Pancreatic Adenocarcinoma: New Strategies for Success

Abstract

Pancreatic adenocarcinoma ranks as the most challenging of human malignancies, with overall 5-year survivorship being measured in a couple of percent. Major progress has occurred regarding the molecular underpinnings and pathogenesis of pancreatic adenocarcinoma, definition of the epidemiology and genetics of this disease, identification of individuals at risk, and in 2008, the preliminary description of the pancreatic genome. However, clinical developments over the past decade have been modest and incremental at most. The core drug and the backbone of treatment in all settings of pancreatic adenocarcinoma— adjuvant, locally advanced, and metastatic—remains gemcitabine. The past decade of research focused initially on combining cytotoxic therapies with gemcitabine, and more recently, on combining newer “targeted agents.” Some success has been observed by combining the platinum analogs and the fluoropyrimidines with gemcitabine in the advanced pancreatic cancer setting. Three- and four-drug combinations have also been assessed, but the data are limited and the major trade-off becomes a toxicity-benefit equation. In relative terms, more limited incremental gains have been observed by combining erlotinib with gemcitabine, while other randomized phase III trials of targeted agents combined with gemcitabine have essentially shown no benefit. Several of the newer generation anti-vascular agents (VEGF-trap, axitinib) are being evaluated in ongoing phase III trials. In the short-term, expectations for advances in pancreatic adenocarcinoma therapy are reserved, with most progress likely to be made in therapy refinement and patient selection. However, it is reasonable to surmise that major progress will evolve as the molecular biology of pancreatic adenocarcinoma continues to be unraveled, as the infrastructure for translational research is strengthened with new preclinical models, and with recognition of the prerequisite requirement for intensive cross-disciplinary collaboration.

Estimates are that nearly 38,000 individuals were diagnosed with pancreatic adenocarcinoma in the United States in 2008.¹ For the majority, the disease is either locally advanced and unresectable or de novo metastatic at the time of diagnosis, conferring median survival durations of 3 to 18 months. Approximately 15% of patients present with localized disease and undergo a potentially curative surgical intervention; however, even in this most select of patient subgroups, the anticipated 5-year survival rates range from 12% to 20%.^2,3 These collective statistics only partly underscore the challenges of this malignancy. Much more difficult to quantitate, but equally important in terms of the ultimate burden of this disease, is the morbidity encountered by patients and the direct and indirect effects on their families, friends, and society at large. Furthermore, while accomplishments in clinical therapeutics have been observed, progress has been incremental and modest at best. More encouragingly, our understanding of the molecular pathogenesis, genetics, and epidemiology, as well as translational research in pancreas adenocarcinoma, have provided important insights into where the next generation of efforts will be focused.^4–7 This article provides a brief overview of the genetics of pancreatic cancer, focusing on identification of at-risk families; an overview of current clinical standards including the highlights and disappointments of the era of “targeted” therapy; and new therapeutic directions.

GENETICS OF PANCREAS ADENOCARCINOMA

It is estimated that approximately 10% of pancreatic cancers may be attributed to genetic factors.^8–10 There are two major groups: the larger group is families with multiple individuals diagnosed with pancreas cancer but with no identifiable genetic defect; the smaller and more welldefined group includes recognizable genetic syndromes (Table 1).¹¹ The importance of defining the genetics of this disease relates to the ultimate goals of identification of high-risk individuals, implementation of risk-reduction strategies, and decreasing mortality from this disease.¹² Early data from the National Familial Pancreas Tumor Registry have demonstrated that healthy members of families that have two or more first-degree relatives with pancreas cancer have a 6.4-fold higher risk of developing the disease. This figure increases to a 32-fold higher risk if three or more first-degree relatives have the disease. Another key observation from these prospective registries is that not smoking reduces the risk of developing pancreatic cancer among these family members.¹³ Smoking cessation is a key public health measure and one potentially modifiable risk factor for this malignancy.

Table 1.

Single gene defects associated with pancreas adenocarcinoma

Syndrome	Mutated Gene	Relative Risk (Estimate)	Reference
Peutz-Jeghers	STK11 (19p13)	132 x	Gastroenterol 200040
Hereditary Pancreatitis	PRSSI (7q35) SPINK1 (5q31)	~ 50 x	Pancreatol 200141
FAMMM (Familial Atypical Multiple Mole Melanoma)	p16 (9p21)	13–22 x	NEJM 199542
FAP (Familial Adenomatous Polyposis)	APC (5q13)	4.5 x	JOP 200843
HBOC (Hereditary Breast-Ovary Cancer)	BRCA1(17q21) BRCA2 (13q12)	2.2 x 3.5 x	JNCI 200244 JNCI 199945
HNPCC (Hereditary Non-Polyposis Colon Cancer)	MLHI (3p21) MSH2 (2p16)	Increased Increased	Cancer 199646
Ataxia Telangiectasia	ATM (11q23)	Increased	Clin Genetics 199947

Our pancreatic registry at Memorial Sloan-Kettering Cancer Center focuses on two groups of individuals: (1) Probands with pancreatic cancer diagnosed at less than 50 years of age, or with one or more first-degree relatives with pancreatic cancer, or two or more second-degree relatives with pancreatic cancer, or known BRCA-1 or BRCA-2 mutation; (2) the family member part of the registry is for individuals not diagnosed with pancreas cancer but with similar entry characteristics as for the proband group. Early registry findings have yielded the identification of two patients with adenocarcinomas following screening. Both patients subsequently underwent resection, including a 58-year-old female (T3,N0,M0) with two first-degree relatives, and a 46-year-old female (T3,Nx,M0) with one first-degree relative with pancreatic cancer. Other findings have included identification of one neuroendocrine cancer and four intraductal papillary mucinous neoplasms (IPMN). The latter observation raises the question of whether IPMNs might represent a precursor to adenocarcinoma in certain familial settings. Clearly, no conclusions can be drawn as these observations are preliminary—but enticing—with regard to identification of at-risk individuals. Consensus recommendations regarding who is at high risk and how to screen remain to be established. However, careful review of family history, particularly in families where pancreatic cancer occurs with breast, ovarian, other gastrointestinal cancers, or a history of pancreatitis, may identify at-risk families.^14,15 These individuals should be considered for referral to a clinical genetics service and, if appropriate, undergo focused testing, eg, BRCA-2, etc, be counseled about smoking cessation, and ideally be enrolled on a prospective registry for early detection and screening.¹⁶ This also points to an emerging concept with potential future therapeutic application: A small but identifiable subset of patients (~ 7%) diagnosed with pancreatic cancer has a mutation in BRCA-2.^8,17 Such patients may have greater susceptibility to DNA-damaging agents such as mitomycin or PARP (poly[ADP-ribose] polymerase) inhibitors.¹⁸ These latter drugs are coming to the clinic for pancreas cancer and information will evolve over the next couple of years.

STATE-OF-THE-ART THERAPY FOR PANCREAS ADENOCARCINOMA 2008

Given that the vast majority of people with pancreatic cancer are diagnosed with locally advanced or metastatic disease, the major focus of drug development has been in the advanced and metastatic settings. Single-agent gemcitabine has been the default standard of care and has stood the test of a decade of challenge in this patient population. Data from Burris et al in 1997 demonstrated an improved clinical benefit (24% vs. 5%, P = .022), modest improvement in median survival (5.6 vs. 4.3 months, P = .0025), and significant increase in 1- year survival (18% vs. 2%) for gemcitabine compared with 5-fluorouracil in what may be one of the smallest randomized trials in any solid tumor to lead to drug approval by the U.S. Food and Drug Administration.¹⁹ Over the past decade, research aimed at improving outcomes in pancreatic cancer has focused on three groups of trials: comparing gemcitabine to other single agents (Table 2); combining gemcitabine with other cytotoxics (Table 3); and in the last 5 years, combining gemcitabine with newer targeted agents.²⁰ For the most part, this has been a relatively disappointing series of ventures. In the first group of trials, other single agents have resulted in inferior outcomes compared with those achieved with gemcitabine, eg, exatecan mesylate (topoisomerase- I inhibitor),²¹ marimastat, and BAY 12-9566 (both matrix metalloproteinase inhibitors).²² These studies have indirectly re-endorsed the value of gemcitabine. Similarly, combinations of cytotoxic agents with gemcitabine, at least in individual trials, have not yielded improved outcomes, with the exception of a gemcitabine and capecitabine combination reported preliminarily from the Medical Research Council (MRC) group in the United Kingdom²³; however, mature outcome data are awaited to confirm the value of this combination. Results of pooled analyses and metaanalyses have provided support for use of selected two-drug, gemcitabine-based combinations, particularly combined with a platinum agent, and to a lesser extent, combined with fluoropyrimidines.^24,25

Table 2.

Median overall survival in patients with pancreatic cancer treated with selected single agents compared with gemcitabine

	New Single-Agent Drug (months)	Gemcitabine (months)	P value
Exatecan Mesylate	4.95	6.46	.993
Marimastat	3.5–4.1	5.6	–
BAY 12-9566	3.2	6.4	.0001

Table 3.

Median overall survival of pancreatic cancer patients treated with selected gemcitabine + drug “X” combinations compared with gemcitabine alone

Drug X	No. pts	Gemcitabine + Drug X (months)	Gemcitabine (months)	P value
Bolus 5-Fluorouracil	322	6.7	5.4	.11
Infusional 5-Fluorouracil	466	5.9	6.2	.68
Pemetrexed	565	6.2	6.3	.85
Capecitabine	319	8.4	7.3	.31
Capecitabine	533	7.4	6.0	.014*
Irinotecan	360	6.3	6.6	.78
Exatecan	349	6.7	6.2	.52
Cisplatin	198	7.6	6.0	.12
Oxaliplatin	313	9.0	7.1	.13

^*Data in abstract form

Another area of research that may have tangible clinical utility is pharmacogenomics as applied to standard cytotoxic agents that are active in pancreatic adenocarcinoma. Genetic polymorphisms, and tumor-specific expression of mRNA and proteins, may affect both the efficacy and toxicity of gemcitabine.²⁶ For example, hENT1 (nucleoside transporter), RRM1 (rate-limiting step in DNA synthesis pathway), and DNA repair polymorphisms may have prognostic implications; low levels of the enzyme cytidine deaminase involved in the pyrimidine salvage pathway may confer toxicity. Examples for other drugs include dihydropyrimidine dehydrogenase (DPD) and thymidylate synthase (TS) levels for 5-fluorouracil, and excision repair cross-complementation group 1 (ERCC-1) for oxaliplatin, which may confer information related to toxicity and sensitivity, respectively. Prospective pharmacogenomic- based biomarker studies are needed to define utility and identify patient subsets where therapy refinement is feasible.

TARGETED THERAPY FOR PANCREATIC ADENOCARCINOMA

The expectations and hope for the value of new biologic agents have, unfortunately, exceeded the reality of the challenges of pancreatic cancer. However, results of a randomized phase III trial comparing gemcitabine and erlotinib vs. gemcitabine in patients with untreated advanced disease provided a small but real hint of activity.²⁷ A modest but statistically significant difference in median survival was noted, along with an improved progression-free survival (PFS) and a more compelling, approximate 40% improvement in 1-year survival (from 17% with gemcitabine to 23% with the combination). Moreover, there was an association between rash severity and survival in this trial, as has been observed in studies in other malignancies where anti-epidermal growth factor receptor (EGFR)-based therapies have had utility. Patients with rash grade ≥ 2 had a near-doubling of their survival to 10.5 months, compared with 5.3 months for patients with no rash. The challenge will be to identify the patients most likely to benefit from treatment in advance of a therapeutic trial. Information on molecular correlates was available from only approximately 21% (n = 117) of the study patients. There was a suggestion that patients with wild-type ras tumors may have had increased benefit from erlotinib; however, the small sample size limited the conclusions that could be made in this post-hoc K-ras mutational analysis.²⁸ For now, the rash and ras story is evolving and there are no specific implications regarding patient care. In contrast to the data on gemcitabine combined with erlotinib, results of a phase III randomized trial comparing gemcitabine and cetuximab vs. gemcitabine in untreated, advanced pancreatic cancer patients have indicated no advantage with the addition of cetuximab (median survival, 5.4 months for the combination and 5.9 months for gemcitabine, hazard ratio [HR] 1.09).²⁹ Of note, the absolute difference in survival for both erlotinib and cetuximab were similar at 0.5 months; however, one study was “positive” and the other “negative,” raising the question of whether there is a difference between a small molecule tyrosine kinase inhibitor and a monoclonal antibody in this disease.

The anti-angiogenesis story is also unfolding in pancreatic cancer. Much promise was put upon an early phase II trial of gemcitabine and bevacizumab in a very select population of patients with untreated advanced disease.³⁰ Unfortunately, results of a randomized phase III trial showed no improvement for the addition of bevacizumab to gemcitabine in a broader population of patients with advanced pancreatic cancer (median survival: gemcitabine, 6.1 months; gemcitabine and bevacizumab, 5.8 months; HR 10.3; P = .78).³¹ Axitinib, an oral vascular endothelial growth factor (VEGF) 1, 2, 3 inhibitor, which demonstrated promise in a small randomized phase II study,³² has completed phase III testing in advanced disease. A press release reported that the primary end point had not been met,³³ and further details are awaited. A phase III study of gemcitabine +/− VEGFtrap (aflibercept) in patients with metastatic pancreatic cancer and Eastern Cooperative Oncology Group (ECOG) performance status of 0–1 is ongoing. The AViTA study was the first trial to combine an anti-angiogenic and an anti-EGFR therapy in the phase III setting in pancreatic cancer.³⁴ This trial evaluated the addition of bevacizumab to gemcitabine and erlotinib in patients with metastatic disease and ECOG performance status of 0–1. The study did not meet its primary end point: overall survival was 7.1 months vs. 6 months (P = .208) with vs. without bevacizumab, respectively; however, PFS improved significantly in the experimental arm, from 3.6 months without to 4.6 months with bevacizumab (P = .0002). Taken together, results of these trials suggest that the anti-angiogenesis story remains to be completely told in this disease.

NEW DIRECTIONS

Given the disappointments of the early era of targeted therapy in pancreatic cancer, the role of cytotoxics currently remains undiminished in this malignancy. Two new taxane derivatives are poised for phase III development. Nab-paclitaxel, a nanoparticle albumin-bound paclitaxel, has been combined with gemcitabine in an early phase I trial. Preliminary data, reported by Von Hoff et al,³⁵ showed interesting activity with partial responses and Ca 19-9 biomarker decline in a significant number of patients, and with the suggestion that SPARC (secreted protein and rich in cysteine) might be a surrogate of poor outcome. A different method of paclitaxel delivery has been chosen for Endotag-1 formulation. Endotag-1 consists of cationic liposomes and paclitaxel. Proposed mechanisms of anti-tumor activity include endothelial cell and vascular disruption as well as cytotoxic activity. A randomized phase II study has explored several dosing schedules in untreated patients with both locally advanced and metastatic disease.³⁶ Allowing for heterogeneity of the patient population, treatment duration, and dose levels, a suggestion of anti-tumor activity has been observed and further development is planned at this time. Other ongoing or imminently accruing phase III trials are summarized in Table 4. Additional areas of development of new therapeutics include insulin-like growth factor 1 receptor (IGF- 1R) signaling inhibitors combined with gemcitabine/erlotinib (Southwest Oncology Group [SWOG] phase II) and other kinase inhibitors,³⁷ PI3/Akt signaling inhibition, PARP inhibition,¹⁸ c-Met inhibition, and antagonists of the tumor microenvironment, to name a few.

Table 4.

Selected ongoing phase III trials in advanced pancreas cancer

	Target	Sponsor
Gemcitabine +/−Axitinib (N= 596)	VEGFR	Pfizer
Gemcitabine +/−Aflibercept (N= 630)	VEGF-A	Sanofi
Gemcitabine +/−Sorafenib (N= 104)	VEGFR, Raf	France
Gemcitabine +/−Sunitinib	VEGF, PDGFR	Pfizer
Gemcitabine vs. S1 vs. Gemcitabine + S1	Fluoropyrimidine	Taiho
Gemcitabine vs. FOLFOXIRI (N= 260)	Cytotoxic targets	FNCLCC-ACCORD
Gemcitabine +/−Curcumin, Celecoxib	COX-2	Tel Aviv Medical Center
Gemcitabine/Capecitabine +/−GV1001	Telomerase (vaccine)	Telovac
Gemcitabine +/−Nab-Paclitaxel	Tubulin	Abraxis

Abbreviations: COX-2 = cyclooxygenase-2; FNCLCC = Fèdèration Nationale des Centres de Lutte Contre le Cancer/National Federation for the Fight Against Cancer; FOLFOXIRI = 5-fluorouracil, folinic acid, oxaliplatin, irinotecan; N = No. patients; PDGFR = platelet-derived growth factor receptor; VEGF = vascular endothelial growth factor; VEGFR = VEGF receptor.

Much hope has been placed on the identification of mouse models for pancreatic adenocarcinoma that mimic the human condition with regard to morphology and clinical behavior, both significant limitations of earlier murine models of this disease. The access of a reliable in vivo model may provide opportunities, previously unavailable, for in vivo drug testing, and may be useful in identifying agents with a potential role in prevention of the disease.³⁸ A preliminary analysis of 24 pancreatic genomes has provided further insights into the molecular biology and underpinnings of pancreatic cancer. Jones et al observed an average of 63 genetic alterations, mostly point mutations, in twelve major signaling pathways,⁵ many of which are being actively targeted by drugs in development. The hope here, however, is that clear directions for therapeutic development will come about with greater understanding of the onconeogenesis of this most challenging of human malignancies.³⁹