The esophagus is a muscular, tubular structure that measures approximately 25 cm in length at extends from the cricoid cartilage, at the level C6 vertebral body level, to the esophagogastric (EG) junction, at the level of the T11 vertebral body level. It is located posterior to the trachea and bronchi in the posterior mediastinum. The esophagus is anterior to the spinal cord and the separation between the esophagus and spinal cord increases toward its gastric interface. The presence of an extensive network of submucosal lymphatics permits the spread of esophageal cancer several centimeters beyond the gross tumor, which is a relatively common event (Czito et al. 2008; Bradley and Mutic 2006). Sakata first demonstrated in 1903 that the submucosal lymphatics drain longitudinally, rather than in a segmental fashion (Sakata 1903). The network of lymphatics within the esophagus results in erratic spread of lymphatic metastases with frequent skip metastases to lymph nodes (van de Ven et al. 1999). Immunohistochemical analyses of surgical specimens for resected esophageal cancers have revealed a 66 % rate of skip metastases (Hosch et al. 2001). It is due to this potential for longitudinal spread that the length of longitudinal surgical margin is a predictor of outcomes after resection of esophageal cancer and that a longitudinal resection margin of 5 cm is recommended (Barbour et al. 2007) (Fig. 2).

The esophagus is generally divided into the cervical, upper thoracic, mid-thoracic, and lower thoracic regions. In the American Joint Committee on Cancer (AJCC) Staging Manual, these regions are defined as: cervical (extending from the inferior border of cricoid to thoracic inlet, at approximately18 cm from upper incisors on endoscopy), upper thoracic (extending from the thoracic inlet to level of the carina, approximately 24 cm from upper incisors), mid-thoracic (extending from the level of the carina to just superior to the EG junction, 32 cm from incisors); and lower thoracic/abdominal, the abdominal portion of the esophagus and the EG junction (40 cm from incisors) (American Joint Committee on Cancer 2002). See Fig. 2 for esophageal anatomy and index distances from upper incisors on endoscopy (Czito et al. 2008). For RT planning, the external border of the esophagus may be defined manually on axial computed tomography (CT) images. In order to obtain an accurate and informative dose-volume histogram, the esophagus should be segmented from its origin at the cricopharyngeus muscle to its termination at the gastroesophageal junction. One recent report suggests that a “correction method” for esophageal segmentation on CT images, based upon physiological principles of the normal esophageal circumference, may improve dosimetric predictions of clinical toxicity outcomes after RT for lung cancer (Kahn et al. 2005), emphasizing the relevance of esophageal anatomy to RT planning.

The esophageal wall contains the basic histological architecture characteristic of the gastrointestinal tract: mucosa, submucosa, two muscular layers (inner circular layer and outer longitudinal layer), and adventitia (Fig. 3a) (Rubin and Casarett 1968; Junqueira et al. 1094). It lacks serosa, a deficiency that is thought to increase the opportunity for radial extension of tumor from the esophageal wall into the periesophageal tissues. The clinical relevance, however, of the lack of serosal coverage is unclear; it is not known how much a thin serosa would protect against extramural extension for a tumor that has invaded through the muscular wall. The esophageal mucosa is composed of non-keratinized stratified squamous epithelium (Squier and Kremer 2001). The components of the esophagus may be characterized using Rubin and Casarett’s classification of radiosensitivity, which is based on cellular reproductive and functional characteristics (Rubin and Casarett 1968). According to the Rubin and Casarett system, the inner germinal stratum of the esophageal epithelium contains vegetative (Group I) and differentiating (Group II) intermitotic cells, which are considered radiosensitive. The outer germinal stratum, adjacent to the esophageal lumen, is composed of fixed postmitotic cells (Group IV), which do not multiply and are considered radioresistant (Rubin and Casarett 1968; Hall and Giaccia 2006). Although the mucosal epithelium is avascular, it receives nutrients by diffusion from capillaries contained in the lamina propria, which is also the site of venous and lymphatic drainage (Rubin and Casarett 1968). Small mucous glands, whose functions are to protect mucosa and facilitate food transport, are present within the submucosa throughout the esophagus (“esophageal glands”) and within the lamina propria in the distal esophagus near the stomach (“esophageal cardiac glands”) (Junqueira et al. 1094). The muscular components of the esophageal wall, governed by both reflexive and autonomic nervous system mechanisms, are responsible for the peristalsis and sphincter regulation necessary for food transport. When considered as an organ using the Michalowski classification, the esophageal radiosensitivity is in agreement with hierarchical (H-type) tissues, which are early-responding cell lines (Michalowski and Hornsey 1986; Wheldon et al. 1982). The esophagus is comprised of structurally undefined functional subunits (FSUs), in contrast to the lung, which implies that repopulation after RT may result from the migration of clonogenic cells from one FSU to another (Hall and Giaccia 2006) (Fig. 3a, b).

The principal function of the esophagus is to deliver food from the pharynx to the stomach, and this rapid transfer requires complex coordination to ensure proper timing and anterograde direction. Although swallowing is initiated voluntarily, esophageal motility is largely under automatic control that consists of input from the brainstem and involvement of vagal parasympathetic, efferent nerve fibers and the enteric nervous system. The upper and lower esophageal sphincters must be relaxed at the appropriate time during swallowing, as these structures are closed at rest in order to prevent retrograde movement of digestive contents. Esophageal peristalsis is activated by the stimulatory effects of distention on mechanoreceptors, triggering a vagovagal reflex that is responsive to food bolus volume and temperature (Barrett 2307). The delivery of a food bolus from the oropharynx to the stomach is a complex, coordinated process dependent upon adequate function of multiple autonomic and reflexive processes.

Disruption of esophageal motility is a frequent occurrence after RT. Goldstein et al. used esophagograms to study RT-induced changes. They found motility disorders to be the most common RT-induced problem and reported loss of peristalsis in the irradiation portion of the esophagus, with adjacent areas of rhythmic, but uncoordinated, contraction after an average 50 Gy of RT (Goldstein et al. 1975). Lepke and Libshitz observed similar dysmotility after RT, which they described as developing 4–12 weeks after RT (Lepke and Libshitz 1983). Technetium scintigraphy has also been used to visualize temporarily prolonged esophageal transit times in most patients after RT doses higher than 40 Gy (Lamanna et al. 1985). Impaired esophageal motility may also be observed by manometry (Kaplinsky et al. 1991). The impaired motility may be related to neuronal and/or muscular injury (Coia et al. 1995; Kaplinsky et al. 1991). Due to the tubular geometry of the esophagus, the development of mural thickening or stricture may severely impair the transit of food (Coia et al. 1995; Fajardo et al. 2001). Esophageal dysmotility and stricture is a serious late morbidity for patients after RT.

Although the pathologic response of the esophagus to radiation injury has been characterized (Rubin and Casarett 1968), the molecular events responsible for the late effects of radiotherapy are complex and not resolved fully (Stone et al. 2003; Brush et al. 2007; Denham and Hauer-Jensen 2002). The response of normal tissue to RT involves a sequence of steps intended to promote healing, including inflammation, epithelial proliferation, collagen deposition, and remodeling (Denham and Hauer-Jensen 2002). When compared to wound healing after non-RT tissue damage, the injury response is often dysregulated after RT. Whereas the tissue-damage response functions to promote successful wound healing in response to other insults, this response contributes to chronic damage after RT. It is possible that cytokine-mediated damage in response to RT is due to changes in the microenvironment, the influence of cell death on nearby tissues, or DNA damage (Brush et al. 2007). Pro-inflammatory mediators, such as chemokines and cytokines, are expressed in tissues, including the esophagus after RT (Vujaskovic et al. 2007; Brush et al. 2007). The significance of oxidative stress and integrin αv6β-mediated stimulation of TGF-β is supported by preclinical work with amifostine in a rat model. Vujaskovic et al. showed that amifostine reduced acute and late pathologic changes after RT with associated decreases in oxidative stress and levels of integrinαv6β and TGF-β (Vujaskovic et al. 2007). It has been shown that manganese superoxide dismutase gene therapy can ameliorate acute and late esophageal injury via modulation of RT-induced elevation of inflammatory cytokines (Epperly et al. 2001). These animal model studies of radioprotective agents for esophageal injury emphasize the relevance of pro-inflammatory mechanisms in the development of late effects after irradiation of the esophagus. In summary, the late clinical effects of esophageal irradiation appear to be initiated by oxidative stress, mediated by cytokines, and guided by end-pathway damage to the muscle wall and submucosal thickening (Fig. 4).

In the assessment of RT-induced esophagitis, endoscopy, and imaging studies provide important information. During the acute phase, endoscopic findings of esophagitis include mucosal erythema, erosion or ulceration (Mascarenhas et al. 1989; Hirota et al. 2001). Hirota et al. performed endoscopic examination during or soon after RT for patients with NSCLC, most treated with concurrent chemoradiotherapy, and found a good correlation between endoscopy-measured esophagitis grade and RTOG toxicity score (rank correlation coefficient = 0.428, p < 0.0001) (Hirota et al. 2001). Endoscopic findings of acute esophagitis from Hirota et al., along with the accompanying score, are displayed in Fig. 7a (Hirota et al. 2001). The agreement between endoscopic appearance of the esophagus during the acute phase and the reported RTOG toxicity score is supportive of the scoring system’s validity (Werner-Wasik et al. 2004). In some instances, however, endoscopic findings of esophagitis may not be confirmed by histology when biopsied (Mascarenhas et al. 1989), producing some false-positive results. Although endoscopy may be useful in the management of esophagitis, endoscopic findings are not a primary determinant of toxicity grade in standard practice. The current CTCAE criteria emphasize patient symptoms of esophagitis, but the description of CTCAEv4.0 Grade 1 esophagitis does include asymptomatic findings on endoscopy or radiography.

Although most endoscopic findings of acute esophagitis resolve without the development of chronic effects, some damage that is evident during the acute period may progress to late esophagitis (Hirota et al. 2001). Endoscopy permits the visualization of strictures that occur as a late effect after RT (Fig. 7b) and provides an opportunity for dilatation by bougie or balloon, the standard therapeutic approach for post-RT benign esophageal stricture (Siersema 2008; Raymondi et al. 2008). Many strictures require more than one dilatation, with a reported median number of 2.5 dilatations delivered after a median time of 5 months between procedures (O’Rourke et al. 1988). The likelihood of requiring more than one dilatation for esophageal stricture is higher for complex versus simple strictures, which may be determined by endoscopy. Simple strictures are short, straight, and wide enough to permit passage of a standard-diameter endoscope. Longer (>2 cm), tortuous strictures, through which a standard endoscope may not be advanced, are categorized as complex and are a challenge for endoscopic visualization and dilatation (Siersema 2008; Giever et al. 2008; Lew and Kochman 2002). Endoscopic dilatation is discussed in further detail below in Sect. 8.2 along with other aspects of management of RT-related esophagitis.

Radiological studies may complement endoscopy and clinical evaluation in the assessment of esophagitis after RT. The esophagogram may be used to demonstrate esophageal stricture and dysmotility after RT (Lepke and Libshitz 1983; Goldstein et al. 1975; Ellenhorn et al. 1993). Goldstein et al. observed deficiencies in peristaltic waves on esophagograms in patients who received mediastinal RT (Goldstein et al. 1975). Double-contrast esophagograms may demonstrate the presence of multiple ulcers or granular mucosa as an acute effect of RT, as well as stricture as a late effect after RT, as shown in Fig. 8 (Collazzo et al. 1997). Manometry and technetium transit scintigraphy may also be used to demonstrate abnormal esophageal motility after RT (Kaplinsky et al. 1991; Lamanna et al. 1985). Findings on esophageal manometry after RT that support the diagnosis of esophagitis include low-amplitude, weakened or absent peristaltic contractions (Collazzo et al. 1997). Among 25 patients who received RT, Lamanna et al. observed protracted transit times using technetium transit scintigraphy in nearly all patients after more than 40 Gy, and some patients displayed transit abnormalities at >2 months after RT (Lamanna et al. 1985). During the initial evaluation of esophageal stricture after RT, esophagography may be helpful in defining the location and extent of stricture. Other modalities, such as manometry or scintigraphy, may provide complementary information regarding esophageal motility. In appropriate cases, endoscopy would then be performed, with biopsy or dilatation as appropriate. In summary, radiography can demonstrate RT-induced esophageal injury and complements endoscopy in the evaluation of acute and late esophagitis.

Although threshold doses for RT-induced esophageal injury had been described earlier (Roswit 1974), Emami et al. provided the first dose-volume recommendations for esophagitis, using the endpoint of clinical stricture or perforation at 5 years, based upon the authors’ clinical experience and review of the literature (Emami et al. 1991) (Table 3). Subsequently, the widespread availability of three-dimensional RT planning has provided informative data for the prediction of esophagitis. The length of irradiated esophagus has been reported to impact esophagitis, as suggested by animal studies that demonstrated that increasing the length of irradiated esophagus lowers the mean effective dose for ulcerative esophagitis (Michalowski and Hornsey 1986). However, findings in clinical studies are contradictory and do not confirm the presence of such a relationship between length or irradiation esophagus and esophagitis in NSCLC patients (Ball et al. 1995; Werner-Wasik et al. 2000; Werner-Wasik et al. 2004; Choy et al. 1999; Langer 1999). Dosimetric parameters have been associated with both acute and late esophagitis in published reports, and are reviewed below (see Tables 3, 4 and 5).