Abstract
Importance
Although patients with resected pancreatic adenocarcinoma are at high risk for disease recurrence, few markers are available to inform patient outcomes.
Objective
To evaluate alterations of the four main driver genes for pancreatic adenocarcinoma and patient outcomes after cancer resection.
Design, Setting, and Participants
We analyzed protein expression and DNA alterations for KRAS, CDKN2A, SMAD4, and TP53 by immunohistochemistry and next-generation sequencing in formalin-fixed, paraffin-embedded tumors from 356 patients with resected pancreatic adenocarcinoma evaluated at three U.S. centers. Associations of driver gene alterations with disease-free survival (DFS) and overall survival (OS) were evaluated using Cox proportional hazards regression with estimation of hazard ratios (HR) and 95% confidence intervals (CI) and adjustment for age, sex, tumor characteristics, institution, and peri-operative treatment.
Main Outcomes
DFS and OS among patients with resected pancreatic adenocarcinoma
Results
Patients with KRAS mutant tumors had worse DFS and OS compared to patients with KRAS wild-type tumors, with median OS of 20.3 versus 38.6 months and 5-year OS of 13.0% versus 30.2%, respectively. Particularly poor outcomes were identified in patients with KRAS G12D-mutant tumors, who had median OS of 15.3 months. Patients whose tumors lacked CDKN2A expression had worse DFS and OS compared to patients whose tumors retained CDKN2A, with median OS of 19.7 versus 24.6 months and 5-year OS of 11.9% versus 19.5%, respectively. SMAD4 status was not associated with DFS or OS, while TP53 status was associated only with DFS (P=0.04). Patients had worse DFS and OS with greater number of altered driver genes. Compared to patients with 0-2 altered genes, those with 4 altered genes had HR for DFS of 1.79 (1.24-2.59; P<0.01) and OS of 1.38 (0.98-1.94; P=0.06). Five-year OS was 18.4% for patients with 0-2 gene alterations, 14.1% for 3 alterations and 8.2% for 4 alterations. Alterations in the four driver genes were not significantly associated with local recurrence as the first site of disease recurrence.
Conclusions and Relevance
Patient outcomes are associated with alterations of the four main driver genes in resected pancreatic adenocarcinoma.
INTRODUCTION
Pancreatic cancer is the third leading cause of cancer-related death in the United States.1 Recent large-scale genome sequencing studies have identified multiple molecular pathways involved in pancreatic adenocarcinoma initiation and progression.2-4 Four main driver genes have been identified, KRAS, CDKN2A, SMAD4, and TP53, that are critical for pancreatic cancer growth. The impact of these driver gene alterations on patient outcomes has not been clearly established. Therefore, we characterized the status of these four driver genes using immunohistochemistry (IHC) and next-generation DNA sequencing (NGS) in a large, highly-annotated, multi-institutional patient population with resected pancreatic adenocarcinoma.
METHODS
The study population consisted of 356 patients with resected pancreatic adenocarcinoma who were treated at Dana-Farber/Brigham and Women’s Cancer Center (DF/BWCC, n=126) between 10/26/2002 and 5/21/2012; at University of Rochester Medical Center (URMC, n=90) between 3/1/2006 and 11/1/2013; or Stanford Cancer Institute (SCI, n=140) between 9/26/1995 and 5/22/2013. Institutional Review Board approval was granted at each center.
IHC for CDKN2A, SMAD4, and TP53 was performed on formalin-fixed paraffin-embedded whole tissue sections (eMethods, eFigure 1). After macroscopic dissection, genomic DNA was extracted from tumor and adjacent normal tissue. Pyrosequencing for KRAS hotspot mutations and NGS using a customized, massively parallel sequencing panel were performed to determine the molecular status of KRAS, CDKN2A, SMAD4, and TP53 (eMethods). For KRAS, tumors were classified as mutant or wild-type based on NGS or pyrosequencing if predefined NGS coverage metrics were not met (eFigure 2). For CDKN2A and SMAD4, tumors were classified as intact or lost based on IHC results. For TP53, IHC and sequencing data were combined to make an integrated call as wild-type or altered (eMethods).
Disease-free survival (DFS) was defined as time between surgery and disease recurrence, and overall survival (OS) as time between surgery and death. We classified disease recurrence as “local” when it occurred in or adjacent to the pancreatic remnant or in the retroperitoneum. Disease recurrence outside these sites was classified as “distant”. Follow-up continued through 6/28/2016 for DF/BWCC, 3/17/2016 for URMC, and 3/11/2016 for SCI. A flow diagram of the study population is presented in eFigure 3.
We evaluated associations of driver gene alterations with DFS and OS using multivariable-adjusted Cox proportional hazards regression (eMethods), calculating hazard ratios (HR) and 95% confidence intervals (95% CI). We generated Kaplan-Meier curves, from which we calculated median, 2-year and 5-year survival. We also analyzed the association between gene alterations and pattern of first recurrence using multivariable-adjusted logistic regression, calculating odds ratios (OR) and 95% CI. All hypothesis tests were two-sided and statistical significance was P<0.05.
RESULTS
Baseline characteristics of the study population by institution and the four main driver genes are presented in eTables 1 and 2, respectively. Median DFS and OS were 13.1 and 21.0 months, respectively, comparable to recent randomized trials.5,6
Activating KRAS mutations were observed in 328 (92%) patients (eTable 3); KRAS mutations affecting two separate codons were found in 11 (3.4%) tumors. Patients with KRAS mutant tumors had worse DFS and OS compared to patients with KRAS wild-type tumors (Table 1), and patients with KRAS G12D mutant tumors had particularly poor outcomes (Figure 1, Table 2, and eFigure 4).
Table 1.
Disease-free survival and overall survival by tumor KRAS, CDKN2A, SMAD4, and TP53 status
| Disease-Free Survival (DFS) | Overall Survival (OS) | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
| ||||||||||||
| No. of patients |
Median DFS (Months) |
2-Year Survival |
5-Year Survival |
HRa (95% CI) | P valueb | No. of patients |
Median OS (Months) |
2-Year Survival |
5-Year Survival |
HRa (95% CI) | P valueb | |
| KRAS | ||||||||||||
| Wild-type | 27 | 16.2 | 30.2% | 20.2% | 1.00 (reference) | 27 | 38.6 | 63.0% | 30.2% | 1.00 (reference) | ||
| Mutant | 308 | 12.3 | 27.5% | 12.4% | 1.72 (1.04-2.84) | 0.03 | 311 | 20.3 | 44.5% | 13.0% | 1.55 (0.96-2.51) | 0.08 |
| CDKN2A | ||||||||||||
| Intact | 111 | 14.8 | 31.2% | 16.9% | 1.00 (reference) | 112 | 24.6 | 53.8% | 19.5% | 1.00 (reference) | ||
| Lost | 224 | 11.5 | 26.0% | 11.5% | 1.62 (1.19-2.20) | <0.01 | 226 | 19.7 | 42.3% | 11.9% | 1.44 (1.08-1.91) | 0.01 |
| SMAD4 | ||||||||||||
| Intact | 172 | 11.5 | 27.1% | 14.4% | 1.00 (reference) | 173 | 21.3 | 49.1% | 15.8% | 1.00 (reference) | ||
| Lost | 163 | 13.6 | 28.4% | 12.3% | 1.18 (0.90-1.55) | 0.25 | 165 | 20.5 | 43.0% | 12.9% | 1.07 (0.83-1.38) | 0.62 |
| TP53 | ||||||||||||
| Wild-type | 118 | 14.8 | 31.4% | 13.9% | 1.00 (reference) | 119 | 24.6 | 50.7% | 18.7% | 1.00 (reference) | ||
| Altered | 217 | 10.8 | 25.7% | 12.6% | 1.33 (1.02-1.75) | 0.04 | 219 | 20.3 | 43.5% | 12.3% | 1.18 (0.91-1.53) | 0.23 |
Abbreviations: CI, confidence interval; HR, Hazard ratio.
Cox proportional hazards regression model adjusted for age, sex, pN stage, tumor grade, lymphovascular invasion, receipt of perioperative treatment, resection margin status, and institution.
P value for multivariable-adjusted Cox proportional hazards regression.
Figure 1.
Kaplan-Meier survival curves for overall survival by A. KRAS mutation status,a and B. number of gene alterations in the four main driver genes (KRAS, CDKN2A, SMAD4, TP53).
Table 2.
Disease-free survival and overall survival by KRAS codon mutation and combined KRAS, CDKN2A, SMAD4, and TP53 gene alterations
| Disease-Free Survival | Overall Survival | |||||
|---|---|---|---|---|---|---|
|
| ||||||
| No. of patients | HRa (95% CI) | P valueb | No. of patients | HRa (95% CI) | P valueb | |
| KRAS mutationc | ||||||
| G12D | 122 | 1.00 (reference) | 123 | 1.00 (reference) | ||
| G12V | 104 | 0.57 (0.41-0.79) | <0.01 | 105 | 0.63 (0.46-0.87) | <0.01 |
| G12R | 44 | 0.67 (0.43-1.05) | 0.08 | 45 | 0.82 (0.54-1.25) | 0.35 |
| Other codon | 25 | 0.63 (0.37-1.10) | 0.10 | 25 | 0.83 (0.50-1.39) | 0.48 |
| Two codon mutations | 11 | 0.27 (0.11-0.69) | <0.01 | 11 | 0.55 (0.26-1.15) | 0.11 |
| Wild-type | 27 | 0.38 (0.22-0.65) | <0.01 | 27 | 0.50 (0.30-0.83) | <0.01 |
| No. of altered genes | ||||||
| 0-2 genes | 126 | 1.00 (reference) | 127 | 1.00 (reference) | ||
| 3 genes | 135 | 1.37 (1.01-1.86) | 0.05 | 136 | 1.22 (0.91-1.64) | 0.18 |
| 4 genes | 74 | 1.79 (1.24-2.59) | <0.01 | 75 | 1.38 (0.98-1.94) | 0.06 |
| Gene combinationsd | ||||||
| 0-2 genes | 126 | 1.00 (reference) | 127 | 1.00 (reference) | ||
| 3 genes | ||||||
| KRAS, SMAD4, TP53 | 35 | 1.16 (0.74-1.82) | 0.53 | 35 | 1.08 (0.69-1.69) | 0.75 |
| KRAS, CDKN2A, TP53 | 64 | 1.51 (1.04-2.20) | 0.03 | 64 | 1.38 (0.96-1.98) | 0.08 |
| KRAS, CDKN2A, SMAD4 | 34 | 1.27 (0.79-2.05) | 0.32 | 35 | 1.28 (0.80-2.06) | 0.30 |
Abbreviations: CI, confidence interval; HR, Hazard ratio.
Cox proportional hazards regression model adjusted for age, sex, pN stage, tumor grade, lymphovascular invasion, receipt of perioperative treatment, resection margin status, and institution.
P value for multivariable-adjusted Cox proportional hazards regression.
Data not reported for patients with KRAS G12C mutations due to small sample size (n = 2).
Data not reported for patients with combination of CDKN2A, SMAD4, and TP53 alterations due to small sample size (n = 2)
By IHC, CDKN2A was lost in 240 (67%) patients. Patients with loss of CDKN2A by IHC had worse DFS and OS compared to patients with intact CDKN2A (Table 1). In sensitivity analyses, we classified CDKN2A status using IHC and sequencing data (eTable 4). CDKN2A loss by IHC was associated with worse DFS and OS regardless of predicted molecular status (eTable 5), likely reflecting the inability of NGS to detect silencing of CDKN2A expression due to methylation and reduced sensitivity for copy number loss in low cellularity tumors.
By IHC, SMAD4 was lost in 175 (49%) patients. Loss of SMAD4 was not significantly associated with DFS or OS in our patient population (Table 1). We utilized our sequencing data to predict whether SMAD4 protein expression would be present or lost (eTable 4). Predicted molecular status of SMAD4 did not influence DFS or OS when SMAD4 was lost by IHC (eTable 5).
For TP53, we used IHC and molecular data to generate an integrated call of TP53 as wild-type or altered (eTable 4). By this approach, TP53 was altered in 231 (65%) patients. Altered TP53 was associated with shorter DFS (P=0.04), but not with OS (P=0.23) (Table 1).
Twenty-four patients received pre-operative therapy, which was not associated with the pattern of driver gene alterations (eTable 6). In sensitivity analyses excluding these 24 patients, associations between driver gene alterations and patient outcomes were largely unchanged (eTable 7).
We analyzed the association of combinatorial gene alterations with DFS and OS (eTable 8). Compared to patients with 0-2 gene alterations, patients with 3 or 4 alterations had worse DFS and OS (Table 2, Figure 1, eFigure 4). The worst outcomes were identified in patients with KRAS-mutant and CDKN2A-lost tumors. Five-year OS rates were 18.4% for patients with 0-2 gene alterations, 14.1% for 3 alterations, and 8.2% for 4 alterations. In our patient population, alterations in the four driver genes were not significantly associated with local recurrence as the first site of disease recurrence (eTable 9).
DISCUSSION
In a large, multi-institutional population of patients with resected pancreatic adenocarcinoma, patient outcomes were associated with alterations of the four main driver genes. Prior studies have individually assessed these genes and patient outcomes using a variety of methods and patient populations with inconsistent results.7-10 A more recent study assessed all four driver genes among 79 patients who underwent rapid autopsy after death from pancreatic adenocarcinoma.11 Tumors were sequenced by polymerase chain reaction for KRAS, CDKN2A, and TP53, and IHC was performed for CDKN2A, SMAD4, and TP53. Although sample size was small and included all stages of disease, patients whose tumors had 3-4 altered genes had worse OS than patients whose tumors had 1-2 altered genes in unadjusted analysis (log-rank P-value=0.04). The primary results of the current study are confirmatory in a large, multi-institutional patient population, but multiple-hypothesis testing should be acknowledged when interpreting data across several markers.
Previous studies have suggested that loss of SMAD4 by IHC was associated with extensive metastatic spread, generating interest in SMAD4 staining as a predictive biomarker for tailored use of radiotherapy.12 However, a subsequent study of 127 patients with resected pancreatic cancer did not replicate these findings.13 In our study population, SMAD4 staining was not a predictor of disease recurrence pattern after surgical resection.
Adjuvant treatment following surgical resection improves patient survival, but outcomes remain suboptimal.5 With the intent of improving cure rates, novel and more aggressive multi-agent treatment programs are currently being devised and evaluated in the adjuvant setting.6 Furthermore, patients are increasingly receiving chemotherapy and radiation before surgical resection, to rapidly initiate therapy against micrometastatic disease and select out those patients with early disease progression unlikely to benefit from surgery.14 In the future, molecular assessment of pancreatic cancer may have utility in guiding the use and components of peri-operative treatment programs.15
In summary, we demonstrate that alterations in the four main driver genes are associated with patient outcomes in a large, multi-institutional population of patients with resected pancreatic adenocarcinoma. Understanding the molecular events that determine patient outcomes has the potential to improve treatment approaches for patients with this aggressive malignancy.
Supplementary Material
KEY POINTS.
Question
Do alterations in the four main driver genes for pancreatic adenocarcinoma affect patient outcomes after pancreatic cancer resection?
Findings
Immunohistochemistry and next-generation DNA sequencing of formalin-fixed, paraffin-embedded pancreatic adenocarcinoma resection specimens identified alterations in the four main driver genes, KRAS, CDKN2A, SMAD4, and TP53. Disease-free and overall survival were associated with the presence and pattern of alterations in these genes, independent of previously identified prognostic factors.
Meaning
Identifying pathogenic alterations in the four main driver genes of pancreatic adenocarcinoma informs patient outcomes.
Acknowledgments
The authors would like to thank Nabeel Bardeesy, Alec Kimmelman, R. Coleman Lindsley, Mandar Muzumdar, Eliezer Van Allen, and Matthew Vander Heiden for their help in designing the targeted sequencing panel.
Role of the Sponsor: The sponsors did not participate in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.
Funding/Support:
Support from Medical Oncology Translational Grant Program of Dana-Farber Cancer Institute to Z.R. Qian; from NIH K07 CA148894 to K. Ng; from NCI R35 CA197735 to S. Ogino; from the Hale Center for Pancreatic Cancer Research, Perry S. Levy Fund for Gastrointestinal Cancer Research, Pappas Family Research Fund for Pancreatic Cancer, NIH R01 CA124908, and NIH P50 CA127003 to C.S. Fuchs; MyBlueDots Fund for A.C. Koong; and from Hale Center for Pancreatic Cancer Research, NIH/NCI U01 CA210171, Department of Defense CA130288, Lustgarten Foundation, Pancreatic Cancer Action Network, Noble Effort Fund, Peter R. Leavitt Family Fund, Wexler Family Fund, and Promises for Purple to B.M. Wolpin.
Footnotes
Author Contributions:
V Morales-Oyarvide and BM Wolpin had full access to all of 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: ZR Qian, DA Rubinson, AF Hezel, AC Koong, and BM Wolpin
Acquisition of data: All authors
Analysis and interpretation of data: All authors
Drafting of the manuscript: ZR Qian, DA Rubinson, JA Nowak, V Morales-Oyarvide, AR Thorner, and BM Wolpin
Critical revision of the manuscript for important intellectual content: All authors
Statistical analysis: V Morales-Oyarvide, C Yuan, and BM Wolpin
Obtained funding: CS Fuchs, S Ogino, AF Hezel, AC Koong, and BM Wolpin
Administrative, technical, or material support: All authors
Study supervision: AR Thorner, JL Hornick, AF Hezel, AC Koong, and BM Wolpin
Financial Disclosures: None.
References
- 1.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin. 2016;66(1):7–30. doi: 10.3322/caac.21332. [DOI] [PubMed] [Google Scholar]
- 2.Jones S, Zhang X, Parsons DW, et al. Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science. 2008;321(5897):1801–1806. doi: 10.1126/science.1164368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Witkiewicz AK, McMillan EA, Balaji U, et al. Whole-exome sequencing of pancreatic cancer defines genetic diversity and therapeutic targets. Nat Commun. 2015;6:6744. doi: 10.1038/ncomms7744. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bailey P, Chang DK, Nones K, et al. Genomic analyses identify molecular subtypes of pancreatic cancer. Nature. 2016;531(7592):47–52. doi: 10.1038/nature16965. [DOI] [PubMed] [Google Scholar]
- 5.Oettle H, Post S, Neuhaus P, et al. Adjuvant chemotherapy with gemcitabine vs observation in patients undergoing curative-intent resection of pancreatic cancer: a randomized controlled trial. JAMA. 2007;297(3):267–277. doi: 10.1001/jama.297.3.267. [DOI] [PubMed] [Google Scholar]
- 6.Li D, O’Reilly EM. Adjuvant and neoadjuvant therapy for pancreatic cancer. Surg Oncol Clin N Am. 2016;25(2):311–326. doi: 10.1016/j.soc.2015.11.010. [DOI] [PubMed] [Google Scholar]
- 7.Gerdes B, Ramaswamy A, Ziegler A, et al. p16INK4a is a prognostic marker in resected ductal pancreatic cancer: an analysis of p16INK4a, p53, MDM2, an Rb. Ann Surg. 2002;235(1):51–59. doi: 10.1097/00000658-200201000-00007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Blackford A, Serrano OK, Wolfgang CL, et al. SMAD4 gene mutations are associated with poor prognosis in pancreatic cancer. Clin Cancer Res. 2009;15(14):4674–4679. doi: 10.1158/1078-0432.CCR-09-0227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Rachakonda PS, Bauer AS, Xie H, et al. Somatic mutations in exocrine pancreatic tumors: association with patient survival. PLoS One. 2013;8(4):e60870. doi: 10.1371/journal.pone.0060870. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Oshima M, Okano K, Muraki S, et al. Immunohistochemically detected expression of 3 major genes (CDKN2A/p16, TP53, and SMAD4/DPC4) strongly predicts survival in patients with resectable pancreatic cancer. Ann Surg. 2013;258(2):336–346. doi: 10.1097/SLA.0b013e3182827a65. [DOI] [PubMed] [Google Scholar]
- 11.Yachida S, White CM, Naito Y, et al. Clinical significance of the genetic landscape of pancreatic cancer and implications for identification of potential long-term survivors. Clin Cancer Res. 2012;18(22):6339–6347. doi: 10.1158/1078-0432.CCR-12-1215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Iacobuzio-Donahue CA, Fu B, Yachida S, et al. DPC4 gene status of the primary carcinoma correlates with patterns of failure in patients with pancreatic cancer. J Clin Oncol. 2009;27(11):1806–1813. doi: 10.1200/JCO.2008.17.7188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Winter JM, Tang LH, Klimstra DS, et al. Failure patterns in resected pancreas adenocarcinoma: lack of predicted benefit to SMAD4 expression. Ann Surg. 2013;258(2):331–335. doi: 10.1097/SLA.0b013e31827fe9ce. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Perez K, Clancy TE, Mancias JD, Rosenthal MH, Wolpin BM. When, what, and why of perioperative treatment of potentially curable pancreatic adenocarcinoma. J Clin Oncol. doi: 10.1200/JCO.2016.70.2134. [published online December 28, 2016]. 2016:JCO2016702134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Valero V, 3rd, Saunders TJ, He J, et al. Reliable Detection of Somatic Mutations in Fine Needle Aspirates of Pancreatic Cancer With Next-generation Sequencing: Implications for Surgical Management. Ann Surg. 2016;263(1):153–161. doi: 10.1097/SLA.0000000000001156. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.