Background
Pancreatic adenocarcinoma is the fourth leading cause of cancer-related deaths in the United States, with mortality nearly equal to incidence. Less than 5% of patients survive five years from the time of diagnosis, and the median survival time is less than 6 months. This year alone, pancreatic cancer will result in approximately 40,000 deaths in the United States [
1]. At the time of diagnosis, surgical resection is unfortunately not an option for many patients due to the advanced stage of disease and distant metastases. In addition, current chemotherapeutic strategies are largely ineffective because of either innate or acquired chemoresistance. Gemcitabine (2’,2’-difluorodeoxycytidine) is the drug of choice for the primary treatment of unresectable pancreatic cancer and adjuvant treatment following resection of pancreatic cancer, but measurable responses are not observed in the majority of patients [
2,
3]. Fortunately, gemcitabine, compared to other chemotherapeutic agents, is relatively non-toxic [
4]; however, its use as chemoprevention in patients with known precancerous lesions has not been explored. Other chemotherapeutic agents, e.g., paclitaxel, have recently been used in patients with precancerous pancreatic lesions with some evidence of regression [
5].
To aid in the search for effective prevention and intervention strategies, clinically relevant animal models are needed and have recently been developed [
6]. For example, a genetically engineered mouse model of pancreatic cancer with targeted expression of mutant
KRAS
G12D
and
Trp53
R172H
genes has been shown to recapitulate human pancreatic neoplasia, from premalignant lesions to invasive cancer and metastasis [
7]. The
LSL-Kras
G12D/+
; LSL-Trp53
R172H
; Pdx-1-Cre mice are a developmental model of pancreatic cancer in which adenocarcinoma form
de novo with close to 100% penetrance. In this mouse model, the Lox-Stop-Lox (LSL) sequence upstream of oncogenic
KRAS and mutant
Trp53 inhibits transcription and translation. Expression of Cre recombinase from the pancreatic-specific promoter Pdx-1, excision of the “Stop” sequences, and subsequent Cre-mediated recombination allow endogenous expression of the mutant Kras and p53 in progenitor cells of the mouse pancreas. Another advantage of this model is that the natural microenvironment of the pancreas is maintained. Thus, preclinical data from these types of animal models may be more predictive of human clinical outcomes.
Due to its critical role in inflammation and multiple tumorigenic processes, the transcription factor nuclear factor-kappaB (NF-κB) is a therapeutic target of interest for pancreatic cancer [
8,
9]. In addition, the p65 subunit of NF-κB, RelA, is constitutively active in human pancreatic adenocarcinoma tissue and in pancreatic tumor cell lines [
10]. It was recently demonstrated in a genetically engineered mouse model that constitutive NF-κB activation, by Kras through AP-1-induced overexpression of interleukin-1α (IL-1α), is required for the development of pancreatic cancer [
11]. These findings implicate NF-κB in the development and progression of pancreatic cancer. Furthermore, experimental evidence suggests that NF-κB may also be a suitable target for chemoprevention [
12,
13]. We have previously examined the anti-cancer activity of dimethylaminoparthenolide (DMAPT), which is a novel and orally bioavailable analog of parthenolide, a sesquiterpene lactone isolated from the medicinal herb feverfew (
Tanacetum parthenium) [
14]. In both xenograft and carcinogen-induced animal models of pancreatic cancer, DMAPT inhibits the activity of NF-κB and shows therapeutic promise in combination with the anti-inflammatory agents sulindac or celecoxib
in vivo[
15,
16].
We and others have also reported that the chemotherapeutic agent gemcitabine induces NF-κB activity in pancreatic cancer cells
in vitro, suggesting that NF-κB activation may play a role in chemoresistance to gemcitabine [
9,
17‐
20]. A viable strategy for improving the therapeutic response to gemcitabine may therefore involve suppression of the NF-κB pathway. In support, we recently demonstrated that DMAPT not only inhibits gemcitabine-induced NF-κB activation but also sensitizes pancreatic cancer cells to the anti-proliferative effects of gemcitabine
in vitro, indicating that the level of NF-κB activity modulates the gemcitabine response [
21]. Furthermore, in a heterotopic xenograft model, gemcitabine exposure activates NF-κB within established pancreatic tumors, suggesting that NF-κB suppression may also improve the anti-tumor effects of gemcitabine
in vivo[
21]. Most recently, we found that DMAPT and/or sulindac in combination with gemcitabine therapy can delay or prevent progression of premalignant pancreatic lesions in the less aggressive
LSL-Kras
G12D/+
; Pdx-1-Cre mouse model of pancreatic cancer [
22]. Due to the low incidence of pancreatic tumors in the
LSL-Kras
G12D/+
; Pdx-1-Cre mouse model, the clinical relevance of this delay on pancreatic tumor formation or metastasis could not be determined. Thus, the chemopreventative efficacy of the most effective combination DMAPT/gemcitabine was further evaluated in this survival study using the
LSL-Kras
G12D/+
; LSL-Trp53
R172H
; Pdx-1-Cre mouse model, which is characterized by near 100% incidence of pancreatic adenocarcinoma development.
Methods
Compounds
Gemcitabine (GEMZAR
®) was obtained from Eli Lilly (Indianapolis, IN). DMAPT [
14] was synthesized by reaction of parthenolide (Sigma-Aldrich, St. Louis, MO) with dimethylamine (Sigma-Aldrich, St. Louis, MO) and isolated as the fumarate salt.
LSL-KrasG12D/+; LSL-Trp53R172H; Pdx-1-Cre mouse model
This study was performed in compliance with federal Institutional Animal Care and Use Committee guidelines. Male
LSL-Kras
G12D/+
; Pdx-1-Cre mice (breeders kindly provided by Dr. Andrew Lowy, University of California, San Diego [
23]) were crossed with female
p53 LSL
R172H
(NCI-Frederick) mice to generate
LSL-Kras
G12D/+
; LSL-Trp53
R172H
; Pdx-1-Cre mice. At 1 month of age, mice were genotyped by PCR analysis of tail genomic DNA. For Kras
G12D, primers were as follows resulting in amplification products of 500 bp (wild-type) and 550 bp (mutant allele):
5′ wild type: GTCGACAAGCTCATGCGGG;
5′ mutant (LSL element): CCATGGCTTGAGTAAGTCTGC
3′ universal: CGCAGACTGTAGAGCAGCG
For Cre, the primers were as follows to generate a 475 bp amplification product:
5′: AGATGTTCGCGATTATCTTC
3′: AGCTACACCAGAGACGG
For p53R172H, primers were as follows generating amplification products of 166 bp (wild-type) and 270 bp (LSL element):
5′ mutant (LSL element): AGCTAGCCACCATGGCTTGAGTAAGTCTGC
5′ wild-type: TTACACATCCAGCCTCTGTGG
3′ universal: CTTGGAGACATAGCCACACTG
This breeding scheme resulted in ~12% positive mice which were eligible for rolling enrollment in the study.
At 1 month of age,
LSL-Kras
G12D/+
; LSL-Trp53
R172H
; Pdx-1-Cre mice were randomized into treatment groups (placebo, DMAPT, gemcitabine, DMAPT/gemcitabine). Placebo (vehicle = hydroxylpropyl methylcellulose, 0.2% Tween 80 [HPMT]) and DMAPT (40 mg/kg body weight in HPMT) were administered by oral gastric lavage once daily. Gemcitabine (50 mg/kg body weight in PBS) was administered by intraperitoneal injection twice weekly. Mouse weight was monitored weekly. Treatment was continued until mice showed signs of lethargy, abdominal distension or weight loss at which time they were sacrificed. Successful excision-recombination events were confirmed in the pancreata of mice by detecting the presence of a single LoxP site [
24].
Upon necropsy, the presence and size of gross pancreatic tumors and metastases were noted. The presence of multiple tumors was determined both by gross examination and palpation of the pancreas since the boundaries between multiple large tumors can be difficult to delineate in hematoxylin and eosin (H&E)-stained specimens. In these cases, gross tumor dimensions were used for analysis. For the smaller tumors, identification of pancreatic ductal adenocarcinoma, as well as their dimensions, was confirmed upon review of ten consecutive H&E-stained sections per pancreas (100 μm apart) by a pathologist blinded to the experimental groups. Tumor volume was calculated using a modified ellipsoidal formula, 1/2(length × width2). Liver, kidney, and lung were also examined for signs of drug toxicity. Pancreatic, liver and lung tissue pieces (3 mm) were frozen in liquid nitrogen and stored at −80°C for analysis; the remaining tissues were fixed in 10% formalin (Sigma, St. Louis, MO) and paraffin-embedded for H&E staining and immunohistochemistry. Serial liver and lung sections (10–15 sections each, 100 μm apart) were examined for metastases. Blood (~ 1 ml) was obtained by cardiac puncture, mixed with 25 μl anticoagulant (EDTA [2 g]/NaCl [0.8 g] in 100 ml water, pH 7.4) and centrifuged (2800 rpm, 15 minutes, 4°C). Plasma aliquots were frozen at −80°C.
Luciferase-expressing p53f/f; LSL-KrasG12D;lucl/l; Pdx-1-Cre mouse model
Floxed-p53/LSL-Kras
G12D
/floxed-STOP-luciferase male mice (designated
p53
f/f
; LSL-Kras
G12D
;luc
l/l
, breeders kindly provided by Dr. Robert Bigsby, Indiana University [
25]) were crossed with
Pdx-1-Cre mice to generate
p53
f/f
; LSL-Kras
G12D
;luc
l/l
; Pdx-1-Cre mice, in which p53 is deleted and mutant Kras and luciferase are expressed. After genotyping, mice were randomized into single agent treatment groups (placebo, DMAPT and gemcitabine) at two months of age as described above. Following injection with D-luciferin (60 mg/kg, in 0.3 mL PBS) into the intraperitoneal cavity, imaging was performed using the NightOWL optical Imager (LB981, Berthold) to detect luciferase expression within the pancreas.
Mouse PanIN (mPanIN) analysis
One section per pancreas, with maximal exposure (greater than 75%) of the pancreas (5 μm) was cut, stained with hematoxylin and eosin (H&E), and examined microscopically for lesions. Sections from all animals in each treatment group were analyzed. mPanINs were counted in a blinded manner according to previously established criteria [
26,
27]. The highest grade lesion in the individual pancreatic lobules within the entire pancreas from each animal was identified for quantification. The percent normal ducts, mPanIN-1, mPanIN-2 and mPanIN-3 lesions was determined relative to the total number of lesions counted per pancreas.
Immunohistochemistry and staining
Immunohistochemistry was performed with primary antibodies NF-κB/p65 (1:400, Lab Vision Corporation, Fremont, CA), phospho-ERK (1:500, Cell Signaling, Danvers, MA), Ki67 (1:50, Dako North America, Carpinteria, CA) and CD31 (1:20, Dianova, Hamburg, Germany). Briefly, slides were deparaffinized and hydrated in running water. Slides were placed in Antigen Retrieval Citrate Buffer pH 6.0 (Dako North America) in a pressure cooker for 15 minutes before placing in 3% H2O2 for 10 min. All slides were placed in Protein Block (Dako North America) for 15 min, incubated with appropriate primary and secondary antibodies, and then counterstained. Percent NF-κB and Ki67 staining was quantified by counting the number of positively staining tumor cells in two fields with the highest density of staining per pancreas and expressed relative to the total number of cells within the field. Intratumoral CD31 staining was quantified using ImageScope software and the positive pixel count algorithm (Aperio Technologies, Vista, CA).
Masson’s Trichrome was performed using Sigma-Aldrich Accustain Trichrome Stains (Masson, kit No. HT-15) and quantified using ImageScope software.
Cytokine analysis
Mouse plasma obtained at the time of sacrifice was analyzed using the Bio-Plex Pro™ mouse cytokine 23-plex immunoassay (Biorad, Hercules, CA) and the Bio-Plex 200 System, as recommended by the manufacturer. Samples were diluted 1:4 with sample diluent supplied in the Bio-Plex kit prior to analysis. Analyte values that were out of range or with a low bead count (< 50) were excluded from analysis.
Statistical Analysis
Median survival was determined by the Kaplan-Meier method and analyzed by the log-rank test. Comparisons between placebo and treatment groups were analyzed by ANOVA with Dunnett’s post-test or Student’s t-test (Prism 5.0 software, Graphpad, San Diego, CA). For incidence, Fisher exact test was performed. P < 0.05 was considered significant (two-tail, 95% confidence interval).
Discussion
In the present study, we evaluated the efficacy of the bioavailable NF-κB inhibitor DMAPT and gemcitabine using a genetically engineered and clinically relevant mouse model of pancreatic cancer. Much research supports a central role for NF-κB in inflammation and pancreatic tumorigenesis [
8,
9,
11]. We have shown both
in vitro and
in vivo that pharmacological suppression of NF-κB by parthenolide or DMAPT has anti-cancer activity [
15,
16,
33]. Gemcitabine, the current standard-of-care therapy for pancreatic cancer, has limited clinical benefit alone, and recent work has indicated that induction of NF-κB activity by gemcitabine may be involved in chemoresistance [
9,
17‐
20].
In this study, we report that gemcitabine and DMAPT/gemcitabine significantly increase median survival and decrease the incidence and multiplicity of pancreatic adenocarcinomas in LSL-Kras
G12D/+
; LSL-Trp53
R172H
; Pdx-1-Cre mice. Ki67, a cellular marker of proliferation, staining is reduced in both gemcitabine and the combination treatment groups, correlating with drug effects on tumor growth. The DMAPT/gemcitabine combination also significantly decreases tumor size as well as the incidence of metastasis to the liver. Although gemcitabine decreases tumor size compared to the placebo group, this effect is not significant. There is also no significant difference between the combination and gemcitabine alone groups with respect to tumor size. Furthermore, whereas liver metastasis in the combination group is significantly reduced compared to gemcitabine, the overall incidence of metastasis is not significantly different between the two groups due to metastasis occurring in the lung. Thus, it is not surprising that despite some additional anti-tumor effects of the combination, no further impact on survival is observed in our study when compared to gemcitabine treatment alone. Clinically, such differences, for example in tumor size, may be beneficial and translate to improved quality of life for the patient, without prolonging survival. Many variables (drug bioavailability and dose, treatment plan, specific mutations present etc.) can influence the effects of the drugs on tumor response in vivo. Thus, further optimization of the agents, experimental design, or animal model may be required to observe differential effects on survival.
No significant difference in premalignant PanIN formation is apparent between the treatment groups. This is likely due to the late stage of disease at the time of sacrifice and development of large pancreatic adenocarcinomas especially in the placebo and DMAPT treatment groups since in many cases, little or no histologically normal or premalignant pancreas was present. Further analysis specifically of pancreata lacking primary tumors reveals the presence of mainly normal ducts or PanIN-1 lesions, with only a few PanIN-2 and −3 lesions. This suggests that not only pancreatic tumors but also high grade pancreatic lesions are prevented in these animals. It will be of interest to further characterize the profiles of these “responders”.
We also report that gemcitabine treatment significantly increases the levels of the inflammatory cytokines IL-1α, IL-1β, and IL-17 in mouse plasma. The expression of these three genes is regulated by NF-κB, so this effect is consistent with gemcitabine-induced NF-κB activation that has been previously reported [
34,
35]. Treatment with DMAPT/gemcitabine reduces the levels of these cytokines to that of DMAPT alone. Furthermore, DMAPT and/or the combination DMAPT/gemcitabine significantly decrease the levels of the inflammatory cytokines IL-12p40, MCP-1, MIP-1β, eotaxin and TNF-α, all of which are NF-κB target genes [
34]. Thus, the effect of DMAPT and DMAPT/gemcitabine on the expression of these cytokines suggests systemic suppression of the target NF-κB in these treatment groups. Target suppression was also demonstrated in our recent study with
LSL-Kras
G12D
; Pdx-1-Cre genetically engineered mice in which all mice were treated for 3 months and then sacrificed 3 hours following the last drug treatment; specifically, combination DMAPT treatment decreased NF-κB expression in the pancreatic lesions [
22]. Thus, at the 40 mg/kg dose, DMAPT consistently inhibits its target in models of either established or spontaneous tumors [
15,
16].
We previously showed that the percentage of normal pancreatic ducts was significantly increased by the combination of DMAPT/gemcitabine compared to placebo in the
LSL-Kras
G12D
; Pdx-1-Cre mouse model; additionally, the percentage of mouse pancreatic intraepithelial neoplasia-2 (mPanIN-2) lesions was significantly decreased by DMAPT/gemcitabine [
22]. This delay in PanIN progression in the
LSL-Kras
G12D
; Pdx-1-Cre mice may in part explain the ability of the combination DMAPT/gemcitabine to attenuate pancreatic tumorigenesis in the present study. Similarly, Fendrich et al. recently showed that treatment with the angiotensin-I-converting enzyme inhibitor enalapril or aspirin, that targets NF-κB, delays PanIN progression in
LSL-Kras
G12D
; Pdx-1-Cre mice and decreases pancreatic cancer development in the
LSL-Kras
G12D/+
; LSL-Trp53
R172H
; Pdx-1-Cre mouse [
36]. The combination of enalapril and asprin was not more effective than the single agents.
It has been previously reported by Olive et al. that gemcitabine therapy alone is ineffective in the
LSL-Kras
G12D/+
; LSL-Trp53
R172H
; Pdx-1-Cre mouse model [
37]. In contrast, gemcitabine alone in our study decreases tumor incidence in
LSL-Kras
G12D/+
; LSL-Trp53
R172H
; Pdx-1-Cre mice. In our study, gemcitabine (50 mg/kg, twice weekly) was administered beginning at 1 month of age until the time of sacrifice/death. In Olive’s study, mice bearing tumors ~5-10 mm in diameter were identified and enrolled for gemcitabine treatment (100 mg/kg, Q3Dx4 or every third day for 4 cycles). In the latter study, gemcitabine was given to tumor-bearing mice for a much shorter time (9 days total); in contrast, for our study, gemcitabine was administered prior to any palpable tumor formation (tumors ~5-10 mm in diameter can be detected by palpation) at 1 month of age until ill health necessitated sacrifice or death occurred (~6 months of treatment). The earlier intervention as well as longer length of treatment may account for the difference in response between these two studies. Thus, based on these findings, gemcitabine administered earlier may have some ability to alter the development of pancreatic cancer. We speculate this may be an effect of delaying PanIN progression and/or possibly improved drug delivery prior to full stromal maturity of the developing adenocarcinomas. Gemcitabine has relatively low toxicity in humans, and at the comparably lower doses used in this study, there was no measurable toxicity in this model. Thus, the use of low dose gemcitabine at the time of diagnosis or in high risk groups may have benefit and merits further investigation.
Conclusions
Although survival statistics for many cancers have improved in recent years, pancreatic cancer remains one of the deadliest diseases with a very poor survival rate due to the lack of early detection and absence of effective treatments. Patients at increased risk of pancreatic cancer include those with inheritable risk factors, patients presenting with premalignant pancreatic lesions, and individuals with multiple epidemiologic risk factors, including smoking, diabetes mellitus and obesity [
38]. Identifying and monitoring such patients may improve their clinical outcome by allowing earlier surgical intervention or intervention with agents that can delay or prevent the progression of disease. Once pancreatic cancer develops, however, curative treatments are limited. Thus, novel and effective treatment strategies remain a critical area of research. In this study, gemcitabine alone and/or the combination of DMAPT and gemcitabine significantly increase median survival as well as decrease tumor size, the incidence and multiplicity of pancreatic tumors, and metastasis to the liver in
LSL-Kras
G12D/+
; LSL-Trp53
R172H
; Pdx-1-Cre mice. The preclinical evidence presented here supports further evaluation of either gemcitabine alone or in combination with DMAPT or related agents as intervention to delay the progression of pancreatic cancer.
Competing interests
P.A.C. has a financial interest in Leuchemix, Inc. All others do not have any competing interest.
Authors’ contributions
MYS designed the study, participated in the animal treatment, performed analyses, and drafted the manuscript. HW bred and genotyped the animals and participated in animal treatment. KS performed the imaging analysis and helped revise the manuscript. NA determined histopathology of the specimens. PC synthesized the DMAPT and helped revise the manuscript. CMS participated in study design and helped draft the manuscript. All authors read and approved the final manuscript.