The GTV generally comprises the primary tumor and any grossly involved lymph nodes (minimum diameter >1 cm) based on computed tomography (CT), magnetic resonance imaging (MRI), and (optionally) ¹⁸F-fluorodeoxyglucose positron emission tomography (PET) [57]. The clinical target volume (CTV) is the area thought to harbor micrometastasis and is usually covered by a 0.5–1.5-cm margin around the GTV. The need for ENI for unresectable or neoadjuvant/borderline resectable cases is controversial, but ENI is typically used even though its benefits have not been proven in a planned prospective randomized study. Findings from some recent trials suggest that omitting ENI can both be effective and allow dose escalation of radiation or chemotherapeutic agents [45, 58, 59]. However, if the dose constraints for normal tissues permit, the retropancreatic space between the root of the celiac and superior mesenteric arteries could be reasonably included in the CTV, because microinvasion of the neural plexus in this region is common and causes severe back pain that reduces the patients’ quality of life. When IMRT is used for adjuvant therapy, the CTV is the preoperative tumor bed and the region at highest risk for residual subclinical disease. Clinical studies have also involved targeting the following volumes as specific regions of interest (ROIs), according to the guidelines used to construct the 3D fields used in RTOG 9704: the celiac artery (the most proximal 1.0–1.5 cm of that artery from where it splits from the aorta and including up to its first branching), the superior mesenteric artery (the most proximal 2.5–3.0 cm of that artery from where it splits from the aorta), the portal vein, the site of pancreaticojejunostomy (may be omitted if clinically appropriate), and the aorta (from the cephalad contour of the celiac artery, and portal vein or pancreaticojejunostomy site [whichever is most cephalad] to the bottom of the L2 vertebral body). The PTV is the CTV plus a setup margin and the internal target volume (ITV). ITV is generated via management of respiratory motion, as described below .

The delivery of highly conformal radiotherapy is usually problematic because the pancreas moves with respiration. According to fluoroscopic, ultrasonographic, MRI, and four-dimensional (4D) CT studies [60–67], the peak-to-peak magnitude of movement associated with respiratory motion can be as large as 20–30 mm [68]. Using cine MRI, Feng et al. [65] found that the motion of the borders of pancreatic tumors was highly variable among patients and reported that a margin of 20 mm inferiorly was required to provide 99 % geometric coverage. Mori et al. [66] used a 256-multislice 4D CT scanner and verified a displacement of the GTV by >10 mm in the superior-inferior direction.

Traditionally, radiation treatment planning for pancreatic cancer involves adding large margins around the CTV to generate the PTV [67] to account for respiratory motion during free breathing; however, the use of large margins can result in high-grade gastrointestinal toxicity [68], as the PTV often contains large volumes of organs at risk.

Respiratory motion of the pancreas is also a critical factor limiting IMRT. The interplay between motion of the multileaf collimator and motion of the target could lead to degradation of the planned dose distribution [69, 70]. Several investigators have proposed various dose delivery methods to resolve problems related to respiratory motion. The use of breath-hold (BH) techniques can reduce the impact of respiratory motion and has been used successfully in the treatment of cancer at various sites [71–76]. Dawson et al. showed the reproducibility of organ position using active breathing control during liver radiotherapy [74]. Hanley et al. applied a deep inspiration BH technique for the treatment of thoracic cancer [75]. Respiratory gating has been used for the treatment of lung and liver cancer [76–79]. Taniguchi et al. showed in a dosimetric analysis that expiratory gating might be preferable to inspiratory BH and free breathing for avoiding normal tissue toxicity [80]. Dynamic tumor tracking [81–84] has reduced the impact of respiratory motion and interfraction motion during radiation treatment. Gibbs et al. found the CyberKnife system, which includes dynamic tumor tracking, to be one of the most effective techniques for treating pancreatic tumors that move with respiration [85].

At Kyoto University, the use of a visual feedback technique during end-exhalation (EE)-BH conditions led to highly reproducible positioning of pancreatic tumors [67, 86]. IMRT in combination with EE-BH could be a reliable method to facilitate dose escalation in locally advanced pancreatic tumors with small margins. This technique and its application in the treatment of locally advanced pancreatic cancer are described in the following sections: