasms of the pancreas,²⁶ patients with these lesions were also excluded from the study.

Hospital and clinic medical records were reviewed for patient demographics, date of diagnosis, type and date of operation, use of neoadjuvant and adjuvant therapy, follow-up, and disease-specific survival. Overall survival (OS) was measured from the date of diagnosis to the date of death or last follow-up. This study was approved by the UCLA Institutional Review Board. Because this study was a retrospective review without direct patient contact, informed consent was not required.

A software program (SPSS, version 22.0; IBM) was used for statistical analysis and data management. The means of continuous variables with normal distributions were compared using the 2-tailed t test. Nonparametric continuous variables were compared using the Mann-Whitney test or Kruskal-Wallis test. The Pearson χ² test was used to analyze categorical data. Survival analysis was performed using the Kaplan-Meier method with comparisons informed by the log-rank test. Multivariable Cox proportional hazards regression analyses were performed using statistically significant univariate parameters with P < .10 as the initial entry criterion. Continuous variables were split at their median for these analyses. Statistical significance was defined as P ≤ .05.

A total of 570 patients underwent pancreatoduodenectomy during the study period. Of these, 44 with intraductal papillary mucinous neoplasm–associated pancreatic cancers and 16 with indeterminate or mixed type tumors were excluded, leaving 510 patients (mean [SD] age, 66.1 [10.9] years; 245 female [48%]) for the final analysis.

The distribution of HP phenotype at each anatomic location is summarized in Table 1. The most common tumor anatomic location was the head of the pancreas (n = 344), followed by the ampulla (n = 110). Ampullary tumors were almost evenly split between the 2 HP phenotypes (57 patients [51.8%] with intestinal type and 53 patients [48.2%] with PB type). Surprisingly, more than one-third of 13 cancers classified as originating in the duodenum were PB type (5 patients [38.5%]). Most of 43 distal CBD cancers were PB type (37 patients [86%]).

Table 2 compares the clinicopathologic factors associated with HP phenotypes. In general, intestinal-type tumors had less aggressive histologic features than PB-type adenocarcinomas and PDAC; they had significantly less nodal involvement and perineural and lymphovascular invasion, and they were of lower histologic grade. Conversely, PB-type adenocarcinomas and PDAC had similar histologic features except that pancreatic cancers tended to be somewhat larger and were more likely to show perineural invasion.

Pancreatic ductal adenocarcinoma was more likely to have positive margins (72 patients [20.9%]) than intestinal-type cancers (3 patients [4.2%]) (P = .01) and PB-type cancers (8 patients [8.4%]) (P = .02). Of 397 patients for whom details about adjuvant therapy were available, most (356 patients [89.7%]) received chemotherapy, including 31 patients with intestinal type and 58 patients with PB-type adenocarcinomas, and 267 patients with PDAC (P < .001). The P value denotes that there was a significant difference in the percentage of patients who received chemotherapy between groups (77.5% of intestinal type, 76.3% of PB, and 95.0% of PDAC).

For the entire cohort, the median OS was 34.3 months, and the median follow-up of survivors was 23.3 months. Similar to previous investigations,⁵ we found on Kaplan-Meier analysis that patients with pancreatic cancer had the shortest survival compared with patients with periampullary adenocarcinomas arising from other anatomic locations (median OS, 31.4 months; P = .01) (Figure, A). However, survival also varied greatly among patients with cancers originating from the same anatomic location, with a median OS of 61.2 months (interquartile range [IQR], 22.0-111.0 months) for the duodenum, 70.4 months (IQR, 26.7-147.7 months) for the ampulla, and 40.6 months (IQR, 15.2-59.6 months) for the distal CBD.

When patients were grouped by histologic differentiation rather than anatomic location (Figure, B), those with pancreatic cancer had survival almost identical to those with PB-type cancers (median OS, 31.4 vs 33.3 months; P = .66), and both groups had significantly shorter survival than patients with intestinal-type adenocarcinomas (median OS, 71.7 months; P = .008). Furthermore, with the exception of the duodenum, which had too few patients with PB-type cancers to determine the median OS, survival for patients with PB-type cancers was similar across the different anatomic locations (31.7 months for the ampulla and 33.0 months for the distal CBD, P = .32) (Table 3). There were also no significant differences in OS between intestinal-type tumors originating from different anatomic locations (90.4 months for the ampulla, 59.6 months for the distal CBD, and 61.2 months for the duodenum; P = .11). Survival was at least doubled for those with intestinal-type tumors compared with PB-type tumors at a given anatomic location. Given the similarity in survival for patients with PB-type tumors and PDAC, these 2 HP phenotypes were subsequently combined for further multivariable analysis.

On univariate analysis, several significant factors associated with survival were identified (Table 4). These variables included histologic grade, nodal status, perineural invasion, lymphovascular invasion, margin status, HP phenotype, and tumor anatomic location in the ampulla or pancreas. However, on multivariable analysis, only histologic grade (hazard ratio [HR], 1.98; 95% CI, 1.56-2.52; P < .001), nodal status (HR, 1.45; 95% CI, 1.12-1.87; P = .05), and HP phenotype (HR, 1.75; 95% CI, 1.16-2.64; P = .008) were significant prognostic indicators. Tumor anatomic location was no longer associated with survival on multivariable analysis.

Patients with adenocarcinomas of the duodenum, ampulla, and distal CBD have variable survival and responses to adjuvant therapy at each of the cancer anatomic locations after tumor resection.^5-11 This variability is problematic given that the current classification scheme for these cancers categorizes them by their anatomic location of origin.^1,2 These cancers can instead be grouped by their HP phenotype (intestinal or PB), which can be reliably determined via H&E staining and may provide a more accurate prognosis.

In the present study, we identified 510 patients who underwent pancreatoduodenectomy for adenocarcinomas originating from the periampullary region at UCLA, and grouped them by HP phenotype. We found that intestinal-type and PB-type tumors of the duodenum, ampulla, and distal CBD were significantly different with regard to pathologic features and patient survival, suggesting that they may represent distinct diseases. Conversely, PB-type cancers and PDAC were similar; both tended to have aggressive pathologic features and poor OS, although PDAC tumors were slightly larger and were more likely to show perineural invasion. These latter differences are likely attributable to delayed symptom onset in patients with PDAC because the tumors are generally located farther from the distal CBD than other periampullary adenocarcinomas, as well as owing to the rich parasympathetic and sympathetic nerve supply to the pancreas. Given their similarities, PB-type cancers and PDAC were combined into a single subgroup that was found to be significantly associated with worse survival (HR, 1.75; 95% CI, 1.16-2.64; P = .008) on multivariable analysis. Poor differentiation (HR, 1.98; 95% CI, 1.56-2.52; P < .001) and lymph node involvement (HR, 1.45; 95% CI, 1.12-1.87; P = .05) also were associated with shorter survival. Tumor anatomic location was not a prognostic factor. Taken together, these results suggest that grouping cancers in this area by HP phenotype rather than anatomic location better estimates survival and may be able to guide clinical decision making.

There are multiple difficulties that arise when treating patients with periampullary adenocarcinomas for which our results can help provide solutions. First, as mentioned previously, determining the precise anatomic origin is often arduous and imprecise. It is routinely done on gross or microscopic pathologic analysis by identifying where the tumor appears to be centered. Given our finding that HP phenotype is more associated with outcomes and aggressive tumor biological features than cancer location, this anatomic approach to defining tumor type may not be necessary or most appropriate. Rather, tumors arising from any of the 4 anatomic locations (duodenum, ampulla, distal CBD, or pancreas) could instead be grouped by HP phenotype, which can be performed solely by H&E staining, as done herein, or with the addition of a small battery of immunohistologic markers to further improve accuracy or differentiate morphologically challenging cases.^18,19

An additional issue with these cancers is that large randomized trials addressing the effectiveness of adjuvant treatment for nonpancreatic adenocarcinomas are lacking; therefore, there is no standard indication or regimen. Because of the low prevalence of these cancers, conducting these types of trials is difficult. Indeed, only one randomized clinical trial¹¹ has addressed the role of adjuvant chemotherapy after resection of periampullary adenocarcinomas, to our knowledge, and despite that 100 centers participated in the study, the incidence of each tumor type was still low. Conversely, the role of adjuvant therapy for PDAC is well defined.²⁵ Our results show that PB-type cancers are similar to PDAC; therefore, this study provides a rationale for applying the current guidelines²⁵ for PDAC to these cancers. In addition, when considering treating patients with intestinal-type tumors with chemotherapy, it is probably best to avoid gemcitabine-based regimens because this drug is not generally effective for carcinomas of intestinal origin.^27,28

Last, our results could also influence the surgical management of these patients. Given that intestinal-type tumors were associated with less aggressive histologic features, our findings complement a prior study,²⁹ which concluded that resection of periampullary adenocarcinoma liver metastases provides a survival benefit but only for those with intestinal differentiation. Further supporting this idea is the similarity between intestinal-type tumors and colon cancer on H&E staining, and the benefit of hepatic metastasectomy for the latter is well known.³⁰ Conversely, patients with metastatic PDAC have a poor prognosis and do not benefit from resection of liver metastases.³¹ Based on our findings, patients with PB-type cancers are likely to have similarly poor outcomes, and metastasectomies would not likely improve survival.

Our study was somewhat limited by its retrospective nature because we were only able to describe similarities and differences between HP phenotypes. However, given the rarity of non-PDAC cancers in the periampullary region, prospective studies are difficult to perform, and large retrospective studies, such as ours, are a reasonable alternative to evaluate these cancers. It is also important to note that intestinal-type PDAC has been previously described.³² However, we do not believe that the inclusion of this PDAC subtype significantly affected our results because we excluded intraductal papillary mucinous neoplasm–based cancers, which comprise most of the subtype. One additional potential limitation was the greater use of adjuvant chemotherapy in patients with PDAC; however, adjuvant treatment was not significantly associated with survival on multivariable Cox proportional hazards regression analysis. Well-annotated treatment regimens were also not included because many patients were treated outside of UCLA. Finally, we cannot comment on the management of patients with mixed type tumors, who were excluded from our study.

The wide variability in survival and response to therapy seen with periampullary adenocarcinomas may be attributable to the current classification scheme that groups these cancers by anatomic location of origin. Alternatively, the cancers can be organized according to HP phenotype (intestinal or PB), which better estimates survival after resection than anatomic location. In addition, PB-type tumors of the duodenum, ampulla, and distal CBD are similar to PDAC with regard to pathologic features and patient survival after resection. Therefore, it may be reasonable to apply current treatment guidelines for PDAC to periampullary PB-type cancers.

Corresponding Author: Timothy R. Donahue, MD, Department of Surgery, David Geffen School of Medicine, University of California, Los Angeles, 10833 Le Conte Ave, 72-215 Center for Health Sciences, PO Box 956904, Los Angeles, CA 90095 (tdonahue@mednet.ucla.edu).

Accepted for Publication: June 20, 2016.

Published Online: October 12, 2016. doi:10.1001/jamasurg.2016.3466

Author Contributions: Drs Williams and Donahue had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Williams, Toste, Vasquez, Sunjaya, Swanson, Hines, Reber, Dawson, Donahue.

Acquisition, analysis, or interpretation of data: Williams, Chan, Toste, Elliott, Vasquez, Sunjaya, Swanson, Koo, Reber, Dawson, Donahue.

Drafting of the manuscript: Williams, Chan, Sunjaya, Donahue.

Critical revision of the manuscript for important intellectual content: Chan, Toste, Elliott, Vasquez, Swanson, Koo, Hines, Reber, Dawson, Donahue.

Statistical analysis: Williams, Chan, Vasquez, Sunjaya, Reber.

Administrative, technical, or material support: Toste, Elliott, Vasquez, Sunjaya, Hines, Reber, Dawson, Donahue.

Study supervision: Reber, Dawson, Donahue.

Conflict of Interest Disclosures: None reported.

Funding/Support: Dr Elliott was supported by training grant T32 NIH-NIDDK DK 41301 from the National Institutes of Health. Dr Donahue was supported by grants R01 CA187678-01 and U01 CA198846 from the National Cancer Institute and by grant DE-SC0012353 from the Department of Energy, Office of Health, Safety and Security.

Role of the Funder/Sponsor: The funding sources aided in the collection, management, analysis, and interpretation of the data; the preparation, review, and approval of the manuscript; the design and conduct of the study; and the decision to submit the manuscript for publication.

Previous Presentation: This study was presented at the 87th Annual Meeting of the Pacific Coast Surgical Association; February 14, 2016; Kohala Coast, Hawaii.