Detection of Disease Recurrence and Monitoring of Therapy

 

The detection of disease recurrence and treat¬ment monitoring pose high demands on diag¬nostic modalities. Whereas serum marker levels in most cases allow an assessment of tumor load and a respective response to therapy, they do not confer information on the localization of disease. Although this diagnostic gap is filled by imag¬ing modalities, most techniques based on mor-phology will come to a limit when fibrotic tissue alterations have to be differentiated from viable tumor tissue in case of suspected recurrence or when residual masses after chemotherapy have to be assessed. The metabolic information on tumor cells gained by fluorodeoxyglucose-posi-tron emission tomography (FDG-PET) imaging appears not only to be more sensitive and reliable in this respect, but also appears to allow assump¬tions on response to therapy, and ultimately on patient prognosis.

12.1    Definition of Tumor Recurrence and Epidemiological Aspects
Pancreatic carcinoma is characterized by its poor prognosis. Only a minority of patients are eli¬gible for curative surgery upon tumor detection (Birk et al. 1998).
Even in those patients operated with curative intent, 5-year survival is only 31%-51% (Panta-lone et al. 2001). This observation can partly be explained by the fact that even in patients with small tumors (<2 cm), lymph-node metastases are present in about 50% of all cases (Cleary et al. 2004; Meyer et al. 2003). Similarly, in a re¬cent analysis of the Japanese National Pancreas
Cancer Registry, only 16.5% of all patients with tumors smaller than 2 cm did not have an infil-tration of adjacent tissue or vessels, lymph node, or distant metastases (Egawa et al. 2004). As a consequence, a small size of the primary tumor does not necessarily signify an early stage of dis-ease. Unfortunately, even a more radical surgical approach often fails to lead to an improvement of survival rate, and at the cost of a reduced quality of life (Riall et al. 2005). Therefore, disease recur-rence has to be expected even in the small sub-group of patients operated with curative intent.

12.2    Disease Recurrence
Apart from clinical symptoms such as pain, weight loss, or jaundice, the increase in serum tumor markers is usually a sensitive indicator for disease recurrence. Probably the most important of these markers is the carbohydrate antigen CA 19-9. Although it is not specific for pancreatic cancer and may also be elevated in other cancers of the gastrointestinal tract, serial assessments have shown its usefulness as a parameter for the monitoring of therapy and the assessment of prognosis (Micke et al. 2003). Whereas factors such as bile retention have to be taken into con-sideration when assessing high tumor marker levels, a greater problem is probably caused by the fact that CA 19-9 is not expressed in all pa-tients with pancreatic cancer. It has been shown that individuals who are negative for the expres-sion of the Lewis (Lea and Leb) blood antigen (accounting for approx. 10% of the general pop-ulation) also do not synthesize CA 19-9. Thus, they cannot not be monitored by its serum deter-mination (Goggins 2005).

Moreover, a drawback of serological tumor markers is the lack of information they provide on the localization of disease recurrence. Locally confined recurrence might be treatable by sur¬gery or interventional procedures in conjunction with systemic chemotherapy. Moreover, even in case of disseminated disease an assessment of tumor response to therapy is desirable. Thus, im¬aging modalities do play a crucial role in recur¬rence detection.
Most recurrences occur locally due to lym-phatic spread or microscopic perineural inva¬sion (Griffin et al. 1990). In contrast to the good results achieved with endoscopic/endoluminal procedures in the assessment of suspicious pan-creatic masses prior to surgery (Harewood and Wiersema 2002), changes in intestinal continuity and the respective anastomoses may limit their use for recurrence detection in patients after Whipple's operation or pylorus preserving pan-creatoduodenectomy (PPPD).
Although cross-sectional morphologic imag¬ing by computed tomography (CT) or magnetic resonance imaging (MRI) is of great value for the assessment of postoperative complications and immediate follow-up (Scialpi et al. 2005), there are problems in the interpretation of findings in the former pancreas bed, where the differentia¬tion of fibrotic scar tissue from tumor recurrence is often difficult (Mortele et al. 2000). Func¬tional imaging by positron emission tomography (PET) offers a solution to these negative or in-determinate findings in morphological imaging
(Fig. 12.1). Unfortunately, the general scarcity of studies on recurrence detection is also true for PET studies, as only a few preliminary studies have addressed this issue (Franke et al. 1999; Jad-var and Fischman 2001; Rose et al. 1999; Higashi et al. 2003).
In a recent study, our group demonstrated the value of fluorodeoxyglucose-positron emission tomography (FDG-PET) for the assessment of suspected pancreas cancer recurrence (Ruf et al. 2005). Included were 31 patients with suspected recurrence after surgery who either showed weight-loss, pain, increased CA 19-9 levels, or a combination of these symptoms. All patients were examined by whole-body FDG-PET and contrast-enhanced multidetector CT (n = 14) or MR (n = 17) imaging. In accordance with the literature, 25/31 patients had local recurrences upon follow-up, which had been diagnosed by imaging modalities in 23/25 patients. FDG-PET detected 96% (22/23) of these cases, whereas morphological imaging by CT or MRI was posi-tive in only 39% (9/23). Liver metastases were, in contrast, better depicted by MRI and CT imag¬ing, with a detection rate of 92% (11/12) vs 42% (5/12) for FDG-PET. The possible explanation for this observation is the improved detection of small hepatic lesions made possible by dynamic multiphase MRI/CT examinations in contrast to PET, which in the case of small lesions suffers from partial volume effects that make tumor-specific FDG uptake indiscernible from physi¬ologic hepatic activity. Despite the drawback of limited anatomical orientation, FDG-PET was again superior to MRI/CT with regards to ab-dominal lymph-node involvement, as focal up-take was more indicative of tumor recurrence than the mere size assessment of lymph nodes by morphologic imaging. Moreover, as FDG-PET is routinely performed in a whole-body technique, unknown extra-abdominal metastases were de-tected in 2 patients.
Since FDG-PET was superior to morphologi¬cal imaging, it is imaginable that it allows for an earlier detection of recurrence and therefore an earlier initiation of therapy. However, other po-tential influential factors on FDG uptake have to be addressed (please also refer to Chap. 3). As pancreatic surgery may lead to a diabetic meta-bolic state (if not already present prior to sur-gery), a potential decrease in FDG sensitivity has to be considered especially in those patients with a low compliance with regards to antidiabetic medication or in which euglycemia is difficult to achieve (Diederichs et al. 1998). Moreover, as in the initial assessment of pancreatic masses, cellu-larity, the expression of glucose transporters and enzymatic activity of glycolytic enzymes, has to be heeded (Higashi et al. 2002).

12.3    Response to Therapy and Prognosis
Apart from the clinical patient reexamination, the response to radiation treatment or chemother¬apy is usually assessed by morphological imaging modalities such as ultrasound, CT or MRI, using both the number of lesions and their respective size as parameters. However, whereas an increase or decrease of the number lesions is relatively easy to discern, changes in lesion size as assessed either by WHO (perpendicular diameter) or the EORTC's area measurement approach (response evaluation criteria in solid tumors, RECIST) may be rendered more difficult by necrotic, fibrotic, and/or cystic transformation of metastases under therapy (Miller et al. 1981; Therasse et al. 2000). The value of metabolic imaging for the differen-tiation of residual vital tumor tissue vs necrotic or fibrotic residue has been demonstrated, e.g., in the neoadjuvant treatment of colorectal cancer, where FDG-PET correlated far better than mor-phological imaging to the histopathology of the resected tumor (Amthauer et al. 2004). In con-cordance with morphological imaging, attempts have also been made to standardize the response to therapy by the use of quantification measure-ments (Young et al. 1999).
Interestingly, apart from the determination of the primary diagnosis, prognosis rather than response to therapy has been assessed by the ma-jority of FDG-PET studies (Table 12.1). A possi-ble reason for this is the simple fact that response to therapy requires at least two studies of the at the time of the rather costly PET examination. In 1997, Nakata and coworkers pioneered with a study on 14 patients suffering from pancreatic cancer that received conventional imaging as well as FDG-PET prior to treatment. FDG up¬take was semiquantified by the determination of the standardized uptake value (SUV) of the pri¬mary tumor, which was correlated to patient sur¬vival. Using a cut-off SUV of 3.0 for high and low glucose uptake respectively, they observed a sig-nificant difference in the patient group with low uptake (14 months) and high uptake (5 months). Using a SUV of 6.1, Zimny and coworkers (2000) were able to reproduce these results in 52 pa¬tients with 5 months survival in case of an SUV exceeding 6.1 vs 9 months survival in the case of an SUV below 6.1 (p = 0.0321), with multivariate analysis revealing the SUV as an independent prognostic factor. These results were strength¬ened by an Italian study on 60 patients, which came to a similar conclusion using a cut-off for SUV of 4 (Sperti et al. 2003).
Although the wide range of cut-offs for SUV in these studies can be explained by differing examination protocols utilized at the respective institutions, it is doubtful that there actually is an "ideal" SUV threshold (please also refer to Chap. 3). Moreover, only the glucose avidity of the primary tumor has been primarily assessed, whereas the correlation of SUV to tumor stage or the SUV of metastases has only been poorly investigated (Higashi et al. 2003).
Nevertheless, these studies indicate the poten-tial of metabolic imaging by FDG-PET in patients diagnosed with pancreatic cancer, leaving aside the pitfalls associated with the primary diagno¬sis of pancreatic cancer, i.e., the differentiation of glucose-avid inflammation from glucose-avid cancer (please refer to Chap. 3). With regards to

one-time examinations, dual-phase imaging, i.e., the measurement of glucose uptake at two time-points during one imaging session, not only ap¬pears to improve the differentiation of benign from malignant disease, but potentially allows for a more precise assessment of prognosis, more ap¬propriately reflecting the kinetics of glucose me¬tabolism. Using a dual-phase approach, Lyshchik and coworkers (2005) determined an intraindi-vidual retention index (RI) based on the respec¬tive SUVs measured 1 h (1) and 2 h (2) after tracer injection (RI = SUV2 - SUV1/ SUV1 x 100%) for each of their 65 patients. RI was an independent marker for survival in the multivariate analy¬sis, and showed the strongest prognostic differ¬ence for an RI cut-off of 10% (UICC stage I—III: 15.3 months vs 11.5 months; stage IV: 9.5 months vs 4.9 months).
Moreover, it is imaginable that a more appro-priate estimation of prognosis can be derived not from a single examination upon tumor detec¬tion but rather when the response to therapy is assessed (Table 12.1). In one study, nine patients underwent FDG imaging both before and 4 weeks after chemoradiation therapy. Instead of the de-termination of a cut-off for SUV, the change in SUV was assessed. The group observed that a de¬crease in SUV larger than 50% was present in all patients that showed good histological response to therapy (Rose et al. 1999). Another study ex¬amined 11 patients both before and 1 month af¬ter chemotherapy (Maisey et al. 2000). Although their evaluation of PET data was rather simple, as only visual analysis was performed, they saw a significantly extended survival in those patients that showed no FDG uptake in the control scan as opposed to the group in which tumor activity was still visible (mean survival: 318 vs 139 days).
Both studies draw their conclusions from serial PET examinations that, just as with dual-phase imaging, might deliver more robust re¬sults, since they are based on intraindividual comparisons in which the biological factor, the patient, remains a constant.
Furthermore, as metabolic changes usually occur earlier than morphological changes, it might be possible to estimate the response of a patient undergoing (rather than after) therapy and to switch to a another therapeutic regimen in case of nonresponse. The potential of FDG-PET in this setting has been demonstrated e.g., in esophageal cancer. In one study on patients with carcinoma of the gastroesophageal junc¬tion, the change in FDG uptake between baseline examination and a control examination as early as 14 days after initiation of neoadjuvant chemo-therapy was indicative for the differentiation of patients who profited from therapy vs those who did not according to conventional follow-up and postoperative histology (Weber et al. 2001). With regards to pancreatic cancer, however, further studies are required.

12.4    Summary
Due to the high percentage of patients with pan-creatic carcinoma that can only be palliatively treated and the high rate of recurrence in pa¬tients operated with curative intent, reliable im¬aging modalities for therapy control, the assess¬ment of prognosis, and recurrence detection are desirable.
Whereas MRI and CT will remain the basic imaging modalities for therapy control and pa-tient follow-up due to their availability, meta¬bolic imaging by FDG-PET has the potential to improve both the monitoring of therapy and the assessment of prognosis.
 

Brachytherapy of Liver Metastases

Brachytherapy for the treatment of liver me-tastases is a novel approach. In this procedure, techniques of locally ablative treatment in in-terventional radiology and radiation therapy are combined. After computed tomography (CT)-guided percutaneous implantation of cath-eters into the hepatic tumor, the irradiation is performed in an afterloading technique. This minimally invasive procedure offers circum-scriptive high-dose rate irradiation of the lesion to treat in a single session, irrespective of breath¬ing motion or potential cooling effects of neigh¬boring vessels. Good local control rates have been achieved in several tumor entities, includ¬ing both secondary and primary malignancies of the liver. This article gives an overview of the ap-plication technique, possible adverse events, and outcome with special attention to the pancreatic cancer scenario.

11.1    Introduction
Locally ablative therapy is an interesting option for patients with irresectable metastatic disease or primary hepatic malignomas confined to the liver. The aim is the complete ablation of all he-patic lesions or at least the achievement of a lo¬cal tumor control. Specifically, radio-frequency ablation (RFA) has captured increasing interest, since it is easy to use even as an outpatient pro-cedure in selected patients (Meyers et al. 2003). Laser-induced thermotherapy (LITT) offers the opportunity of real-time therapy monitoring by magnetic-resonance thermometry, which is thought to be advantageous to other locally ab-lative procedures (e.g., RFA) (Nolsoe et al. 1993; Vogl et al. 1995). However, these procedures have limitations concerning number and localization, as well as size and shape, of tumor lesions.
More recently, there is growing interest in applying radiotherapy to hepatic malignancies. Compared to local thermoablative procedures, radiation efficacy is not affected by cooling ef-fects of neighboring vessels or bile ducts, which are known to be a potential source of local tumor progression after RFA or LITT. Adjacent organs such as the colon or the hilar bile ducts play a minor role for possible complications. The size and shape of the radiation target volume is not restricted to less than 5 cm in diameter and spheroid lesions. However, as the tolerance dose of liver parenchyma is lower than that of most tumor tissues, the therapeutic efficacy of percu-taneous irradiation interferes with the manda¬tory maintenance of a sufficient liver function. The main problems are the breathing excursion of the liver and the flat dose shoulder surround¬ing the target volume, resulting in a relatively high radiation exposure of the normal hepatic parenchyma. Even though there do exist in-novations such as respiratory gated irradiation, stereotactic irradiation, or tomotherapy devices, these problems have not generally been solved to date (Herfarth et al. 2004; Wurm et al. 2006).
The drawbacks of external beam radiotherapy can be overcome when irradiation is brought next to or into the tumor, offering a steep dose decrease to the periphery around the irradiated focus and independency from breathing motion. This approach is referred to as brachytherapy. As a high dose rate, hypofractionated radiation therapy, this technique is used, e.g., for endo-bronchial or endovaginal irradiation of lung and cervical cancer, or interstitial irradiation of superficial tumors (e.g., breast cancer, head and neck cancer). It is usually realized in an af-terloading technique, where a radiation source, e.g., iridium-192, is inserted into prepositioned catheters, a technique that offers the opportunity of high dose rate irradiation (HDR, >12 Gy/h) with minimal exposure of neighboring tissues. The sites accessible for traditional noninvasive brachytherapy are limited, as body cavities such as the trachea or the vagina are needed to insert the afterloading catheter, or invasive implanta¬tion of catheters is required.
Concerning brachytherapy of hepatic ma-lignomas, intraoperative radiation therapy has been successfully used in the past. However, as most of the indications are palliative approaches, minimally invasive procedures with a low risk of morbidity and mortality are warranted. This can be achieved by employing radiological interven-tional procedures, which are mainly based on im¬age guidance. To treat hepatic malignancies, the afterloading technique is combined with the pro¬cedure of image-guided interventional, locally ablative treatment. Inserting a radiation source into the tumor in an afterloading technique via transhepatic catheters implanted percutaneously under computed tomography (CT) guidance is largely independent of breathing motion and of¬fers a steep dose reduction toward the periphery around the target volume for optimally focused dosing of the tumor. This technique is invasive and thus requires a short radiation time and is therefore applied as a single session HDR brachy-therapy (Ricke et al. 2004a).
In the following sections, aspects of interstitial brachytherapy with CT-guided afterloading will be discussed including patient selection, treat-ment planning, procedures, technical consider-ations, adverse effects, and clinical outcome.

11.2    CT-Guided
High Dose Rate Brachytherapy via Interstitial Afterloading

11.2.1  Background
High dose rate brachytherapy for the treatment of unresectable liver metastases has been used previously in an intraoperative setting. Efficacy and safety have been proved in several studies. In these trials the minimal target doses covering the entire tumor ranged between 15 and 30 Gy; internal dose inhomogeneities inside the target volume, depending on the radiation technique, were tolerated (Nauta et al. 1987; Dritschilo et al. 1988; Thomas et al. 1993). In a study with 22 pa-tients suffering from irresectable liver metastases, irradiation was realized with laparotomy and in-terstitial HDR brachytherapy using iridium-192 with doses in the tumor periphery ranging from 20 to 30 Gy (Thomas et al. 1993). There was no acute or chronic radiation toxicity observed at a median follow-up of 11 months. Median actuar-ial local control at irradiated sites was 8 months, with 26% actuarial local control at 26 months by CT or magnetic resonance imaging (MRI). This phase I/II trial demonstrates the feasibility of single fraction HDR brachytherapy in the treat-ment of liver metastases.
However, as most of these procedures are palliative, a minimally invasive approach with¬out the risk of laparotomy is favorable. This may be achieved by image-guided procedures, e.g., CT-guided percutaneous puncture of the hepatic tumor and catheter placement for subsequent af-terloading as described by Ricke et al. (2004a). This interventional radiological approach has been successfully employed for treatment of sev-eral secondary and primary hepatic malignomas, and also in malignomas of the lung and other sites (Ricke et al. 2004a, 2005a).


11.2.2 Patient Selection
In general, locally ablative treatment should be preserved for patients with a limited number of tumor deposits, regionally confined disease, or symptomatic lesions. Local overtreatment as an unnecessary risk should be avoided. Whether a patient might profit from locally ablative treat-ment depends on the tumor entity, tumor spread, and the overall clinical condition. Fluorodeoxy-glucose-positron emission tomography (FDG-PET) has proved to be a valuable adjunctive to the conventional staging modalities in the evalu¬ation of patients prior to locally ablative treat¬ment (Amthauer et al. 2006). The indication has to be made individually after thorough ex¬amination and careful assessment of alternative treatment options. Clinical and paraclinical par¬ameters such as comorbidity, liver function, and blood coagulation have to be taken into consid¬eration, as well as the patient's wishes.
Theoretically, minimally invasive interstitial afterloading is applicable to many potential tumor localizations. It has already been success-fully applied for treatment of pulmonary and he-patic malignomas, but also in the mediastinum, in the retroperitoneum, and bone. The main limitation is of course the technical feasibility of catheter placement, as a minimal risk of the pro-cedure has to be ensured. Additional limitations are surrounding tissues at risk for adverse effects of irradiation, such as bowel, stomach, spinal ca-nal, skin, neuronal tissue, etc. The number and size of the lesions to treat is also an issue. How-ever, the procedure can be adapted to achieve a sufficient dose coverage in target volumes that are much larger than those suitable for thermal ablation techniques (Ricke et al. 2004b). In large tumor volumes or numerous target lesions, the procedure can be completed with several step-by-step sessions. Additionally, it has been shown that CT-guided interstitial brachytherapy is in¬dependent from the cooling effects of large ves¬sels or bile ducts in the ablation zone, which have been identified as potential causes of inadequate heating and local recurrent tumor growth in thermal ablation techniques (Ricke et al. 2004b). Furthermore, as one of the major advantages compared to thermal ablation techniques, the shape of the ablation zone can be adapted to the irregular geometries of target lesions after the catheter implantation by modulating the dwell locations and dwell times of the radiation source inside the afterloading sheaths. Thus, the achiev¬able ablation volume is not only defined by the CT-guided puncture, but also by the planning af¬ter a contrast-enhanced CT scan is obtained with optimal demarcation of the tumor lesions and the implanted catheters.
The widest experience has been with hepatic malignomas and the following will mainly focus on interstitial HDR brachytherapy of liver me-tastases via percutaneous transhepatic afterload-ing. However, these technical aspects of locally ablative treatment apply not only for hepatic malignomas but also for other organs invaded by tumors, such as the lung, where patients may benefit from interstitial therapy, predominantly in a palliative scenario (Ricke et al. 2005a).


11.2.3 Therapy Procedure
The application of CT-guided interstitial brachy-therapy consists of five steps. After appropriate patient selection with staging and assessment of the feasibility based on local findings, the steps are (1) planning the access, (2) CT-guided cath-eter implantation, (3) radiation planning, (4) ac-tual irradiation via afterloading, and (5) removal of applicators.
For access planning, cross sectional imag¬ing (whether CT or MRI) is needed (Fig. 11.1a). The chosen imaging modality must be capable of outlining the tumor precisely against the sur-rounding tissue, vessels, and central bile ducts. A gross planning of catheter positions is done us¬ing the axial slices, as this is the orientation of the fluoroscopic CT monitoring of the intervention. Besides the puncture direction, dose coverage of the clinical target volume and sparing of sur-rounding tissues at risk has to be respected dur-ing catheter placement. It is recommendable to place the needle that way, that a bridge of liver parenchyma lies between the liver capsule and the tumor border. This buffer provides a hold for the catheter and prevents bleeding of commonly hypervascularized tumors and the potential spill-ing of tumor cells into the abdominal cavity.
The patient is positioned supine in the CT scanner and monitored for blood oxygenation and heart frequency. The intervention is per¬formed under aseptic conditions and CT guid¬ance after intravenous analgosedation as well as local anesthesia at the cutaneous puncture site. The puncture is monitored by CT fluoroscopy. A guide wire is inserted through the needle, and then a flexible catheter sheath replaces the needle (opaque on X-ray). After removal of the guide wire, an afterloading catheter is placed into the catheter sheath. The system is stitched to the skin for fixation. If more than one afterloading cath-eter is needed, the procedure is repeated.
Upon completion of catheter placement, a contrast-enhanced scan of the liver is acquired

for documentation of the exact catheter location in relation to the tumor (Fig. 11.1d). These data are the basis for 3D reconstructions and radiation therapy planning on a dedicated workstation by outlining the gross tumor volume, the catheters, and the surrounding risk tissues (e.g., bowel, stomach wall, gall bladder, kidney, spinal canal, skin) (Fig. 11.1b, e). This technique with retro-spective registration of the catheter positions is highly accurate and less complex as compared to prospectively arranged catheter positions with templates or intraoperative raster placement (To-nus et al. 2001; Kolotas et al. 2003). With a se¬lected minimal target dose, usually 15-25 Gy, an afterloading plan is generated giving dwell loca¬tions and dwell times for the iridium-192 source [half-life, 78.8 days; decay, beta (672 keV) and gamma (<469 keV)] inside the afterloading cath¬eters. This plan needs control and can be adjusted manually if necessary. The goal is to modulate a planned target volume covering the entire gross tumor volume and a safety margin while spar¬ing healthy surrounding tissue, especially risk organs, as much as possible. An optimization of target volume definition and consecutively the dose coverage of the tumor can be achieved by registration of the previously acquired FDG-PET images with the CT data (Denecke et al. 2006). Although the number of catheters is theoretically unlimited, it is recommendable not to exceed 6-8 catheters, depending on the tumor size and shape. Because of the stress situation, the radia¬tion time should be limited to a maximum of 1 h, depending on the patient's condition.
Using this plan, the afterloading procedure is performed and subsequently the catheters are slowly removed, sealing the puncture channels with tissue glue or other thrombogenic material to prevent bleeding.


11.2.3 Undesired Side Effects
Complications can be subdivided into acute complications, occurring during or immedi¬ately after treatment, and late complications. The inadvertent acute events are mainly due to mechanical alterations caused by the puncture and catheter placement (e.g., bleeding, perfora-tion of bowel, stomach, or gall bladder). These inadvertent events, however, occur very rarely, as CT-guided puncture of the liver is a safe way to avoid severe injuries of nontarget tissues. Major bleeding from the liver is an extremely rare com-plication and can be prevented by sufficient seal-ing of the puncture channel during retraction of the catheter sheaths. Other acute side effects are emesis, pain, and shivers, which are to be treated medically.
Delayed side effects besides infectious com-plications are mainly related to radiation expo-sure of nontarget tissues. Treating hepatic ma-lignomas, exposure of surrounding healthy liver tissue to a relevant radiation doses is desired as a safety margin. However, a sufficient hepatic re¬serve has to be ensured before treating hepatic malignomas, particularly in patients with large and/or multiple lesions, preexisting liver disease (e.g., hepatocellular carcinoma in cirrhosis, por¬tal vein thrombosis), previously irradiated liver
(dose accumulation), or otherwise impaired liver function reserve due to prior chemotherapy. The tolerance dose of a healthy liver is approximately 30 Gy to the whole organ or 50 Gy to one-third of the liver volume. For external radiotherapy, the clinical endpoints are liver failure and severe hepatitis. If the irradiated volume of normal liver tissue is reduced to approximately 100 ml or less, the tolerated doses are much higher—in prin¬ciple, without any upper limit with respect to the clinical endpoints mentioned. Additionally, the different radiobiological effects of a single high dose fraction to the tissue compared to fraction¬ated strategies has to be considered. It is well known that healthy tissue tolerates larger doses applied in multiple fractions. The options for the irradiated liver tissue are either destruction or recovery to normal liver function. Additionally, compensative mechanisms of the remaining non-irradiated liver parenchyma has to be taken into account. A recent study showed that for intersti¬tial brachytherapy in an afterloading technique with an iridium-192 source, the tolerance dose causing an early function loss of hepatocytes as determined in MRI with hepatocyte specific con¬trast material 6 week after irradiation was 9.9 Gy (±2.3 standard deviation) (Ricke et al. 2004b). This and the careful assessment of the hepatic re¬serve have to be taken into consideration when planning the treatment to avoid posttherapeutic hepatic failure.
Other tissues at risk are, e.g., bile ducts, gall bladder, gastrointestinal tract, skin, kidney, and spinal cord. Previously described complications have included rare events such as strictures of the common bile duct or gastric ulcers (Ricke
et al. 2004a; Streitparth et al. 2006). Concern¬ing gastric complications, a threshold dose of 15.5 Gy/ml tissue for the clinical endpoint ul-ceration of gastric mucosa has been estimated (Streitparth et al. 2006). This in vivo assessment is in accordance with tolerance data by Emami et al. (1991). Regarding the small and large bowel, dose thresholds have not been estimated yet, but it has been hypothesized that they are similar to those described for the gastric wall; overall, however, these complications and late ef¬fects are rare, to which patients with repeated ir¬radiation close to the risk tissues are more prone (Ricke et al. 2004a, 2005b). Concerning gastric exposure, a proton pump inhibitor therapy is be¬ing recommended as ulcer prophylaxis. Potential risk and benefit have to be thoroughly evaluated before treatment initiation and during radiation planning.


11.2.4 Efficacy
Local tumor ablation has become a valuable tool in oncological treatment concepts. The majority of locally ablative procedures are performed by applying thermal ablation, such as RFA or LITT. However, with respect to the limitations of ther-mal ablation modalities (i.e., tumor size, tumor shape, tumor location, adjacent risk structures), novel techniques combining brachytherapy with interventional techniques have demonstrated favorable outcomes. In contrast to thermal abla-tion, CT-guided brachytherapy is independent of complex geometric configurations of the lesions, as dwell times and dwell locations of the source within the applicators can be adjusted to fit the outlines of the tumor (Ruhl and Ricke 2006). Fur-thermore, adjacent ducts and vessels do not in-fluence the ablation zone as brachytherapy is not prone to disturbing cooling effects. In contrast to external beam radiation, breathing motions are not an issue, because the afterloading catheters move with the tumor (Ricke et al. 2004b).
Early studies on CT-guided brachytherapy showed local tumor control rates of 87% after 6 months at minimal dose levels of 12-20 Gy (Ricke et al. 2004a). An analysis of the treatment of 200 colorectal liver metastases between 1 and 11 cm (median 4 cm) recruited for a phase III study revealed a local tumor control rate of 96% after 12 months when applying 25 Gy, and 67% when applying 20 Gy as the minimal tumor dose. Major adverse events were hemorrhage in 3 pa-tients (2%), which ceased after blood transfusion
(Ricke et al. 2005c).
The use of interstitial brachytherapy is not limited to its application inside the liver, as treatment of lung malignancies has also dem-onstrated promising results with respect to local tumor control and side effects. In a phase I trial, 15 patients with 28 lung metastases and nons-mall cell lung cancer in 2 cases were treated with a single fraction of at least 20 Gy inside the entire clinical target volume. No major adverse events were reported. Minor events included radio-graphically visible local hemorrhage in 2 patients (Ricke et al. 2005a). In contrast to thermal abla¬tion techniques, air cavities in the lung were not seen. Radiobiologically, the cytotoxic effect after single-fraction, high-dose rate irradiation shows within weeks to months, with only moderate acute injury (Manning et al. 2001).

11.3   The Role of CT-Guided Interstitial
Brachytherapy in Pancreatic Cancer
The widest experience with CT-guided brachy-therapy has been with colorectal cancer, breast cancer, and hepatocellular carcinoma. In con¬trast, pancreatic neoplasias are less favorable for locally ablative treatment, depending on the his-tological subtype.
Regarding liver metastases from neuroen-docrine gastroenteropancreatic tumors, locally ablative treatment has been used successfully for cell reduction, tumor eradication, and symptom relief in functionally active tumors (Elvin et al. 2005; O'Toole and Ruszniewski 2005). In this context, a special focus of CT-guided brachy-therapy is the treatment of lesions unfavorable for LITT or RFA because of size, shape, and lo¬cation. Even though some of these tumors are known to be less sensitive to radiation therapy, it has to be emphasized that in the setting of single session HDR brachytherapy with doses above 20 Gy and core doses above 50-100 Gy, the cri¬terion radiosensitivity of tumor tissues plays a minor role. Therefore, CT-guided brachytherapy for locally ablative treatment of liver metastases appears to be a promising tool in this subset of pancreatic neoplasias.
Concerning the treatment of pancreatic duc-tal adenocarcinoma, there are no data available yet assessing the use of this novel technique. Ductal adenocarcinoma represents the majority (90%-95%) of pancreatic malignancies and is known to have a poor prognosis (5-year overall survival, 1%-2%) (Wagner et al. 1999; Tsiotos et al. 1999; Brand and Tempero 1998; Rosewicz and Wiedenmann 1997). The resection of the primary tumor currently being the only poten¬tially curative treatment (5-year overall survival after R0 resection, approximately 20%), most of the affected patients (approximately 80%) are ir-resectable at the time of first diagnosis, owing to distant metastases or local tumor extent (Lopez-Hanninen et al. 2002; Wagner et al. 1999; Tsiotos et al. 1999; Brand and Tempero 1998; Rosewicz and Wiedenmann 1997). Even though there are promising developments in both surgical tech¬nique and chemotherapeutic agents, the outcome remains rather poor. This emphasizes the need for innovative treatment strategies as adjuncts to the traditional therapeutic approach.
The efforts being made toward treating the primary pancreatic ductal adenocarcinoma by lo-cally ablative procedures such as RFA and LITT, as well as external beam irradiation for neoadju-vant or palliative therapy, have met with varying success (Stroszczynski et al. 2001; Varshney et al. 2006). As most adenocarcinomas are located in the pancreatic head (approximately 75%) the le¬sions are difficult to reach by CT-guided local ab¬lation. Surrounding vessels and risk structures, such as the bile ducts, the duodenum, and the stomach, as well as the pancreatic parenchyma itself, limit the ablation volume for either modal¬ity. Even in intraoperative RFA of pancreatic ma¬lignancies, there have been severe complications observed, such as necrotizing pancreatitis (Elias et al. 2004). The primary tumor, especially when judged as irresectable, is a diffuse mass with an irregular growth pattern. Therefore, the inclu¬sion of the entire tumor is often impossible and it can be questioned whether local tumor destruc¬tion exposes the patient unnecessarily to risks that outweigh the potential benefit. Even though CT-guided brachytherapy is more flexible re-garding the configuration of the target volume, these limitations apply here as well. On the other hand, retarding the local tumor progression only by a few months has to be considered a success, as alternative treatment options are rare and of¬fer only moderate response rates and rather poor outcomes.
Referring to hematogenous distant metas-tases, these are most commonly located in the liver. Even though the progression of the disease is rapid and, in most cases of metachronous liver metastases, further micrometastases are present that are invisible to diagnostic imaging, hepatic metastases in theory can be considered as a lim¬ited disease. Thus, it can be considered whether a patient might profit from removal of the me-tastases if the primary tumor has been resected. There are currently no data showing an advan¬tage of hepatic surgery in such a scenario, and any unnecessary risk has to be kept as low as pos¬sible in the palliative setting. Therefore, the use of minimally invasive ablation of liver metastases is a promising alternative to surgery, not only in irresectable patients. Treating hepatic metastases by thermal ablation in pancreatic cancer patients implies an additional issue. As most adenocar-cinomas are seated in the pancreatic head, the resection includes a biliodigestive anastomosis. This condition allows ascending bacterial colo¬nization of the biliary tree, which supports the development of abscesses in the necrotic ablation zones after thermal tumor coagulation (Thomas et al. 2004). A similar condition is present in palliative treatment of irresectable patients with stents in the common bile duct. In contrast to thermal ablation, brachytherapy induces a pro-longed tumor inactivation along with ongoing organization of the developing necrosis, which is accessible to the immune system, and therefore the rate of abscesses is extremely low compared to thermal tumor destruction. Infection of the ablation zone therefore appears to be a minor problem for interstitial afterloading in this spe-cific patient group.
To create reasonable indications for the use of locally ablative treatment in pancreatic cancer patients, it has to be implemented into a mul-tidisciplinary approach including surgery and chemotherapy in an individualized therapeutic strategy. A possible indication for locally abla¬tive treatment would be the eradication of liver metastases to facilitate resection of the primary tumor. For illustration of this multidisciplinary individualized approach, we present a case with an undifferentiated osteoclastic-type giant cell tumor of the pancreatic corpus (Fig. 11.1).
The patient presented with two liver metasta-ses in the right lobe and a portal vein thrombosis of the left branch at initial diagnosis. Chemo¬therapy maintained stable disease, but caused in¬tolerable sensory irritation in the hands and the feet. The patient was referred to our department for CT-guided interstitial HDR brachytherapy of the liver metastases in order to retard local tumor progression. Because of inaccessibility and the previously mentioned risks of locally ablative therapy inside the pancreas, the primary tumor was not treated. After ensuring a good local control of the liver metastases by MRI 3 months after brachytherapy of the two liver lesions and exclusion of new intra- and extrahepatic metas-tases by CT, a R0 resection of the pancreatic tail and corpus with the primary was performed. Eighteen months after brachytherapy the patient is still free of tumor with a good quality of life.
To summarize, CT-guided brachytherapy of liver metastases may be reasonable in patients with local resectability and a limited number of liver metastases, when stable disease is ensured by a sufficient surveillance interval. Here and in previously resected patients with metachronous liver metastases, the palliative use of CT-guided brachytherapy might prove beneficial in the fu-ture and appears to be advantageous compared to thermal ablation because of the reduced rate of abscesses in the ablation zone.

11.5    Conclusion
Combining the features of substantially different therapies, such as the safety of CT-guided or¬gan puncture, efficacy of brachytherapy, and the principles of the afterloading technique, a novel therapeutic approach in radiation oncology has been developed. Percutaneous afterloading of hepatic malignomas enables effective treatment, even in those patients who are not suitable to undergo surgical or thermal ablations because of tumor size or location. There are promising data for the treatment of hepatocellular carcinoma and colorectal liver metastases, and further stud¬ies are pending. For pancreatic cancer, the poten¬tial indications for CT-guided brachytherapy are currently limited to the treatment of liver metas-tases. Despite the lack of controlled trials, there is probably a role for this therapeutic approach in an individualized multidisciplinary therapeutic approach in a carefully selected patient group.

Primary Advanced Unresectable Pancreatic Cancer

Median as well as overall survival of pancre¬atic cancer patients in the advanced stage is ex¬tremely low despite advances in cancer therapy regarding tumor cell biology, therapy resistance, and diagnosis. In matters of chemoradiation therapy (CRT) in locally advanced pancreatic cancer, favorable positive effect has been reached with different radiotherapy proceedings such as intraoperative radiation therapy with or with¬out external chemo-/radiation therapy or with CRT alone with regard to local tumor pain, lo¬cal tumor remission, or local control of disease and overall survival. Primary (chemo-) radia¬tion therapy only rarely leads to local remission. Intraoperative radiation therapy (IORT) merely reaches pain palliation in most cases. By admin-istering up-to-date primary CRT, especially with gemcitabine-associated CRT, local remission in up to 50% of patients can be observed. By ap-plying neoadjuvant CRT, better resectability and the reduction of postoperative positive lymph node metastasis has been seen in patients with resectable or possibly resectable pancreatic can-cer. With primary CRT, resectability can also be achieved in patients with primary unresect-able pancreatic cancer. It has been shown at the evaluation of patients' progression samples—ei-ther treated with neoadjuvant or primarily with radiotherapy (with conventional radiation tech-nique)—that the rate of local recurrence or local progression can be reduced in comparison with historical cohorts. By contrast, the rate on distant metastases was not affected. Whereas concurrent CRT leads to favorable local tumor control, this procedure has a minor effect as to the survival in most of the studies. Because metastases occur mostly out of the irradiation field and because of partly advanced local tumor progression, the concept of combined CRT with continuing che-motherapy was developed.
Median survival of pancreatic patients in the advanced stage is approx. 3-5 months, with a 12-month survival probability of 10% despite ad-vances in cancer therapy. On the other hand, the 5-year survival probability is 0.4%-3.0% (Bram-hall et al. 1995, 1998).
The causes of such a dismal prognosis can be understood first of all in the commonly late diag-nosis (Haycox et al. 1998), second in the aggres-sive tumor cell biology with continuing therapy resistance (Magee et al. 2001), and finally because an acceptable resection rate can be achieved only in specialized centers (Birkmeyer et al. 2002; Ne-optolemos et al. 1997).
Only 10%-15% of patients can be resected af-ter the diagnosis of pancreatic cancer. Resection is considered a potential curative therapy. How¬ever, median survival of these patients amounts to only 13-18 months, with a 5-year survival of 10%-20% (Bramhall et al. 1995; Yeo et al. 1997). The survival rate did not improve with a radical resection and extended lymphadenectomy (Pedrazzoli et al. 1998).
Furthermore, 15%-30% of primary nonmeta-static pancreatic cancer is unresectable due to extended vessel infiltration at time of diagnosis. The prognosis for these patients is very dismal due to lack of specific therapy; moreover, median overall survival is a maximum of 6-8 months (Niederhuber et al. 1995; Shinchi et al. 2002).
10.1    Chemotherapy of Advanced
Pancreatic Cancer
Although more chemotherapeutic agents have been examined for the purposes of the therapy of advanced pancreatic cancer, only 5-FU, mito-mycin-C (MMC) (Haycox et al. 1998) and, lately, gemcitabine (Burris et al. 1997; Moore et al. 1995; Rothenberg et al. 1996) have shown reproduct¬ible outcomes with objective results.
A 5-FU-based combined chemotherapy has shown a clear survival advantage compared to patients without treatment in randomized con-trolled studies (Glimelius et al. 1996; Mallinson et al. 1980; Palmer et al. 1994). However, com¬pared to monotherapy with 5-FU, a toxicity in¬crease without additional improvement of sur¬vival has been reported (Cullinan et al. 1990).
Gemcitabine belongs to a series of new che-motherapeutic agents tested for pancreatic can¬cer. It has shown superior efficacy both in mono-therapy and in combined therapies (Berlin et al. 2002; Burris et al. 1997; Heinemann et al. 1999b, 2000). Nevertheless, the application of fluoro-pyrimidine continues to be of interest in trials where the efficacy of 5-FU after portal vein in¬fusion (PVI) application and the development of orally applicable chemotherapeutic agents is sought (Neoptolemos et al. 2004). The National Cancer Research Institute in Britain has recently started a Gem-Cap phase III trial where gem-citabine (Gem) will be applied with capecitabine
(Cap).
The survival advantage with gemcitabine is minor compared to bolus 5-FU (Burris et al. 1997). Nevertheless, it is being used in advanced pancreatic cancer increasingly as the standard therapy.
However, a significant breakthrough in the therapy of advanced pancreatic cancer with che-motherapy has not been found yet.

10.2    Chemoradiation in Locally
Advanced Pancreatic Cancer
A favorable positive effect has been reached with different radiotherapy proceedings such as in-traoperative radiation therapy (IORT) with or without  external chemo-/radiation therapy or with chemoradiation (CRT) alone with regard to both local tumor symptomatic (local tumor pain), local tumor remission, or local control of disease and overall survival (Fossati et al. 1995; Nishimura et al. 1997; Staley et al. 1996). There-fore, chemoradiation with a total irradiation dose of 45.0-50.0 Gy (eventually up to 60.0 Gy) with conventional fractionation and a concur¬rent chemotherapy with 5-FU (eventually PVI during the whole therapy with 200-225 mg/m2 per day) was recommended as the standard and most effective therapy procedure for patients in good general condition (German Cancer Asso¬ciation 2002).
Randomized trials reporting significant im-provement of median survival after chemoradia-tion are listed in Table 10.1 (Moertel et al. 1981; Li et al. 2003; Shinchi et al. 2002).
For decades, 5-FU has been considered the agent of choice with regard to the chemothera-peutic agents administered concurrently or sequentially to radiation. Combined chemo¬therapies such as FAM (5-FU, doxorubicin, mito-mycin-C) or SMF (streptozotocin, mitomycin-C, 5-FU), or the Mallinson regimen (5-FU, cyclo-phosphamide, methotrexate and vincristine) re¬sulted in increased toxicity and no improvement as to survival (Bruckner et al. 1993; Cullinan et al. 1990). Even newer agents tested recently for pancreatic cancer such as paclitaxel, docetaxel, irinotecan, topotecan, and oxaliplatin could not be established as treatment (Ashamalla et al. 2003; Kamthan et al. 1997). Only after the intro-duction of the pyrimidine analog gemcitabine it was possible to reach an improved response rate for unresectable (Epelbaum et al. 2002; Kornek et al. 2001; Okusaka et al. 2004; Safran et al. 2002) or metastatic (Burris et al. 1997; Carmichael et al. 1996; Casper et al. 1994; Heinemann et al. 1999a; Rothenberg et al. 1996) patients in different stud-ies. Gemcitabine has a favorable side effect pro-file: positive clinical benefit response, practically no hepato- or nephrotoxicity. Only hematotoxic-ity can be seen as a dose-limiting factor. Because of that, radiation-sensitizing effects have been experimentally proved for gemcitabine (Law¬rence et al. 1996; McGinn et al. 1996; Mose et al. 1999; Shewach et al. 1994), which suggests some hope for successful administration of this agent concurrent to radiation.

10.3    Local Remission After Primary (Chemo-)Radiation Therapy
Radiation therapy alone or the combination of external beam radiation therapy (EBRT) with IORT leads to local remission only rarely. The administration of IORT has been explained pri-marily with providing the benefit of good to best pain palliation [pain control in 57% (Okamoto et al. 1994; Tuckson et al. 1988) to 100% (Manabe et al. 1988)].
Whittington et al. (1984) and Mohiuddin et al. (1988) have reported on a patient cohort where the combination of EBRT with iodine-125 seed implantation increased local control (clinical, lo¬cal symptomatic) from 22% (historical patient group) to 81%.
It has been possible to observe objective re-mission in images only recently, since the advent of the concurrent administration of radiation sensitizing agents (chemotherapeutic agents, in the first-line 5-FU or combination therapies with 5-FU) for EBRT (Aristu et al. 2003; Luderhoff
et al. 1996).
The frequency of local remission in up to 50% of patients can be observed when administering up-to-date primary chemoradiation (Ikeda et al. 2002; Li et al. 2003; Rich and Evans 1995; Spitz et al. 1997; Wolff et al. 2000).
It is possible to achieve these response rates especially with gemcitabine-associated chemo-radiation therapy. Remission was seen in 50% vs 13% of the patients in a randomized comparison
of gemcitabine-CRT with 5-FU-CRT. Of a total
of 18 gemcitabine-CRT patients, 4 have shown total remission (Li et al. 2003).
Bruckner et al. (1998) observed downstag-ing in 30% of a group of unresectable patients in International Union Against Cancer (UICC) stages II and III. The therapy concept included radiation therapy (54 Gy) with 5-FU and strep-tozotocin and cisplatin, followed by systemic chemotherapy with 5-FU-folic acid. Local tumor regression was surprisingly positive in resected patients. Fibroses were found in the histology of five of the resected patients, but no cancer cells.
A histological response (tumor cell destruc-tion) has been seen in more than 5 in 10 (50%) of the resected patients from a group of 34 that were examined (Joensuu et al. 2004). These pa-tients were treated primarily with concurrent chemoradiation with gemcitabine. Of these,
3    patients (11%) had a tumor cell destruction of more than 90%.
Wilkowski et al. (2004) have shown in an analysis of 47 patients with primary unresectable carcinoma that, especially during a concurrent sequential CRT with gemcitabine and cisplatin (GC), a high rate of local remission (69% dur¬ing GC-ssqCRT) can be achieved, which can be proved by imaging diagnostics. It is possible to reach complete pathological remissions: R0 re-section was achieved in 13 patients. Of these,
4    patients had no histologically verified tumor.

10.4    Secondary Resection After Primary
(Chemo-)Radiation Therapy
Better resectability and the reduction of postop-erative positive lymph node metastasis has been seen especially in patients with resectable or pos-sibly resectable pancreatic cancer after neoadju-vant chemoradiation.
The restaging after chemoradiation has led the enrolled cancer patients in UICC stage II and III to resection rates between 43% and 74% in Evans et al. (1992), Ishikawa (1996), Rich et al. (1985), and Hoffman et al. (1998). Patients in UICC stage IV were studied by Jeekel and Treurniet-Donker (1991), Bruckner et al. (1998), Todd et al. (1998), and Kim et al. (2002)). Accordingly, resection rates were between 3.4% and 19% af¬ter neoadjuvant multimodality therapy in UICC stage IV.
Resectability can be achieved in patients with primary unresectable pancreatic cancer with primary chemoradiation too. This has been re-ported first of all after concurrent chemora-diation with gemcitabine (Ammori et al. 2003; Brunner et al. 2003; Crane et al. 2002; Epelbaum et al. 2002; Pipas et al. 2001; Wilkowski et al. 2004). R0 resection has been reached only in in¬dividual cases after concurrent administration of PVI 5-FU (application during radiation). Con¬current combination chemotherapies (CDDP+5-FU+/-paclitaxel) (Aristu et al. 2003) has led to complete remission in the framework of a resec¬tion only in individual cases.

10.5    Local and Systemic Progression After Primary (Chemo-)Radiation Therapy
It can be shown at the evaluation of patients' pro-gression samples—either treated neoadjuvant or primarily with (chemo-) radiotherapy (with conventional radiation technique)—that the rate of local recurrence or local progression can be reduced in comparison with historical cohorts; moreover, it can be expected only in 6%-27% of the patients (Ishikawa et al. 1994, 1998; Kornek et al. 2001; Luderhoff et al. 1996; Okusaka et al.
2001).
By contrast, the rate on distant metastases, especially peritoneal carcinosis and liver metas-tasis, was not affected. This rate was assessed in 97% of the patients as a cause for therapy failure
(Okusaka et al. 2004; Poggi et al. 2002; Shinchi et al. 2002).
Time to distant metastasis was even extended in the patient groups if a systemic chemother¬apy was integrated in the treatment regimen as sequential chemotherapy. Kornek et al. (2000) report a progression-free survival of 10 months after concurrent sequential chemoradiation with 5-FU/leucovorin and cisplatin. Favorable ef¬fect was achieved with local liver perfusion with 5-FU too (Ishikawa et al. 1998).
The prognosis was favorable for the group of patients undergoing resection after neoadjuvant or primary (chemo-) radiation therapy (Al-Suk-hun et al. 2003). Downstaging has been found in only 2 of 16 patients in a historical comparison by Jessup (1993), who administered neoadjuvant radiotherapy with 54 Gy and continuous 5-FU infusion. Median survival was 8 months in this group. The two successfully treated and resected neoadjuvant patients had a tumor-free interval of 20 and 22 months, respectively.
Ishikawa et al. (1994) have found in a case control study that preoperative radiation therapy resulted in downstaging and consequently in oncological resection in 17 of 23 patients. These patients had a reduction in local recurrence and have not died of the consequences of them. The first manifestations of disease development in the group of Ishikawa were frequently liver metasta-ses. These metastases have caused lethal conse¬quences more frequently in a year compared to a historical control group.
Median time to progression (TTP) of only 4.4 months was observed by Azria et al. (2002) after sequential chemo-radiotherapy (sCRT) with 5-FU (600 mg/m2, day 1-5, week 1 and 5) and cisplatin (100 mg/nr*, day 2, weeks 1 and 5).
Okusaka et al. reporton a median progres-sion-free interval of 5.8 months and a therapy failure due to distant metastasis (78%) in a series of 41 primary unresectable patients treated with ssqCRT (concurrent cisplatin 5 mg/m2 per day, sequential 5-FU). Only 21% of the patients de-veloped a local recurrence.
A median TTP of 2.7 vs 7.1 months (p = 0.019), a median TTLP of 2.7 vs 7.4 months (p = 0.0016), and a median time to systemic progression (TTSP) of 3.1 vs 6.1 months (n.s.) have been ob¬served in a randomized trial by Li et al. (2003) where 5-FU-CRT vs gemcitabine-CRT has been compared.

10.6    Overall Survival After Primary (Chemo-)Radiotherapy
In 1969, Moertel et al. were able to achieve a bet-ter median survival in locally advanced pancre-atic cancer with the administration of the com-bination of external radiation (EBRT) and 5-FU compared to radiation alone. Hereby was the im-portance of radio sensitizing through concurrent chemotherapy established.
5-FU, as an integrative part in CRT, has not been substituted yet by other agents, new combi-nations, or regional applications. The role of PVI 5-FU application, the administration of gem-citabine as radio sensitizer, or hyperfractionation of radiation during CRT is not clear.
Significant advantage of gemcitabine-CRT (concurrent gemcitabine, weekly 600 mg/m2 per day 6x and sequential gemcitabine) compared
to 5-FU-CRT (bolus 5-FU 500 mg/m2 per day,
days 1-3 and sequential gemcitabine) has been shown only in one randomized study (Li et al. 2003). However, there have been only 18 vs 16 patients compared, and 5-FU has been applied as bolus on three subsequent days of weeks 1, 3, and 5 of radiotherapy. Median survival of 14.5 months was achieved with gemcitabine-CRT, 6.7 months with 5-FU-CRT.
However, a median survival of 13.2 months has been reported in a further randomized trial
by Shinchi et al. (2002) with PVI 5-FU-CRT
(200 mg/m2 per day). Median survival was 6.4 months in case of an untreated group of pa¬tients.
We should conclude in comparing these two studies alone that gemcitabine-CRT can be seen as equivalent to PVI 5-FU-CRT. A retrospective analysis by Mehta et al. (2001) does not support this conclusion. He has compared 27 patients
with either bolus 5-FU or PVI 5-FU CRT. Me¬dian survival was 6 months for both groups in this trial. However, bolus 5-FU-CRT shows no improvement as to overall survival in compari¬son with untreated patients.
Whereas concurrent CRT leads to favorable local tumor control, this procedure has a minor effect as to the survival in most of the studies. This is because of the development of metastases out of the irradiation field. The concept of com-bined CRT with continuing chemotherapy was developed as a logical consequence due to these metastases and because of partly advanced local tumor progression. The Gastrointestinal Tumour Study Group (GITSG) completed a trial in 1981 that divided the patients in three study arms as follows: 60 Gy EBRT without radio sensitizing with 5-FU, 60 Gy EBRT with radio sensitizing with 5-FU and subsequent 5-FU application, 40 Gy EBRT with radio sensitizing with 5-FU and subsequent 5-FU application. Median sur¬vival was 23, 40, and 42 weeks, respectively. That is, a high dose of chemotherapy had no addi¬tional use. Better results were reached especially with concurrent or subsequent chemotherapy.
A median survival of 8.2 months after chemo-therapy alone with weekly bolus 5-FU has been reported in the phase III study of the Eastern Cooperative Oncology Group (ECOG) of a to¬tal of 91 patients with unresectable pancreatic cancer. A median survival of 8.3 months has been seen in the direct comparison within the group treated with 5-FU-CRT and subsequent chemotherapy (Klaasen et al. 1985). Significantly higher toxicity (51%) has been observed in the study arm with combination therapy compared to 5-FU therapy alone (27%).
It has been possible to show in the phase I/II study of the Radiation Therapy Oncology Group (RTOG) enrolling 81 patients that, in spite of the fact that a prophylactic liver radiation can reduce the incidence of hepatic metastasis, this proce-dure can affect neither the local tumor control nor the intraabdominal tumor spread, and con-sequently, prophylactic liver irradiation cannot be generally recommended (Komaki et al. 1992).
A median survival of 17 months was seen in recent studies after CRT combined with triple chemotherapy (5-FU, streptozocin, and cispla-tin) (Terk et al. 1997). Favorable median survival of 14 months was shown in a further trial of combined chemoradiation with 5-FU, leucovo-rin, and cisplatin (Kornek et al. 2000).
Hypofractionation (Luderhoff et al. 1996) or hyperfractionation (Crane et al. 2001; de Lange et al. 2002) of the irradiation in the framework of chemoradiation and sequential chemotherapy has been also administered. These regimens can¬not be recommended as routine therapies due to considerable toxicities.
No survival advantage was achieved for pa-tients with combined or IORT or brachytherapy alone (Nishimura et al. 1988; Calvo et al. 1991; Fossati et al. 1995; Kasperk et al. 1995; Okamoto
et al. 2004).
Although recent studies have been report¬ing about acceptable survival after CRT, the results are not convincing in comparison with chemotherapy alone (especially using gem-citabine) (Fisher et al. 1999; Saad et al. 2002); possibly, it would be seen as even more unfavor¬able for the total cohort of patients (Chauffert
et al. 2006).

10.7    Effect of Radiation Dose
on Progression and Overall Survival After Primary (Chemo-)Radiotherapy
The initial data about the irradiation of pancreatic cancer have shown that "very high" doses would have been considered necessary in order to treat local tumors effectively. The effect of radiation has been reported as depending on dose, and a median survival of 10 months was achieved with a total dose of 68-75 Gy in the target field (Wie-gel et al. 2000). A cure after radiation alone has been assessed as barely possible. Accordingly, the further aspect of local palliation (reduction of pain symptomatic in 65%-70% of the cases) has been formulated as the treatment aim. Necessary dose escalation was reached through IORT, even¬tually followed by EBRT. It was not possible to show a convincing remission rate or a substantial improvement of progression-free and total sur¬vival in studies with EBRT alone (Moertel et al. 1969, 1981), with IORT (Abe et al. 1993; Goldson 1991; Nishimura et al. 1988), or with a combina¬tion of EBRT and IORT (Kawamura et al. 1992;
Gilly et al. 1990), although a median survival of 12.0 months has been reported in patients in stages I and II, respectively, with the combined use of EBRT and IORT (Manabe et al. 1988; Abe et al. 1991, 1993; Okamoto et al. 1994).
Downstaging for potential resectable patients has been reported first with the administration of chemoradiation with 5-FU and the adminis¬tration of a radiation dose of 45-54 Gy (Scherer 1987). However, it has been shown that with the additional administration of IORT, a me¬dian survival of 18 months can be reached only for resectable patients, and the median survival amounts to 8.0-16.5 months for unresectable pa-tients in this case (Kojima et al. 1991; Wood et al. 1982; Shipley et al. 1984).
The concept of dose escalation with IORT has been discussed in recent studies in combination with chemoradiation as a rule (Okamoto et al. 2003). Convincing data were not shown in com-parison with chemoradiation alone. Accordingly, Okamoto et al. (2003) have reported a median overall survival of 8.6 months.
Thus, it can be formulated that the "effectuat-ing" of the therapy for locally advanced pancre-atic patients can be strived for with radiation dose escalation only in cases where the "optimization" of the concurrent or sequential chemotherapy regimen can affect the systemic progression pro¬cedure of the carcinoma significantly. This is sup¬ported by the fact that local progression results after chemoradiation can be seen substantially later in the median survival than in the overall survival of the patients. That means that patients have no local relapses or local progression after chemoradiation with a local dose of 50.0 Gy, which is considered moderate regarding possible late consequences.

10.8    Primary Concurrent
Chemoradiation with Gemcitabine
Since the effectiveness of gemcitabine compared to 5-FU or 5-FU-associated polychemotherapies has been proved in the treatment of pancreatic patients with metastasis (Burris et al. 1997), this agent has been examined also as a combination agent for concurrent chemoradiation in the treat-ment of locally advanced pancreatic cancer.
The selected therapy applications of concur-rent chemotherapy with gemcitabine were and are:
-    Normal gemcitabine chemotherapy dose (or combination therapy (Muler et al. 2004)) with total radiation dose- and/or volume reduc¬tion with accelerated radiation (McGinn et al. 2001; Muler et al. 2004)
-    Reduced gemcitabine chemotherapy with to¬tal dose reduction with accelerated radiation (Crane et al. 2001)
-    Reduced gemcitabine chemotherapy dose with hypofractionated radiation (de Lange
et al. 2002)
-    Escalation of gemcitabine chemotherapy dose with weekly applications up to a toxic dose with normal dose radiation (Morganti et al. 2003; Poggi et al. 2002; Safran et al. 2002)
-    De-escalation of gemcitabine chemotherapy dose with weekly applications to a tolerable dose with normal dose radiation (Brunner
et al. 2003)
-    Reduced gemcitabine chemotherapy dose with weekly applications and a normal dose radiation (Ikeda et al. 2002; Okusaka et al.
2004)
-    Reduced gemcitabine chemotherapy dose with 24-h infusion and weekly applications and normal dose radiation ((Kudrimoti et al. 1999; Mohiuddin et al. 2002) respectively)
-    Reduced gemcitabine chemotherapy dose with weekly applications and "split course" ra-diation (Van Laethem et al. 2003)
-    Reduced gemcitabine chemotherapy doses twice a week and normal dose radiation (Pi-pas et al. 2001; Blackstock et al. 1999, 2003)
-    Reduced gemcitabine chemotherapy doses three times a week and normal dose radiation
(Epelbaum et al. 2002)
-    Combined chemotherapy, gemcitabine with 5-FU and normal dose radiation (Talamonti
et al. 2000; Wilkowski et al. 2000)
-    Combined chemotherapy, gemcitabine with cisplatin (dose reduced or normal dose) and normal dose or accelerated radiation (Mar-
tenson et al. 2003; Wilkowski et al. 2002)
Tolerable concurrent gemcitabine doses with weekly application are defined as 250 mg/m*-300 mg/m2 with conventional dose radiation (up to 50.4 Gy and 28 fractions) and with the involvement of regional lymph nodes in the ra-diation volume (TV II) (Wolff et al. 2001; Ikeda
et al. 2002; Morganti et al. 2003). Gemcitabine
doses could be increased to 440 mg/m2 per day with weekly applications when there is a reduc-tion of radiation volume (only with the involve-ment of macroscopic tumor = TV I) (Poggi et al.
2002).
A gemcitabine dose of 300 mg/m2 has been applied with hypofractionated radiation (3 x 8.0 Gy) (de Lange et al. 2002).
A maximum gemcitabine dose of 300 mg/m2 or 75 mg/m2 per day, and weekly application for each, has been determined for the purposes of a concurrent combined chemotherapy with cis-platin or paclitaxel, respectively (Brunner et al.
2003; Safran et al. 2002).
It has also been proved to be possible to ap¬ply a dose of 40-90 mg/m2 per day gemcitabine every 2 weeks concurrent with normal dose ra-diation (Blackstock et al. 1999, 2003; Yavuz et al.
2001).
Kornek et al. have examined the concurrent 24-h infusion of gemcitabine and determined an applicable dose of 130 mg/nr* per day and weekly application.
It can be seen on the basis of the data above that neither the gemcitabine dose, the applica¬tion rhythm, nor the applicable radiation dose or radiation volume could have been recommended as standard up to now.
The problem is that the different combinations of gemcitabine with radiation have led to differ-ent toxicity profiles which partly correspond to grade 3 or 4 toxicity.
Generally, a normal dose or weekly reduced gemcitabine dose, even with a reduced radiation dose or reduced volume, has led to increased gastrointestinal complications (McGinn et al. 2001; Muler et al. 2004). Hematotoxicity (leuko-cytopenia and/or thrombocytopenia grade 3-4 up to 66%) has been seen with high probability with the dose reduction of gemcitabine and normal dose radiation  (Wolff et al. 2001;
Blackstock et al. 2003; Okusaka et al. 2004).
Dose-reduced combined chemotherapies with normal dose radiation have led to significant hematotoxicity too (Wilkowski et al. 2002; Brun-
ner et al. 2003).
One possible solution for the problem has been seen in the application of gemcitabine as sequential chemotherapy alone (Li et al. 2003; Kachnic et al. 2001; Ben-Josef et al. 2004) and the concurrent application of a chemotherapy with 5-FU or capecitabine (Ben-Josef et al. 2004).
It has been shown that for the majority of pa-tients treated concurrently with gemcitabine, a local remission could have been achieved with the possibility of secondary R0 resection in comparison with patients treated with concur¬rent 5-FU (Crane et al. 2002), independently from toxicity, which was assessed as moderate (Brunner et al. 2003) to intolerable (Talamonti et al. 2000). Response rates have been reported in
29%-50% of the patients (de Lange et al. 2002;
Ikeda et al. 2002; Li et al. 2003), and absence of tumor after neoadjuvant treatment and second-ary resection (ypT0 stages) has been observed
(Brunner et al. 2003; Epelbaum et al. 2002). Thus, pathologically complete remissions are possible after gemcitabine-associated CRT.
Median TTP has been reported between 4.4 (Okusaka et al. 2004) and 7.1 months (de Lange et al. 2002; Li et al. 2003). Local therapy failure (local progression or local relapse after remis¬sion) has been observed in up to 65% (de Lange et al. 2002) and distant metastasis in 75% of the patients (de Lange et al. 2002). Median overall survival was between 8.3 (Kornek et al. 2001)
and 14.5 months (Li et al. 2003).
Concurrent chemoradiation with 5-FU has been compared with gemcitabine monotheapy only in one prospective randomized study; se-quential chemotherapy has been administered in both arms with gemcitabine monotherapy until progress (Li et al. 2003). Significantly bet¬ter remission rates (13% vs 50%) and signifi¬cantly improved median progression-free (2.7 vs 7.1 months) and median overall survival (6.7 vs 14.1 months) have been observed only in the case of a few patients (16 vs 18 patients) during gemcitabine-CRT.

10.9    Primary (Chemo-)Radiotherapy: Does It Make Sense and Is It Efficient?
Different studies (Crane et al. 2002; Li et al. 2003) point out that sequential concurrent chemora-diation, especially with gemcitabine, positively affect the progression-free and overall survival in primary unresectable pancreatic patients in comparison with concurrent chemoradiation and radiotherapy alone. Prospective randomized studies are warranted to prove the significant differences as to different combinations of con¬current and sequential chemotherapy with 5-FU +/-gemcitabine or gemcitabine+/-cisplatin.
IORT alone or in combination should not be induced because there have been negative results with possible deterioration of the prognosis.
The most important positive prognostic fac¬tor is remission after chemoradiation. Especially patients undergoing secondary R0 resection after image-proved remission show a significant im-provement in progression-free and overall sur-vival and reach can 5-year survival (24.9% of the patient cohort; R. Wilkowski, unpublished).
The therapy aim for locally advanced primary unresectable nonmetastatic pancreatic cancer is the local treatment of the disease, considering its systemic aggressiveness. A proper combination agent should be found for concurrent chemora-diation that has a high systemic effect and/or is capable of achieving this effectiveness in sequen¬tial application.

Neoadjuvant and Adjuvant Strategies for Chemoradiation

 

There is an increasing body of evidence showing that patients with resectable pancreatic cancer might benefit from adjuvant therapy. Based on phase III trials, potential options for adjuvant treatment are chemotherapy alone or a multi-modal approach involving radiotherapy. Avail¬able data are heterogeneous and have been dis¬cussed controversially. Hitherto, a worldwide standard of care has not yet been established. Adequate patient selection might be the key ele-ment for a tailored adjuvant treatment. Clinical research currently focusses on gemcitabine alone or in combination, and some molecular biologic approaches with epidermal growth factor recep-tor monoclonal antibodies (EGFR-MoABs) and anti-angiogenic drugs. Recent advances in ra-diooncology offer better dose conformality and reduced morbidities. Currently, the co-operative Radiotherapy and Gastrointestinal Groups have launched a multicentric European Organization for Research and Treatment of Cancer (EORTC) trial investigating the impact of radiotherapy in combination with gemcitabine in R0-resected pancreatic head cancer.

9.1      Introduction
Only about 10%-20% of pancreatic cancer pa-tients are deemed to be resectable and could be ideal candidates for adjuvant or neoadjuvant treatment strategies (Evans 2005; Kelly and Ben-jamin 1995; Sener et al. 1999). Surgery is the only possibly curative treatment option. Remarkable progress has been made in terms of clear resec¬tion margins, which could be increased from 26% to 43%. From the 1970s through the 1990s, peri-operative mortality improved from 30% to 0.9%, respectively. Five-year survival rates increased from 14% in the 1970s to more than 30% in the 1990s (Yeo et al. 1995; Yeo and Cameron 1999). In the 1990s more than 65% of patients received some form of adjuvant therapy compared to less than 25% in the 1980s. This reflects a major change in the treatment paradigms from thera-peutic nihilism to intensified adjuvant therapy.
Prognostic factors after surgery are perfor-mance status, extent of tumour spread and tu-mour size, nodal status, grading and status of resection margins (Kalser et al. 1985; Kalser and Ellenberg 1985; Neoptolemos et al. 2001, 2004; Sohn et al. 2000). Additional and well-known factors are blood loss during pancreaticoduo-denectomy and time to recovery. Less common ampullary carcinoma and intrapancreatic bile duct carcinoma have a more favourable prog¬nosis than pancreatic ductal adenocarcinoma (Magee et al. 2002). For pancreatic cancer, a pro¬longed median survival of about 15-20 months can be expected after R0 surgery compared to 8 to 12 months after R1 surgery (Evans et al. 1998). Even after R0 surgery, however, relapses occur regularly. Local recurrences account for the ma¬jority (80%), recurrence in the peritoneal cavity for 25%, and liver metastasis for about 50% of all cases (Wayne et al. 2002). Metastases to other re¬gions such as lung are rare and usually occur at a late stage.
Though there is remarkable progress of pan-creatic cancer treatment, survival data are still dismal. Therefore, adjuvant therapy is of major importance.
9.2     Adjuvant Chemoradiation
The Gastrointestinal Tumor Study Group (GITSG) GI-9173 data published in 1985 has proved that post-operative chemora-diation (CRT) is highly effective (Kalser and Ellenberg 1985). This randomized trial investi¬gated surgery followed by a 40-Gy split-course radiotherapy (RT) combined with 5-fluorouracil (5-FU) versus surgery alone. The treatment arm (n = 21) was superior to surgery alone (n = 22), resulting in median survival of 20 months ver¬sus 12 months. These results were confirmed by the GITSG in a non-randomized controlled phase II study in 1987 (Gastrointestinal Tumor Study Group 1987). Both studies had great im¬pact on adjuvant treatment of pancreatic cancer in the USA. This treatment regimen has become the new standard of care, and until recently it was in widespread use in the United States. A number of additional studies confirmed these re¬sults (Foo et al. 1993; Foo and Gunderson 1998; Mehta et al. 2000; Paulino 1999; Yeo et al. 1995, 1997). The combination of 5-FU with leucovorin seemed to be only marginally effective (Abrams et al. 1999). A recent study from Johns Hopkins University compared two schemes of chemora-diation with surveillance. Chemoradiation was either intensified 50-57 Gy to the pancreas and a prophylactic dose of 23-27 Gy to the liver com-bined with 5-FU or a standard dose of 40-45 Gy to the pancreas combined with 5-FU/leucovo-rin. Patients receiving adjuvant therapy had a median survival of 19.5 months compared with only 13.5 months after resection alone. More in¬tense treatment schemes did not appear to fur¬ther improve survival (Sohn et al. 2000; Yeo et al. 1997). A European Organization for Research and Treatment of Cancer (EORTC) multi-centre trial including 207 evaluable patients combined split-course RT with 5-FU and also found a sta-tistically insignificant improved survival trend of 24.5 versus 19 months (Klinkenbijl et al. 1999). Subgroup analysis indicated a benefit for pan-creatic head cancer patients with survival of 17.1 versus 12.6 months.
Some of these studies have methodological limitations. Patient accrual was either slow or the numbers of patients were statistically insufficient.
Patient selection was non-uniform, thereby in-cluding patients with pancreatic cancer and those with periampullary cancer, who have a better prognosis. Mono-institutional studies suffered from selection bias. Up to 25% of the patients did not receive the planned radiochemotherapy because of withdrawal of consent, lack of post-operative recovery or rapid tumour progression. There was no stratification for tumour sites, nor was there a detailed analysis of resection mar-gins. From the current point of view, treatment was often suboptimal, with split-course RT and a heterogeneous dose distribution, and 5-FU given as bolus instead of as continuous infusion.
A rather modern treatment scheme was ap-plied to 52 patients by Mehta et al. (2000). RT
was intensified from 45 Gy (R0) to 54-60 Gy (R1) and in a few cases intraoperative RT was added. This was combined with a continuous infusion of 5-FU. The resulting median survival was a promising 32 months and morbidities were only moderate.

9.3     Adjuvant Chemotherapy
The first randomized trial to demonstrate a posi-tive effect of adjuvant chemotherapy without RT was published in 1993 (Bakkevold et al. 1993). Sixty-one radically resected patients were ran-domized either for post-operative adjuvant com-bination chemotherapy using 5-FU, doxorubicin and mitomycin C (AMF), or as controls (no ad-juvant chemotherapy). The median survival in the treatment group was 23 months compared to 11 months in the control group. The authors con-cluded that adjuvant chemotherapy prolongs the incidence of recurrence during the first 2 years following radical surgery; however, an increased cure rate was not observed.
The benefits of chemoradiation have been questioned since the publication of the European Study Group for Pancreatic Cancer (ESPAC) 1 trial (Neoptolemos et al. 2001, 2004). This was the most ambitious and largest adjuvant trial in pancreatic cancer with 548 patients involved. It had a 2x2 factorial design: of 541 eligible pa-tients, 289 were assigned to chemotherapy versus radiochemotherapy, 68 to chemoradiotherapy only versus no chemoradiotherapy with record of background chemotherapy and 188 to chemo-therapy only versus no chemotherapy with record of background chemotherapy. The 289 patients of the first group were assigned to observation, chemoradiotherapy, chemotherapy, or chemo-radiotherapy and chemotherapy. The treatment concepts were those of other published trials: surgery followed by 5-FU/FA bolus, surgery fol¬lowed by 40 Gy split-course RT+5-FU bolus, or surgery followed by both. In summary the 2x2 factorial design showed no survival benefit with CT or CRT to observation. When all patients were pooled, a survival benefit for adjuvant CT, not for adjuvant CRT, was observed. The Kaplan-Meier analysis showed a reduced survival rate following radiochemotherapy compared to no radiochemotherapy of 29% versus 41%, respec-tively, after 2 years and of 10% versus 20% after 5 years, which was statistically significant.
These data actually suggest that adjuvant ra-diochemotherapy might even harm patients af-ter a potentially curative resection. This trial had substantial impact on therapeutic decisions in Europe, where radiochemotherapy has virtually been abandoned and adjuvant chemotherapy has become the European standard of care.
The results of ESPAC-1, however, need to be discussed, and substantial criticisms have been raised. The trial used various randomization pro-cedures. Because of interaction between the ther-apy arms, the study might be regarded as under-powered. Background chemotherapy was allowed, which was not part of the study medication. Ad¬ditionally, about 40% of the patients did not re¬ceive the originally planned treatment. There was no standardized quality assurance for surgery nor for the radiochemotherapy procedures, and 30% of the RT patients received a non-uniform dose or even no radiation at all. Chemoradiation started on average 2 weeks later than chemotherapy. Nei¬ther RT nor chemotherapy was optimal accord¬ing to modern standards. And last but not least, from a tumour biologic point of view, there is no rationale conceivable for a shorter relapse-free survival with chemoradiation. Insufficient treat-ment quality for the chemoradiation arm might also have contributed to the enhanced treatment-related toxicity (Choti 2004; Koshy et al. 2005). In summary, the published results of the ESPAC-1 trial have to be considered premature and are in no way suited to rule out adjuvant chemoradia-tion for pancreatic cancer.
Recently, a meta-analysis summarized indi-vidual data of 875 patients treated in five ran-domized controlled trials (Stocken et al. 2005). The ESPAC-1 trial contributed 550 patients. The pooled hazard ratio reduction was 25% corre-sponding to a significant reduction of the death risk after chemotherapy. The median survival was 19 months with, and 13.5 months without, chemotherapy, respectively. In this meta-analy-sis an increased risk of death with the addition of RT to chemotherapy could not be observed. The median survival was the same (15.8 months with chemoradiation and 15.2 without). Sub¬group analysis revealed that chemoradiation was more effective compared with chemotherapy in patients with positive resection margins. Still, the authors concluded that the initial use of chemo-radiotherapy might have delayed the effective use of chemotherapy and thereby reduced survival.
The ESPAC-3 trial has meanwhile launched to confirm the beneficial role of adjuvant che-motherapy and to differentiate between the ef-ficacies of 5-FU and gemcitabine (Neoptolemos et al. 2003). A randomization between 5-FU/FA and gemcitabine is used; for ampullary cancer a third arm with observation only is installed. Af¬ter amendment, 680 patients need to be enrolled, so final results will take some time.
Gemcitabine has come into focus recently. It is a potent radiation sensitizer of pancreatic tu¬mour cells in vitro (Lawrence et al. 1996). It is active in pancreatic cancer and improves clini¬cal benefit (23.8 versus 4.8%) and survival rate (5.7 versus 4.4 months) over 5-FU, which led to the Gemzar (Eli Lilly, Indianapolis) registration (Burris and Storniolo 1997; Burris et al. 1997; Rothenberg et al. 1996). Toxicity is minimal and it can be administered on an outpatient setting.
Preliminary data of the CONKO-001 study, presented at the American Society of Clinical Oncology annual meeting (ASCO) 2005, con-firmed the efficacy of gemcitabine. In a random-ized manner, median survival with gemcitabine was 14 months, without it, 7 months (Neuhaus
et al. 2005).
9.4     Pre-operative Chemoradiotherapy
The concept of pre-operative radiochemotherapy is alluring, but its value for resectable cancer is not yet clear. Using radiochemotherapy first, a possibly systemic disease could receive a sys-temic treatment without the delay of post-opera-tive recovery. Multi-modal treatment ought to start with the least toxic treatment, thus sparing surgery in patients, who currently suffer from rapid progression or metastasis. Further possible advantages have been suggested, including the prevention of intraoperative tumour spread and improved radiosensitivity, as tumour oxygen-ation has not been hampered by surgery and the higher probability of R0 resection (Bergenfeldt and Albertsson 2006; Crane et al. 2006b; Evans
2005).
There have been very promising data from a sequence of trials performed by the MD An¬derson Cancer Centre. Only patients with a pos¬sibly resectable disease were included, surgical techniques and pathological evaluation were standardized. First, 5-FU and concomitant RT with 50.4 Gy did result in a median survival of
18    months. Toxicity, though, was severe enough to necessitate hospital admission in one-third of the patients (Evans et al. 1992). The subsequent trials did use an accelerated, so-called "rapid frac¬tionation" programme delivered over 2 weeks, with a 30-Gy total dose and a 3-Gy single frac¬tion dose. During surgery an additional intraop-erative RT of 10-15 Gy was administered. Com¬bining this scheme with 5-FU led to a median survival of 25 months (Pisters et al. 1998). Pacli-taxel was less effective, with a median survival of
19    months (Pisters et al. 2002). Best results were achieved when combining RT with gemcitabine. Median survival was 36 months, toxicities were manageable, with a hospitalization rate of 43% (Wolff et al. 2001). These trials all suffered from a positive selection bias. Recently, results of a French phase II study have been published com-bining 5-FU with cisplatin and concurrent RT to 50 Gy in a pre-operative setting for potentially resectable pancreatic cancer. Among 40 evalu-able patients, 15 did not undergo resection of the pancreatic tumour because of local or metastatic progression. Median survival for those patients, who completed treatment, was 11.7 months. The scheme was concluded to be feasible, but the use of more efficient drugs such as gemcitabine and optimized RT seemed justified (Mornex et al.
2006).
Numerous trials investigated pre-operative chemoradiotherapy (Bergenfeldt and Albertsson 2006) based on heterogeneous patient populations including patients with locally advanced disease. Pre-operative chemoradiation seems to offer a downstaging effect, shifting patients from locally advanced to potentially resectable stages (White et al. 1999, 2001). Median survival in those pa¬tients who were resected exceeded 16 months. When comparing 5-FU-based chemoradiation in a pre-operative and post-operative setting, actuarial survival rates at 2, 3 and 5 years were
39% versus 52%, 35% versus 40% and 28% ver-sus 40%, respectively, in favour of adjuvant treat¬ment. This difference did not reach statistical significance and was attributed to larger, more locally advanced tumours in the preoperative therapy group (Spitz et al. 1997). Pathological findings after pre-operative chemoradiotherapy showed fewer involved lymph nodes, more nega-tive resection margins, similar toxicity and a non-significant median survival difference of 20 versus 25 months (Pendurthi et al. 1998).
So far, randomized controlled trials proving an overall survival benefit for neoadjuvant treat¬ment approaches are missing.

9.5     Intraoperative Radiotherapy
A possible advantage of intraoperative radiother-apy (IORT) is the ability to deliver high doses of radiation to sites at high risk of local recurrence while organs at risk can be shielded. IORT has been used in the adjuvant and neoadjuvant situa-tion, alone or in combination with chemotherapy (for an overview see Bergenfeldt and Alberts-son 2006). IORT has been proved effective over surgery alone in pancreatic cancer with respect to local recurrence, which was reduced by half (Reni et al. 2001). This trial also found a survival benefit for selected stage I-II patients. Operative morbidity and mortality were not increased. In a subgroup analysis, a combination of IORT with
RT and chemotherapy improved survival signifi-cantly (Di Carlo et al. 1997).
IORT can act as a valuable partner in a combined modality treatment setting that in-cludes chemoradiation. However, accelerated re-population during the interval between IORT and external RT has to be taken into account
(Wilkowski et al. 2005).

9.6     Radiotherapy Treatment Planning
Many of the patients treated in the aforemen-tioned trials had their RT based on 2D-treat-ment planning, which led to large dose burden on the surrounding healthy normal tissue and consequential acute and late radiation damage. The introduction of a split-course technique and a total dose not exceeding 40 Gy was mainly driven by the acute toxicities observed. 2D dose distributions in the target volume often were less than optimal, leading to cold and hot spots. In the meantime, technical progress enables 3D-treatment planning with optimized protec¬tion of normal tissue, thus offering the opportu¬nity to treat without a split to higher total doses of 54 to 60 Gy. Still, the critical dose-limiting struc¬tures neighbouring the target volume is the small bowel, which has a tolerance dose of 45-50 Gy, depending on the single fraction dose, and the kidney, which has a tolerance dose of about 23 Gy to the whole organ. With regard to a potentially enhanced toxicity by combining with chemo¬therapy, no more than 30% of the kidneys should reach a dose level of 20 Gy. In contrast to these structures, the liver, as an organ with great regen¬eration capacities, tolerates much higher doses, up to 50 Gy in up to 1/3 of the liver volume. A rather conservative approach with regard to pos¬sible toxic damage of the liver is 12.5 Gy to 75%, 25 Gy to 50% and 37.5 Gy to 25% of liver volume (Emami et al. 1991; Wilkowski et al. 2005).
Pancreatic cancer carries a high risk, exceeding 80%, for dissemination to loco-regional lymph nodes. The lymphatic drainage from the pancreas consists of peripancreatic nodes and along the upper mesenteric artery, the a. gastroduodena-lis, a. hepatica communis, a. lienalis and coeliac trunk. Involvement of nodes near the portal vein and para-aortal and para-caval nodes happens frequently (Kayahara et al. 1995, 1996, 1999).
Treatment planning should be based on a pre-operative 3D data set with intravenous con-trast medium and contrast enhancement of the small bowel. The patients should be immobi¬lized with raised arms. A clinical target volume (CTV) should be defined, which includes the primary tumour region, involved lymph nodes and any subclinical region at risk, with an addi¬tional margin of 0.5 cm. To reduce toxicity, part of the pancreatic tail, which is not involved, may be excluded. Loco-regional lymph nodes should be included, at least between the upper mes-enteric artery and coeliac trunk. The planning target volume is equal to the CTV plus a safety margin to account for patient and organ move¬ment of 1-3 cm. An extension of the target vol¬ume to include the liver for prophylactic reasons did not result in an increased survival (Yeo et al. 1997). Shrinking of the planning target volume (PTV) after 45 Gy is recommended. Dose should be prescribed according to the guidelines of the International Commission on Radiation Units, report 50 (ICRU-50), dose heterogeneity should not exceed ±5% (see EORTC 40013 (Wilkowski et al. 2005)). When sophisticated techniques such as intensity modulated RT (IMRT) or stereotac-
tically guided RT (SRT) are involved, ICRU-50
criteria might not be fulfilled; still, dose homoge-neity should be aimed for.

9.7     Advances in Radiotherapy
Stereotactically guided RT is a means to further reduce the irradiated volume by shrinking the PTV (Fig. 9.1). The technique is still emerging since set-up uncertainties caused by breathing and organ motion are of concern. A stereotacti-cally guided boost of 25 Gy following an IMRT treatment of 45 Gy combined with 5-FU enabled dose escalation up to 70 Gy with a tolerable acute toxicity. Of 19 patients, 16 completed treatment, local control was excellent and overall survival was not influenced in these patients suffering from advanced disease (Koong et al. 2005). On the other hand, a trial from Denmark reports unacceptable toxicity, poor outcome and a questionable palliative effect of a fractionated stereo-tactically guided treatment of 15 x 3 Gy, again in locally advanced cancer (Hoyer et al. 2005).
Inverse treatment planning and IMRT allows specific dose distributions in order to either es-calate the total dose given to the tumour with¬out a further protection of surrounding organs at risk, such as the small bowel, or to better pro¬tect the organs at risk at the same total dose level (Landry et al. 2002). In general, the high dose volume is reduced, whereas the low dose volume is increased. A dosimetric analysis using differ¬ent planning techniques revealed an advantage for using an integrated boost with doses of up to 64.8 Gy, which could have been given safely (Brown et al. 2006). A first phase I study combin¬ing gemcitabine 350 mg/m2 with IMRT 33 Gy in 11 fractions had to be closed due to exces-sive toxicity. Even after reducing gemcitabine to 250 mg/m2, patients suffered from dose-limit¬ing gastrointestinal toxicity and myelosuppres-sion (Crane et al. 2001a). A trial using IMRT as a boost following conventionally fractionated RT of 30 Gy with another 21-30 Gy, and a 3-Gy single fraction dose combined with 5-FU had a tolerable acute toxicity and some palliative ef¬fect in locally advanced cancer (Bai et al. 2003). Without dose escalation, IMRT has been found to be effective and tolerable in combination with capecitabine (Ben-Josef et al. 2004). Currently, the PARC study is investigating IMRT in combi¬nation with cetuximab and gemcitabine (Krem-
pien et al. 2005).

9.8     Advances in Chemotherapy
in the Combined Therapy Setting
Gemcitabine has been widely investigated in the adjuvant and neoadjuvant setting to replace 5-FU-based regimens. In the United States, the randomized Radiation Therapy Oncology Group
(RTOG) 9704 trial tested the sequence 5-FU/
FA-5-FU/RT-5-FU/FA versus gemcitabine (Gem)->5-FU/RT->Gem. The data were pre¬sented at the ASCO 2006 meeting (Regine and Abrams 1998; Saif 2006). From 1998 to 2002, 538 patients entered the trial (stage T1-T4, N0-N1); 381 had pancreatic head carcinoma, and 442 pa-tients were eligible and analysable. Haematologi-cal toxicity was elevated in the gemcitabine arm, but manageable. For patients with pancreatic head carcinoma, median survival improved to 36.9 months compared to 20.6 months without gemcitabine. There was no improvement in sur¬vival for patients with tumours of the pancreatic body or tail. It was concluded that gemcitabine might be considered as a new standard adjuvant therapy, at least for pancreatic head carcinoma.
Data on gemcitabine combined with RT is emerging, but dose finding is still an issue. About 170 patients were treated in approx. 15 phase II trials. The gemcitabine dose was limited to 300 mg/m2 weekly. The RT dose was 50.4 Gy for a limited target volume of less than 1,500 cm3 (Van Laethem et al. 2003). There was a promising activity combined with moderate and manage-able haematological and gastrointestinal toxic-ity. Median survival was 15 months, disease-free survival was 6 months; grade 3-4 haematologi-cal and non-haematological toxicity was 25%-36%. The Eastern Cooperative Oncology Group (ECOG) 4201 study evaluating gemcitabine plus RT versus gemcitabine alone was closed recently.
There were dose-finding phase I/II-studies of gemcitabine with concurrent radiation for advanced pancreatic cancer with a significant grade 3-4 toxicity of anorexia and dehydration when doses exceeded 350 mg/m2 (McGinn and Zalupski 2003; Wolff et al. 2001). Toxicity was related to radiation dose. The infusion rate is re-lated to the systemic efficacy (Crane et al. 2001b). Since gemcitabine as a prodrug must be phos-phorylated to its active metabolites there is tissue saturation. Administration with a fixed dose rate is feasible (Tempero et al. 2003); data on combi¬nation with RT is not available.
In general, the therapeutic index seemed to be rather narrow (Crane et al. 2002). A dose of 40 mg/m2 twice weekly concurrent to RT fol¬lowing an induction therapy of irinotecan/gem-citabine was feasible too (Blackstock et al. 1999, 2002) but of only moderate activity (Mishra et al. 2005). Combination of gemcitabine with cisplatin and concomitant with RT was feasible (Wilkowski et al. 2003). Other partners are un¬der evaluation. Of special interest are non-cyto-toxic partners such as trastuzumab or epidermal growth factor receptor (EGFR) antibodies. Inhib-itors of the ras protein, metalloproteinases, COX inhibitors, and vascular-endothelial-growth-re-ceptor (VEGF) inhibitors are being investigated pre-clinically or in early clinical studies (Wayne
et al. 2002).
Three major randomized studies are evaluat-ing the role of bevacizumab and cetuximab with gemcitabine and irinotecan with docetaxel (Saif 2006). The ECOG 2204 trial, activated in Febru-ary 2006 is a phase II randomized study of ad-juvant therapy comprising bevacizumab versus cetuximab in combination with gemcitabine, capecitabine, and RT in patients with completely resected carcinoma of the pancreas (see the ECOG homepage). Activity is of interest for a possible combined regimen with RT as the toxic-ity profiles are favourable. A phase I trial com¬bining capecitabine and bevacizumab with RT in locally advanced pancreatic cancer had shown promising results, although there was ulceration and bleeding in the RT field (Crane et al. 2006).
In Europe, the EORTC is recruiting patients in the protocol 22012/40013 (Fig. 9.2; see the EORTC homepage). This is a multi-institutional trial of the gastrointestinal tract cancer group and the RT group, with the Federation Francophone de la Cancerologie Digestive (FFCD) cooperating. There is a phase II feasibility part with a planned 80 patients, followed by a phase III part with a planned 540 patients. Endpoint will be an improvement in disease-free survival by 10%. A strict patient selection will take place, as only patients with pancreatic head carcinoma after R0 pancreaticoduodenectomy will be included. Tak-ing into account the ESPAC-1 data, an initially planned surveillance arm was dropped. After a first amendment, the standard arm now offers gemcitabine 1,000 mg/m2 for four cycles. The ex-perimental arm offers gemcitabine 1,000 mg/m2 for two cycles followed by chemoradiotherapy with gemcitabine 300 mg/m2 once weekly with concurrent 50.4 Gy RT. Eligibility criteria include R0 pancreaticoduodenectomy for pancreatic head cancer with a complete recovery within 8 weeks. So far the accrual has reached 80 patients.
In 2006, during the German Cancer Congress, the first data from the Heidelberg Phase III Trial CapRI (post-operative cisplatin, interferon alpha-2b, and 5-FU combined with external radiation treatment versus 5-FU alone for patients with re¬sected pancreatic adenocarcinoma) was reported. In all, 52 patients were enrolled. The treatment scheme was less toxic than expected; patients could be treated on an outpatient basis. The main common toxicity criteria (CTC) grade III toxici-ties are leukopenia, hand-foot-syndrome, stoma¬titis, fatigue syndrome and hypo-calcemia. The treatment scheme was deemed to be feasible, but an experienced interdisciplinary group is needed (Knaebel et al. 2005).
A phase II study of the CAO/ARO/AIO evaluating a pre-operative radiochemotherapy for potentially resectable patients with cancer of the pancreatic head has started. Gemcitabine 300 mg/m2 weekly is combined with cisplatin 30 mg/m2 weekly and concurrent RT to 50.4¬55.8 Gy followed by surgery versus surgery alone (see the AIO homepage, www.aio-portal.de).
9.9     Summary and Conclusion
As surgery has improved, the outcome has re-mained predictable by factors such as tumour size, resection margin status, N-stage, grading and blood loss at surgery.
In the United States, the adjuvant standard of care is a combined chemoradiation with 5-FU based on the GITSG study results from 1985 with a very limited number of patients. Results have been confirmed by several authors and the randomized EORTC 40891 study, which was, however, underpowered.
In Europe, most patients receive adjuvant che-motherapy only, with 5-FU or with gemcitabine based on the ESPAC-1 data, which have been criticized for considerable inherent limitations concerning statistical power and quality assur-ance. Gemcitabine is currently evaluated in com-bined modality treatment with other partners, varying doses and administration forms, and es-pecially with concurrent RT. Still, in the United States and Europe there is a strong belief in adju-vant therapy based on a survival advantage in the major randomized studies.
Many trials investigating adjuvant therapy in pancreatic cancer are outdated or statistically questionable. This is why the recent phase III studies, ESPAC-3 and EORTC 22012/40013, de-serve support.
When planning adjuvant therapy, some pre-cautions have to be taken into consideration.
-    Patients should be selected carefully.
-    Chemoradiotherapy may be more effective for patients with cancer of the pancreatic head than pancreatic body and tail and for patients with positive resection margins.
-    Treatment planning should be state-of-the-art to minimize treatment-related toxicity.
-    RT planning should be three-dimensional, and should aim for 50-Gy doses without split.
-    IMRT or IORT should be evaluated.
The concurrent chemotherapy can be 5-FU, given as continuous infusion with 200-250 mg/m2 per day, 7 days per week during the entire treatment cycle. Possible partners are interferon-a and cis-platin, but toxicity has to be monitored carefully. Finally, gemcitabine to a dose of 300 mg/m2 once weekly concurrent to RT is a safe and possibly more effective treatment, which is being evalu¬ated in a randomized multi-centre trial of the
EORTC.