© 2003 AlphaMed Press
The Medical Management of Pancreatic Cancer: A ReviewDepartment of Oncology, Belfast City Hospital, Belfast, Northern Ireland Correspondence: Martin Eatock, M.D., Department of Oncology, Belfast City Hospital, Lisburn Road, Belfast BT9 7AB, Northern Ireland. Telephone: 02890-329241 ext 3911; Fax: 02890-263744; e-mail: MartinEatock{at}bch.n-i.nhs.uk
Access and take the CME test online and receive one hour of AMA PRA category 1 credit at CME.TheOncologist.com
Pancreatic carcinoma is a commonly occurring cancer that tends to present late in its course when potentially curative surgical treatment is not possible. The majority of patients are, therefore, candidates for systemic therapy. We review the patient and disease-related factors that contribute to the difficulties in the medical management of this condition and discuss new methods of assessing response to treatment, including the introduction of more clinically relevant novel end points such as clinical benefit response. We review the current trial literature examining the use of conventional cytotoxic agents in this disease, both as single agents and in combination. We also review the use of more novel targeted agents and examine their potential utility in this disease. The use of the farnesyl transferase inhibitors, matrix metalloproteinase inhibitors, epidermal growth factor receptor antagonists, and angiogenesis inhibitors is discussed. Key Words. Pancreatic ductal carcinoma • Antineoplastic agents • Tissue inhibitor of metalloproteinases • Epidermal growth factor receptor • ras gene • Angiogenesis inhibitors
Adenocarcinoma of the pancreas, with an annual incidence of 910 cases per 100,000 per year, is a relatively common oncological diagnosis [1]. The tumor has a poor prognosis, with an associated 2-year survival rate of 10% [2]. Surgery is the only potentially curative treatment but, unfortunately, due to the tendency of the tumor to present late in its course, less than 20% of patients have resectable disease at presentation. The majority of patients with pancreatic cancer, therefore, are potential candidates for systemic management. For a number of reasons, the medical management of this disease presents a considerable therapeutic challenge to oncologists. First, almost 70% of patients are 65 years or older at diagnosis, so many will have premorbid conditions that limit their treatment options [2]. In addition, more than 80% of patients have disease-related symptoms, such as pain, asthenia, weight loss, and poor performance status, that limit the ability to deliver potentially toxic chemotherapy [3]. As well as these patient-related factors, adenocarcinoma of the pancreas is less chemosensitive than other commonly occurring solid malignancies, with best response rates to conventional agents of less than 10% [4]. The reasons for this chemoresistance are poorly understood. There are also specific difficulties in assessing the activity of anticancer agents in this disease. Traditionally, end points such as response rate and survival have been used as markers of useful clinical activity in the assessment of new agents. Pancreatic tumors, however, are often associated with a dense desmoplastic reaction, which may be difficult to differentiate from tumor on conventional imaging. This makes it difficult or impossible to assess response using conventional criteria [5], and new methods for assessing activity in this disease are under investigation to ensure that agents are not wrongly discarded as inactive. Positron emission tomography (PET) scanning, for example, has already been demonstrated to have some utility in differentiating benign from malignant pancreatic lesions [6], and dynamic imaging with PET scanning may allow a more accurate radiological assessment of response [7]. Tumor markers such as serum CA19-9 may also have utility in assessing disease response if they are elevated prior to commencing treatment [8]. Furthermore, it is being increasingly recognized that worthwhile palliation of symptoms and improved quality of life may result from chemotherapy even in the absence of an improvement in survivalthe traditional end point of phase III trials. Several recent trials have used novel end points to improve clinical relevance in the assessment of new agents. Some have reported response rates in individual parameters such as performance status, pain, or weight gain [9]. Other investigators have developed tools to integrate information regarding response in these different parameters to give a single end point assessing clinical benefit. Clinical benefit response (CBR) is one such end point that has now been used as the primary outcome in a number of studies [10, 11]. CBR is a composite analysis of pain, performance status, and weight gain. A positive CBR requires improvement in at least one of these parameters maintained for at least 4 weeks with no deterioration in the other parameters. While these studies have added to our understanding of the utility of chemotherapy in pancreatic cancer, the variation in end points used to examine clinical benefit among studies makes comparisons difficult. The recent development of a pancreatic cancer-specific quality-of-life (QOL) module (QLQ-PAN 26), supplementing the European Organization for the Research and Treatment of Cancer core cancer QOL questionnaire (QLQ-C30), may represent a major step forward in standardizing QOL assessment in this disease [12]. In addition to the problems outlined above, the optimum evaluation of novel agents poses several other problems in clinical trial design. First, many novel agents target specific molecular pathways. In order to identify those patients most likely to benefit, assays to detect the target protein and/or downstream markers of biochemical response must, therefore, be identified. This usually requires tumor biopsies, which may be difficult to obtain in pancreatic cancer patients. Second, novel clinical trial designs are required to evaluate cytostatic rather than cytotoxic agents. The conventional strategy of initially testing new agents in the advanced setting may not, therefore, be appropriate, as these agents may, perhaps, be optimally used in the setting of minimal residual disease. Caution must, therefore, be exercised before rejecting an agent as inactive based on a negative study in pancreatic cancer.
This review highlights the agents that have been examined in this disease over the last decade (summarized in Table 1
Gemcitabine is an antimetabolite analog of deoxycytadine, which requires activation to the nucleotide level for its activity [13]. It has been investigated in pancreatic cancer using a weekly regimen, administered as a short infusion. In phase II studies, response rates of 10%15% have been demonstrated [14, 15]. Despite this, 27% of patients experienced clinical benefit as determined by a prospectively defined composite assessment of pain, performance status, and weight. This suggests that clinical benefit may occur even in the absence of a radiological response [14]. Subsequently, a multicenter randomized phase III study was performed comparing weekly 5-fluorouracil (5-FU) with gemcitabine in a group of patients with locally advanced or metastatic pancreatic cancer [11]. While conventionally measured response rates for both agents were unimpressive (0% for 5-FU, 5.4% for gemcitabine), improvements in clinical benefit (23.8% versus 4.8%) and survival (median survival, 5.65 months versus 4.41 months; survival rate at 1 year, 2% versus 18%) were noted for gemcitabine. As a result of that study, gemcitabine has become the standard of care for the treatment of pancreatic cancer. It is possible that the effects of gemcitabine in this disease may be improved by optimizing the schedule of administration. Gemcitabine is phosphorylated intracellularly to convert it to its active metabolite. This is a saturable process, and there is preclinical and clinical evidence that the drug may be more active when administered using a fixed dose rate of 10 mg/m2/min [16, 17]. Studies are ongoing investigating this approach in pancreatic cancer [18, 19]. Troxacitabine is a newly developed cytadine analog [20]. Unlike other naturally occurring and manufactured D-stereoisomeric forms, this L-stereoisomeric nucleoside analog does not require nucleoside transporters for entry into the cell and is not a substrate for cytidine deaminasean enzyme that degrades other nucleoside analogs and has been linked with resistance to these agents [21]. Troxacitabine has preclinical activity in pancreatic tumor models [22] and comparable activity with gemcitabine in a phase II study in pancreatic cancer [23].
5-FU is a pyrimidine analog whose antineoplastic activity is related to its ability to inhibit the enzyme thymidylate synthase and also to directly inhibit of DNA and RNA synthesis. Bolus administration of 5-FU has little activity in this disease [11]. As 5-FU is a cell-cycle-specific agent with a short duration of action, investigators have attempted to improve the efficacy of this drug using infusional or chronomodulated administration or with the addition of folinic acid (FA). Unfortunately, there has been no clear evidence of additional clinical benefit with these strategies [2427]. Therefore, manipulations that have been proven to improve the activity of 5-FU in other cancers remain of uncertain benefit in pancreatic cancer. Capecitabine is a fluoropyrimidine prodrug, which can be administered orally and which is activated preferentially in tumors [28, 29]. Its mode of action has been likened to that of infusional 5-FU, and its preferential activation in tumors results in an improved side effect profile [30, 31]. In a phase II study in chemotherapy-naïve pancreatic cancer patients with locally advanced or inoperable pancreatic cancer, a response rate of 7.3% and a CBR rate of 24%, comparable with that observed with gemcitabine, were noted [10]. Clearly, comparison with gemcitabine in a phase III study is warranted. Pemetrexed is a rationally designed antifolate with a number of potential advantages over 5-FU [32]. Observation of partial responses in patients with pancreatic cancer in a phase I trial [33] led to investigation of this drug in the phase II setting [34]. A somewhat disappointing response rate of 5.7% was noted; however, a median survival of 6.5 months, with 28% of patients alive at 1 year, was observed, suggesting that this agent may have some useful activity. Perhaps more interesting is the preclinical observation of synergy with a number of other chemotherapeutic agents including platinum analogs, 5-FU, and gemcitabine [35]. Investigations of pemetrexed in combination with other cytotoxic agents in pancreatic cancer are currently ongoing [36].
The taxanes are a class of spindle inhibitors, which are believed to exert their effect by inhibiting the dynamic process of microtubule assembly [37]. Preclinical activity has been demonstrated in pancreatic cancer models [38, 39], and taxanes have subsequently been examined in clinical trials. Three phase II trials have examined the use of docetaxel in pancreatic cancer. A dose of 100 mg/m2 on an every-3-weeks schedule resulted in a response rate of 15%, but CBR was not assessed [40]. A second phase II study examined the activity of docetaxel, 100 mg/m2, administered every three weeks with G-CSF support [9]. Although a response rate of only 6% was demonstrated, 36.4% of patients were still alive at 1 year and 67% of patients experienced a CBR. These results must be interpreted with caution, however, as the median survival in patients with progressive disease was 24 weeks, reflecting the strict criteria for entry into this study. Furthermore, the tools used to assess QOL differed from those used by Burris et al. in the gemcitabine study [11], making comparisons difficult. A further study suggested that docetaxel at 60 mg/m2 was inactive [41]. Paclitaxel has also been assessed for activity in this tumor. In an initial phase II trial in 14 patients using standard dose paclitaxel (175 mg/m2) given every 3 weeks, no responses were observed [42]. Dose-intense paclitaxel treatment (250 mg/m2) with G-CSF support resulted in a response rate of 8% with a median survival of 5 months [43]. Although no prospective assessment of CBR was made, 85% of patients reported problems with asthenia and fatigue, and 74% reported problems with nausea, vomiting, or anorexia, suggesting that the toxicity of the regimen probably outweighs any potential clinical benefit. In a retrospective study of weekly paclitaxel as second-line therapy in 18 patients with pretreated pancreatic cancer, one patient achieved a complete response while five achieved stable disease with an acceptable toxicity profile [44]. A number of new agents with similar mechanisms of action to the taxanes are in early clinical development. In preclinical studies, BMS-247550, an epothilone derivative, appears to be twice as potent as paclitaxel in promoting tubulin polymerization, and cell line studies indicate that this agent may have activity in paclitaxel-resistant pancreatic cancer cell lines [45]. Available evidence at present, therefore, suggests that taxanes have limited activity in pancreatic cancer.
Camptothecin analogs act by stabilizing the cleavable complex between topoisomerase I and the free 3' phosphate group of DNA [46]. The resulting enzyme-linked DNA breaks cannot be religated resulting in S-phase cytotoxicity. Three analogs, irinotecan, 9-nitrocamptothecin, and exatecan mesylate, have been investigated in pancreatic cancer. An every-3-weeks schedule of irinotecan at 350 mg/m2 resulted in a response rate of 9% and a median survival of 5.2 months. Clinical benefit response, however, was not assessed [47]. A phase II study of 107 patients with pancreatic cancer treated with 9-nitrocamptothecin showed a response rate of 31.7%, with an equivalent proportion of patients having stable disease [48]. Forty-four percent of patients, however, received less than two cycles of chemotherapy, and those patients were not included in assessment of response. A further phase II study of 19 patients [49] demonstrated responses in 28.6% of patients and symptomatic improvement in 71.4% of patients; however, the criteria to define a CBR were less stringent than those reported by Burris et al. [11]. Despite the impressive response rates, the median survival of 21 weeks is comparable with other studies, and the fact that 36.8% of patients experienced dosing delays with this treatment suggests significant drug-related toxicities. Exatecan mesylate has also been assessed in the phase II setting, with a response rate of 5% and a median survival of 5.6 months. Fifty-one percent of patients had received previous chemotherapy [50]. As with the taxanes, camptothecin derivatives appear to have modest activity in this disease.
Combination chemotherapy has also been investigated in pancreatic cancer, and a number of the relevant trials are outlined in Table 2
Phase II combination studies, in general, demonstrate higher response rates than those observed in single-agent studies. The combination of cisplatin and gemcitabine, for example, had a 26% response rate and a 7.1 months median survival [51], while the four-drug combination of cisplatin, epirubicin, 5-FU, and gemcitabine in patients with stage IV disease showed a 51% response rate and a median survival of 10 months [52].
Randomized controlled phase II and III trials have so far failed to deliver confirmatory evidence of better outcomes with combination therapy (Table 3
A further, recently reported phase III trial compared single-agent gemcitabine with gemcitabine and 5-FU in this disease. Again, no significant difference in median survival (5.4 months versus 6.7 months) was noted [57]. A randomized phase II trial, reported thus far only in abstract form, compared the combination of gemcitabine and capecitabine with single-agent gemcitabine. Again, no benefit for the combination was seen either in terms of response rate (17% versus 14%) or median overall survival (9.5 months versus 8.2 months) [58]. To date, no single agent or combination of agents has demonstrated greater activity when compared with single-agent gemcitabine.
A number of recently developed novel agents are currently being investigated to define their roles in the systemic treatment of pancreatic cancer (Table 4
Ras Inhibitors The ras oncogenes encode a family of proteins important in many of the signal transduction pathways involved in cell growth, differentiation, and apoptosis. Thirty percent of all tumors harbor gain-of-function mutations, suggesting a role for Ras proteins in oncogenesis. Preclinical studies demonstrating that the loss of mutated ras within a cell line results in attenuation of the malignant phenotype lend support to this hypothesis.
The functioning of wild-type and mutated Ras in signal transduction pathways is dependent upon the protein undergoing modification by farnesylation (Fig. 1
A number of FTIs have been evaluated in phase I clinical trials. They appear to be well tolerated and antitumor activity has been demonstrated [64]. As approximately 95% of all pancreatic tumors harbor ras mutations, it was hoped that FTIs may be useful in this disease, although the recent controversies over the exact mechanism of action of these agents makes this seem less certain. Two phase II studies have examined the activity of the FTI R115777 alone in pancreatic cancer, and no activity has been noted [65, 66]. In preclinical models, synergism between FTIs and chemotherapy has been noted. Consequently, a number of phase II and III studies were initiated examining the addition of an FTI to chemotherapy in patients with pancreatic cancer. In a randomized study of R115777 with gemcitabine versus gemcitabine with placebo, no improvement in median survival or response rate was seen [67]. At present, the place of FTIs in the treatment of pancreatic cancer remains uncertain.
Metalloproteinase Inhibitors In preclinical models, inhibitors of these enzymes (MMPIs) slow tumor growth and diminish metastatic potential [68]. A number of MMPIs are in clinical development. The dense desmoplastic reaction associated with many pancreatic tumors and the observation that MMPs are overexpressed in many pancreatic tumors make pancreatic cancer a rational disease in which to investigate the activity of these agents [69, 70]. BAY12-9566 is a rationally designed, nonpeptidomimetic MMPI. It has a narrow spectrum of activity, inhibiting the action of MMP-2 and MMP-9, and was the first MMPI to be compared directly with gemcitabine in a randomized clinical trial [71]. This trial, presented thus far only in abstract form, was terminated early after an interim analysis showed inferior progression-free survival and overall survival in those patients receiving BAY12-9566 compared with those receiving gemcitabine. Marimastat is a peptidomimetic MMPI with activity against all naturally occurring MMPs and, as such, it has been postulated that it may be expected to be more active than BAY12-9566. Two published studies have examined the activity of this agent in pancreatic cancer. The first examined the activity of marimastat at three different dose levels in a group of patients with advanced pancreatic cancer [72]. Patients were assessed for response after 28 days of treatment. Although no patients had evidence of a conventional response to treatment and survival in the group was similar to historical controls, marimastat was associated with stabilization or reduction in pain, immobility, and analgesic use scores in 50% of patients in this population, and 30% of patients were noted to have a stable or falling CA19-9 level hinting at some activity in this disease. A randomized study was, therefore, conducted using gemcitabine with marimastat at three dose levels in a group of previously untreated pancreatic cancer patients [73]. Gemcitabine had a superior response rate, median survival, and clinical response rate to marimastat. At the present time, MMPIs have not been demonstrated to have significant activity in pancreatic cancer. The trials conducted to date, however, have important flaws in their design. First, the doses used may be inadequatephase I trials measured the plasma level of the drug, and this may not correlate with drug concentration at the tissue level [74]. This hypothesis is supported by the suggestion of greater activity at higher doses in the trial from Bramhall et al. [73]. Second, the duration of treatment may be important with these agents, with patients requiring treatment over prolonged periods. Practically, this may be limited by the musculoskeletal toxicities associated with higher doses of these agents. Furthermore, the mechanism of action of these drugs indicates that they may have the greatest effect in those patients with minimal disease. A recent study examining the use of marimastat in patients with inoperable gastric cancer supports this hypothesis, demonstrating a greater survival at 1 year in those taking marimastat compared with those taking placebo for a subgroup of patients who had responded to palliative chemotherapy [75].
Epidermal Growth Factor Receptor Antagonists
Two approaches have been used to target the EGFR family. First, antibodies that target members of the EGFR family have been developed. Their mode of action is not entirely clear, but current evidence suggests that they act by blocking ligand binding and, possibly, downregulating the receptor by promoting its degradation. In so doing, these antibodies inhibit EGFR signaling, leading to cell cycle arrest and/or cell death. They may also function by identifying cells to the Fc receptor expressing immune effector cells leading to antibody-directed cell-mediated cytotoxicity. Two antibodies have already demonstrated exciting clinical activity: trastuzumab, an anti-Her-2 antibody, is active in Her-2-overexpressing breast cancer and cetuximab, an antibody to EGFR, has activity in colorectal cancer [78, 79]. In both instances, as predicted by preclinical data, these agents appear to interact synergistically with chemotherapy [80]. Small molecular inhibitors of EGFR tyrosine kinase activity, which prevent activation of downstream signal transduction pathways, have also been developed. These include ZD1839 (IressaTM) and erlotinib (OSI-774), which have demonstrated activity in early clinical studies [81, 82]. Her-1 is overexpressed in 30%50% of all pancreatic tumors and, in many cases, this is accompanied by overexpression of EGF [83], while Her-2 is overexpressed in approximately 20% of pancreatic tumors [84]. Several trials are, therefore, ongoing to determine the efficacy of anti-EGFR therapies in this disease. As these agents are synergistic with conventional chemotherapy, most ongoing trials are examining these agents in combination with cytotoxic agents known to have some activity in pancreatic cancer. Preliminary results from phase II trials examining the activity of trastuzumab and gemcitabine [85] and cetuximab and gemcitabine [86] show encouraging early evidence of activity with manageable toxicities. The final results of these studies are awaited with interest.
Antiangiogenics The inhibition of angiogenesis may represent a rational target for anticancer therapy development. Microvessel density and vascular growth factor production have been correlated with prognosis in a number of tumor types. Furthermore, with the exception of the endometrium and during wound healing, the normal human endothelium is nonproliferative, giving a degree of specificity and reducing toxicity. Third, a number of agents (e.g., thalidomide, paclitaxel) long known to have anticancer action have recently been recognized to function at least partly by inhibiting angiogenesis. A number of agents that target diverse parts of the angiogenic process have been developed and shown to have activity in preclinical models. These include antagonists of vasoactive growth factors, such as the anti-vascular endothelial growth factor (VEGF) bevacizumab, endothelial cell transduction inhibitors (e.g., SU-5416), and direct endothelial cell toxins (e.g., TNP 470 and combretastatin). While many of these agents are now in the clinical phase of development, there is concern regarding the choice of appropriate study end points for these drugs. End points such as serum VEGF levels, associated with effect in animal models, and markers of endothelial cell damage have not been validated and are of uncertain significance. Other investigators have used dynamic methods to assess tumor vasculature such as magnetic resonance imaging and ultrasound. It is likely, from preclinical evidence and on theoretical grounds, that these agents will work best with minimal residual disease. Expression of angiogenic factors and microvessel density correlate with a poor prognosis in patients with pancreatic cancer [8890]. In addition, pre-clinical studies of several angiogenesis inhibitors demonstrate activity in pancreatic tumor models [91]. Studies are now ongoing examining the activity of antiangiogenics in clinical trials. As preclinical evidence suggests that these agents may be more active in combination either with other antiangiogenics or with conventional cytotoxics, most trials are assessing the combination of gemcitabine with an angiogenesis inhibitor.
Some progress has been made in the medical management of pancreatic cancer over the last few years. There is now a standard treatment with clear evidence that gemcitabine can confer a clinical benefit and small survival advantage in patients with good performance status with locally advanced or metastatic disease. The fact remains, however, that gains from treatment are small and confined to a minority of the treated population. Endeavors to improve outcomes in this patient group with new chemotherapeutic agents and combination regimens are ongoing. Thus far, no single agent or combination of agents has been proven to be superior to gemcitabine. This may be partly due to deficiencies in trial design, and optimal investigation in future will require consensus on the most appropriate end points for trials assessing the activity of new agents. Advances in understanding of the molecular processes underlying this condition have led to clinical trials of novel molecularly targeted agents. While the results of clinical testing have been somewhat disappointing, it should be borne in mind that the designs of many of the trials in which these agents were assessed were suboptimal. With all agents, it is likely that only a subset of patients will benefit from treatment. Methods to identify both clinical and laboratory markers that predict response to therapies should allow optimization of treatment for individual patients in the future.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||