Background
The prognosis for pancreatic adenocarcinoma remains dire, with a 5-year survival rate of <5 % [
1]. The clinical response to gemcitabine is significant, but its effect on overall survival is modest [
2]. In addition, gemcitabine’s toxic side effects can be dose limiting and some pancreatic tumors are inherently resistant to the drug. At present, conventional chemotherapy for pancreatic cancer consists of maximum tolerated doses (MTD) of gemcitabine, in which the patient is given the highest possible drug dose that does not cause life-threatening side effects. The inherently toxic nature of MTD treatment requires drug-free breaks to allow the patient to recover from systemic drug toxicities before resuming the treatment. Unfortunately, the tumor often reestablishes itself during the drug-free breaks, and sometimes with acquired resistance to gemcitabine, that renders subsequent cycles of treatment ineffective. More effective strategies for treating and controlling pancreatic cancer are thus needed.
An alternate treatment regimen for pancreatic cancer under investigation is metronomic chemotherapy where low doses of a cytotoxic drug are administered frequently without prolonged drug-free breaks [
8]. Drug doses lower than MTD, even if given more frequently and without rest breaks, are tolerated better by patients and cause fewer side effects. Interestingly, studies also indicate that drug resistant tumors can still respond to metronomic dosing of the same drug [
3,
4], implying that direct cytotoxic kill of cancer cells is not the only mechanism of tumor control. Cytotoxic agents also target proliferating endothelial cells in the tumor [
5], and it is not unreasonable that metronomic chemotherapy exerts non-specific effects on tumor vasculature that limits the supply of oxygen and nutrients and subsequently impedes tumor growth [
6‐
8]. Moreover, the lack of treatment breaks in metronomic chemotherapy also means that the continuous presence of drug prevents damaged vasculature from recovering [
3]. The effects of metronomic chemotherapy on tumor vasculature are largely unknown except for a handful of studies where increased vessel perfusion was observed [
9‐
11].
We now report here the different effects that metronomic gemcitabine and DC101 treatment have on two patient derived pancreatic tumor lines grown orthotopically in mice. Our data show that metronomic therapy with gemcitabine not only improves the tumor’s vascular function transiently, but also significantly retards cell proliferation and metabolism in primary human pancreatic cancer xenografts. In contrast, DC101, a monoclonal antibody that targets mouse vascular endothelial growth factor receptor-2 (VEGFR-2) decreases vascular density as expected but had no effect on tumor growth in the same models. The molecular and in situ physiological data reported here provide more insight on the activity of metronomic gemcitabine in primary pancreatic tumors. The information garnered suggests that the activity of Met-Gem treatment is not only cytotoxic, but also affects tumor vasculature effects as well. The data further suggest a role for non-invasive imaging technologies in monitoring changes in the tumor microenvironment, which could be used to guide the development of more effective therapies and dosing parameters specifically for pancreatic cancer.
Discussion
The aim of this study was to better understand the anti-tumor effects of metronomic gemcitabine (Met-Gem) in the context of pancreatic adenocarcinoma. Our data clearly show that Met-Gem is very effective at controlling pancreatic tumor growth, with the average volume of Met-Gem-treated tumors almost 10 times smaller than that of Veh-ctrl tumors (Fig.
1a). The avidity of Met-Gem-treated tumors for FDG, a radiolabelled glucose analog, was measured with PET to evaluate the tumors’ glucose requirements and metabolic activity as a surrogate marker for in situ cell viability and proliferation. The PET data show that FDG is distributed homogenously in Met-Gem-treated tumors, which have few necrotic areas, but that the tumors were significantly less avid for FDG compared to Veh-ctrl tumors. Interestingly, FDG uptake in tumors treated with Met-Gem was lower as soon as day 3, whereas tumor volume data gave no indication of any response to metronomic treatment at this early time point (Fig.
1a, c). The levels of Ki67 and TUNEL staining in frozen sections from the Met-Gem-treated tumors (Fig.
2) were also lower than those in the Veh-ctrl tumors, consistent with the FDG data (Fig.
1). Met-Gem treatment, therefore, appears to exert a cytostatic effect on the tumor wherein the cancer cells are still alive, but not actively proliferating.
Metronomic dosing with other drugs has been reported to decrease tumor perfusion [
3,
5,
30,
31], thus treatment induced changes and function in the tumors’ vasculature were examined using immunohistochemistry and DCE-MRI. The density and functionality of the tumor vasculature in our study were quantified, and the data (Fig.
4) indicate that Met-Gem treatment increases vascular density and improved function compared to Veh-ctrl tumors. Hypoxic cells were virtually absent in Met-Gem-treated tumors (Figs.
3,
4) and this is likely due to a more functional vascular system which delivers more oxygen and lower metabolic rates. Similar findings have also been reported in a breast cancer model treated with orally administered, low-dose gemcitabine [
11].
The
K
trans data (Fig.
5) indicate blood flow is homogenous throughout the tissue following Met-Gem treatment. Moreover, the average
K
trans values increase by day 7 compared to Veh-ctrl tumors even though the average tumor volume in all groups are similar. The
K
trans values, however, decrease by day 21, suggesting that the changes in tumor perfusion due to Met-Gem is dynamic and that the initial increase in perfusion observed is transient. Other studies report that treatment with Met-Gem decreases tumor perfusion [
30], but this may be a reflection of when the scans were carried out with respect to the treatment as few longitudinal studies such as ours have been carried out.
Since Met-Gem treatment had such dramatic effects on the density and function of tumor vessels, we treated the same tumor models with DC101 to examine the effects of a dedicated antiangiogenic agent on tumor growth, metabolism, and vascular function. DC101, the murine analog of ramucirumab [
32], was used because the vasculature that develops in the orthotopic tumors is of murine origin. DC101 therefore only targets actively growing tumor vasculature and has no effect on the human cancer cells. Our results indicate that DC101 treatment is much less effective than Met-Gem. Treatment with DC101 only reduced the volume of PaCa13 tumors by about 3 times after 21 days compared to Veh-ctrl tumors. The uptake and distribution of FDG in Veh-ctrl- and DC101-treated tumors were also similar—uptake was concentrated at the periphery, and the tumors had large central necrotic areas with no uptake. Moreover, the avidity of the DC101-treated and Veh-ctrl tumors for FDG was higher than that in Met-Gem-treated tumors indicating that the DC101-treated tumors, albeit highly necrotic, were still actively proliferating at the periphery, a pattern typical of rapidly growing tumors (Fig.
1). Similarly, levels of cell proliferation and apoptosis (Ki67 and TUNEL, respectively; Fig.
2) in DC101-treated tumors were similar to those present in Veh-ctrl tumors. Our data further show that DC101 was toxic to the vasculature (Fig.
4), but the consequences were not detrimental to tumor growth at least in these models and time points. As expected, DC101 reduced vascular density in the primary pancreatic tumor xenografts in our study as reported previously [
13]; the same group also reported that DC101 reduced tumor volume [
13], whereas no significant effects on tumor volume were seen in our study. Unfortunately, it is not possible to draw any definite conclusions between the two studies as our group used primary orthotopic human tumor xenografts [
9], which bears the histology of the original resected patient tumor, whereas the other group used tumors derived from cell lines which tend to contain homogeneous sheets of cancer cells and a well distributed vascular system.
DC101 treatment did not change in situ tumor perfusion as evaluated with DCE-MRI over time either, and in fact, the median
K
trans values in tumors treated with DC101 were similar to those in Veh-ctrl tumors. Perfusion in Veh-ctrl- or DC101-treated tumors was also limited to the periphery of the tumor (Figs.
5a, b). The aggregate data for DC101 treatment indicate that specifically targeting tumor vasculature (via VEGFR-2) in these primary pancreatic tumors has a minor effect in one tumor line, but that in general, the tumors in both groups behaved like Veh-ctrl tumors and continued expanding outwards from a central necrotic core.
Since Met-Gem and DC101 affected the tumor vasculature, and vascular growth and remodeling is a delicate balance of pro- and anti-angiogenic factors [
33], selected molecular factors implicated in vascular reorganization were assayed (Additional File 2). Met-Gem reduced the levels of proangiogenic factors VEGF and PDGF-BB compared to Veh-ctrls, suggesting that the treatment has some antiangiogenic effect. VEGF is known to control endothelial cell growth, migration, and survival, but overexpression of VEGF typically causes abnormal, random, and disorganized vasculature [
34,
35]. Since the levels of VEGF decreased with Met-Gem treatment, the tumor vasculature would be expected to become less haphazard and disorganized as seen here. The results reported here would appear consistent with the hypothesis originally put forth by Jain et al. [
35] that metronomic therapy may induce vascular normalization. DC101 treatment increased the levels of both mouse and human VEGF, indicating that VEGFR-2 was successfully blocked, but somewhat surprisingly did not change vasculature patterns. Tumors are known to become resistant to antiangiogenic treatment by using alternative pathways to compensate for the inhibition of VEGFR-2 [
36]. However, in this case, no change in P
lGF, PDGF-BB, and SDF-1α was seen over the 3-week treatment period. It appears that the primary pancreatic tumors used in our study can survive VEGFR-2 inhibition without activating other pathways.
The differences in response to DC101 (a targeted drug) and Met-Gem (low-dose cytotoxic) provide some insight into how each affects tumor growth. The observations from tumors treated with Met-Gem are somewhat paradoxical since the tumors appear ‘healthier’—they had more functional blood vessels, less necrosis, and virtually no hypoxia—and yet the tumors did not proliferate, or at least proliferated slowly compared to Veh-ctrl- or DC101-treated tumors. Gemcitabine would initially cull proliferating endothelial and cancer cells to leave behind a tumor with mature blood vessels that are more functional and fewer cancer cells. The continuous delivery of gemcitabine subsequently appears to dampen the proliferative capacity of the cells. Since the drug is present more often during metronomic therapy, the cancer cells and endothelial cells may have no chance to recover as they do during chemotherapy breaks in conventional MTD therapy. The result is a tumor in stasis where cells are still viable, but less proliferative than normal. In contrast, targeting VEGFR-2 with DC101 in these primary tumors may not be effective because the VEGF pathway is not an important therapeutic target in pancreatic cancer [
37], or because alternative pathways may be utilized. Similar results have also been reported in the clinic where targeted anti-VEGF treatment in pancreatic cancer was not successful, and failure of DC101 to Veh-ctrl tumors in our study mirrors these findings where inhibition of a single target was insufficient to achieve tumor control [
38].
At present, the consequences of long-term Met-Gem treatment are unknown. The tumor may eventually be eradicated, or it could become refractory to Met-Gem and re-grow. However, it is tantalizing to speculate that if the treatment holds tumor growth in check indefinitely, Met-Gem treatment could be used as an additional therapeutic option where the goal is to attain a cytostatic state rather than to reduce tumor volume at all costs. Several intriguing implications arise from the effects of Met-Gem. Using non-invasive imaging technologies, as in our multi-modality study, may also be useful in determining therapeutic opportunities that arise from Met-Gem treatment. The homogenous perfusion due to improvements in vascular function may provide better access for a second drug to all cancer cells as was previously shown by our group [
17] and could potentiate its cell-killing effects. DCE-MRI could be used to determine how sequential drug treatments could be used. The classic end point for drug activity is reduction of tumor volume [
39]. However, new therapies that induce a cytostatic state [
40] require other end points for the assessment of response and efficacy [
41]. FDG-PET may be useful as a complementary measure of drug activity [
42‐
44] because the technique is non-invasive, applicable to multiple scans, and measures a tumor’s metabolic activity which may change before tumor volumes are affected [
45,
46]. This is exemplified by the FDG-PET scans in our study which show decreased cell viability before changes in tumor volume are observed after 3 days of treatment. There has only been a handful of studies of FDG-PET response with metronomic treatment of other drugs and other cancer models [
47,
48], and its utility as an early surrogate response marker remains promising. We are the first to study FDG uptake with metronomic gemcitabine in pancreatic cancer and further studies will be required to confirm the predictive benefits of FDG-PET in this setting.
We have previously shown that Met-Gem has better efficacy at lower doses than conventional maximum tolerated doses [
9]. In this study, we further confirmed that Met-Gem treatment is cytostatic and improves vasculature function transiently and that changes in the tumors’ viability can be detected before changes in their volume. DC101 treatment indicated that a treatment specific for VEGFR-2 may be less effective if the targeted pathway is no longer important in disease progression or if other pathways can compensate for its activity. However, a therapeutic strategy such as Met-Gem which provides the continuous presence of low-dose gemcitabine may control not only the proliferation of endothelial cells, but also cancer cells in a two-pronged attack. This approach appears to control cell growth well, as evidenced by the FDG-PET and Ki67 data, precisely because it is not targeted to a specific component of the disease.
There is clinical interest in combining metronomic therapies with dedicated anti-angiogenic agents as a therapeutic strategy for cancer treatments [
49‐
53]. The therapeutic effects of combinations such as DC101 and metronomic vinblastine in neuroblastoma [
8], DC101 with metronomic cisplatin or doxorubicin in breast cancer [
54], and bevacizumab with metronomic irinotecan in colorectal cancer [
55] have been examined in preclinical models. Phase I/II clinical studies have also been carried out in glioblastoma with bevacizumab and metronomic irinotecan [
56,
57] and breast cancer with bevacizumab and metronomic cyclophosphamide [
58]. The studies showed that the combination treatments are well tolerated, and in some cases produced stable disease. However, few conclusions can be made about the mechanisms of the combination therapies due to variation in study methods and small population patients. The consensus at the moment is that a better understanding of how metronomic therapies interact with anti-angiogenic therapies is needed and that more predictive biomarkers and imaging techniques are required to facilitate the sequencing (or monitoring effects) of the two treatments [
49]. In our study, the data indicate that Met-Gem causes a transient period of improved perfusion and low hypoxia which could be advantageous for the delivery of a second drug, or radiation therapy, respectively. Met-Gem and DC101 were not combined in this study and so we are unable to comment on any potential synergies between the two treatments. However, our study also shows the utility of imaging technologies to assess changes in the physiology of the tumor to guide changes in initial treatment to take advantage of transient treatment opportunities in a combination or sequential setting.