Gastrointestinal Imaging Original Research Samadi et al. Feline Esophagus and Gastroesophageal Reflux Gastrointestinal Imaging Original Research Faraz Samadi 1 Marc S. Levine 1 Stephen E. Rubesin 1 David A. Katzka 2 Igor Laufer 1 Samadi F, Levine MS, Rubesin SE, Katzka DA, Laufer I Keywords: barium study, esophagus, fluoroscopy, gastroesophageal reflux, gastrointestinal radiology DOI:10.2214/AJR.09.3352 Received July 21, 2009; accepted after revision September 30, 2009. M. S. Levine and S. E. Rubesin are consultants for Bracco Diagnostics. 1 Department of Radiology, Hospital of the University of Pennsylvania, 3400 Spruce St., Philadelphia, PA 19104. Address correspondence to M. S. Levine (marc.levine@ uphs.upenn.edu). 2 Department of Medicine, Hospital of the University of Pennsylvania, Philadelphia, PA. AJR 2010; 194:972 976 0361 803X/10/1944 972 American Roentgen Ray Society Feline Esophagus and Gastroesophageal Reflux OBJECTIVE. The purposes of this study were to reassess the relation between a feline esophagus (transient transverse esophageal folds) and gastroesophageal reflux (GER) and to determine whether a feline esophagus is observed more often during swallowing or during reflux of barium. MATERIALS AND METHODS. A computerized search of double-contrast esophagrams was performed to generate four equal groups of 56 patients with marked, moderate, mild, and no GER. The imaging findings were reviewed to determine the frequency of a feline esophagus in these groups and whether this sign was detected during swallowing or reflux of barium. The presence of a feline esophagus also was correlated with the presence of a hiatal hernia, reflux esophagitis, a peptic stricture, and esophageal dysmotility. RESULTS. A feline esophagus was detected in 20 of 224 patients (9%). It was detected during reflux of barium in 17 patients (85%), swallowing of barium in two patients (10%), and both in one patient (5%). GER was present in all 20 patients with a feline esophagus and in 148 of the 204 patients (73%) without a feline esophagus (p = 0.0068). A significant relation also was found between a feline esophagus and the presence of a hiatal hernia (p = 0.0116) but not between a feline esophagus and the presence of reflux esophagitis, a peptic stricture, or esophageal dysmotility. CONCLUSION. All patients with a feline esophagus at barium esophagography had associated GER. These transverse folds were observed mainly during reflux of barium from the stomach rather than during swallowing of barium. When a feline esophagus is detected during barium studies, the patient is extremely likely to have GER whether or not GER is seen T ransverse esophageal folds were originally described in 1970 by Bremner et al. [1] as a normal anatomic feature of the cat esophagus and by Goldberg et al. [2] as a normal finding on esophagrams obtained by insufflation of tantalum powder into the cat esophagus. Since that time, similar folds have been recognized on double-contrast esophagrams of humans. The folds appear as thin, closely spaced horizontal striations extending across the circumference of the esophagus without interruption [3]. Because of the close resemblance to the transverse folds in cats, this appearance in humans has been termed a feline esophagus [4]. When they occur in humans, the transverse folds are usually observed as a transient finding at double-contrast esophagography [3]. Because of the fleeting nature of the folds, the term esophageal shiver also has been used to describe the phenomenon [4]. In a study by Williams et al. [5], gastroesophageal reflux (GER) was found in 23 of 44 patients (52%) with transverse folds on barium esophagrams but in only 17 of 211 randomly imaged patients (8%), so the frequency of GER was significantly higher in patients with a feline esophagus (p < 0.001). Those authors therefore recommended that recognition of these transverse folds on barium studies prompt careful evaluation for GER and reflux-associated complications in patients without symptoms [5]. Our anecdotal experience has been that there is an even stronger association between a feline esophagus and GER and that these transverse folds are more likely to be found when barium refluxes into the esophagus than when it is swallowed. The purposes of our study therefore were to reassess the relation between a feline esophagus and GER at 972 AJR:194, April 2010
Feline Esophagus and Gastroesophageal Reflux double-contrast esophagography and to determine whether these transverse folds are more likely to be observed during swallowing or during reflux of barium. Materials and Methods Patient Population A computerized search of the radiology database at our university hospital was performed to identify the cases of all patients who underwent double-contrast esophagography during a recent 2-year period. The original radiographic reports were reviewed retrospectively by one author to classify each case as no GER, mild GER, moderate GER, or marked GER. A finding of no GER required a report in which it was specifically stated that no GER had been found during the fluoroscopic examination. Mild GER required a report in which it was stated that GER was mild or minimal. Moderate GER required a report in which it was stated that there was reflux to the level of the midthoracic esophagus or aortic arch. Marked GER required a report in which it was stated that there was reflux to or above the thoracic inlet. Patients were excluded from analysis if the presence or degree of GER was unclear or if the most proximal level of GER was not mentioned in the reports. Patients also were excluded for one or more of the following reasons: the barium study was incomplete or suboptimal; the barium study was performed by residents or fellows rather than attending gastrointestinal radiologists; and esophageal or gastric surgery had been performed. Patients who fulfilled the study criteria were placed into one of four groups (no GER, mild GER, moderate GER, and marked GER) until each group contained 56 patients. The study sample therefore was arbitrarily designed to consist of 224 patients assigned to four equal groups that included 56 patients with no GER, 56 with mild GER, 56 with moderate GER, and 56 with marked GER. This study design resulted in an artificially high prevalence (75%) of GER in the patient sample. However, it enabled us to compare the frequency of a feline esophagus in patients with no GER with the frequency in patients with mild, moderate, or marked GER. Examination Technique All 224 patients in the study group underwent double-contrast esophagography performed as a biphasic examination. The studies routinely included upright left posterior oblique double-contrast views of the esophagus obtained with an effervescent agent (Baros, Lafayette Pharmaceuticals) and a 250% weight/volume high-density barium suspension (E-Z-HD, E-Z-EM) and prone right anterior oblique single-contrast views obtained with a 50% weight/volume low-density barium suspension (Entrobar, Lafayette Pharmaceuticals). Esophageal motility was evaluated by having the patient take multiple (usually three to five) separate swallows of low-density barium while in the prone right anterior oblique position. Esophageal motility was considered abnormal when two or more of five separate swallows showed no normal progression of primary esophageal peristalsis through the thoracic esophagus to the gastroesophageal junction. At the end of the study, the patient was routinely rotated to the supine position and then to the right lateral position for assessment for spontaneous GER. If necessary, provocative techniques were used to elicit GER, including straight-leg raising or a Valsalva maneuver to increase intraabdominal pressure. When reflux was observed at fluoroscopy, at least one radiograph of the esophagus during the act of reflux was obtained for documentation. All of the studies were performed with digital fluoroscopic equipment (Diagnost 76, Philips Healthcare; Sireskop SD, Siemens Healthcare). All of the studies were performed and interpreted by one of two experienced attending gastrointestinal radiologists. Review of Images The 224 double-contrast esophagrams were reviewed randomly at a PACS workstation by a consensus of two attending gastrointestinal radiologists who had no knowledge of the clinical findings or the findings in the radiographic reports. All of the images were reviewed for the presence or absence of transverse folds in the esophagus (i.e., a feline esophagus) and whether the folds, if present, were detected during swallowing or during reflux of barium or both. This determination was possible because all esophagrams were obtained in a standardized manner, and the spot images were stored in the PACS in the same chronologic order in which they were obtained. Initial upright left posterior oblique double-contrast spot images obtained during swallowing of high-density barium were followed by prone right anterior oblique spot images obtained during swallowing of low-density barium and testing for GER with the patient in the supine, right posterior oblique, and right lateral positions (one or more spot images were obtained in these positions to document GER when it occurred). In this way, it was possible to determine whether a feline esophagus was present during swallowing or reflux of barium or both. When a feline esophagus was present, the images were reviewed to assess the following parameters of the folds: duration (transient versus persistent), distribution (upper, middle, or distal third of the thoracic esophagus), expanse (25%, 50%, 75%, or 100% of the esophageal circumference), and thickness or width of the folds (in millimeters). The images also were reviewed for the presence or absence of a hiatal hernia, reflux esophagitis, and peptic strictures, and the original radiographic reports were reviewed for the presence or absence of esophageal dysmotility. Study Design The imaging findings and radiographic reports were used to determine the frequency of a feline esophagus in each of the four equal study groups and whether transverse folds were detected as a persistent or transient finding during swallowing or reflux of barium, or both. The presence of transverse folds also was correlated with the presence and degree of GER; the presence and size of hiatal hernia (defined as small if it involved less than 10% of the stomach, moderate if it involved 10 25% of the stomach, large if it involved 26 50% of the stomach, and giant if it involved more than 50% of the stomach); and the presence or absence of reflux esophagitis, a peptic stricture, and esophageal dysmotility. The statistical significance of differences between the study groups for these various parameters was determined with a chi-square test (Excel, Microsoft). Our institutional review board approved all aspects of this retrospective study and did not require informed consent from any of the patients whose radiographic images or medical records were included in the study. This investigation was compliant with HIPAA. Results Feline Esophagus A feline esophagus was detected on doublecontrast barium esophagrams of 20 of the 224 patients (9%) who underwent this examination (Figs. 1 4). The transverse folds had a thickness of 1 2 mm in all patients and traversed the entire circumference of the esophagus in 19 patients (95%) and 25% of the circumference in one patient (5%). The transverse folds were observed in the middle and distal thirds of the thoracic esophagus in 19 patients (95%) (Figs. 1 3) and only in the middle third in one patient (5%) (Fig. 4). The folds were observed as a transient finding at fluoroscopy in all 20 patients. The folds were detected only during reflux of barium in 17 patients (85%) (Figs. 1 3), only during swallowing of barium in two patients (10%) (Fig. 4), and during both swallowing and reflux of barium in one patient (5%). Relation Between a Feline Esophagus and GER GER was detected in all 20 patients with a feline esophagus and in 148 of the 204 patients (73%) without a feline esophagus (p = 0.0068). When the 20 patients with a feline esophagus and GER were stratified on the basis of degree AJR:194, April 2010 973
Samadi et al. Fig. 1 76-year-old woman with a feline esophagus. Supine steep right posterior oblique view from double-contrast esophagram shows moderate gastroesophageal reflux with transient transverse folds (arrows) in middle and distal thirds of esophagus that were seen only as barium refluxed from stomach. Large hiatal hernia also is present. of reflux, 10 had marked GER (50%), seven had moderate GER (35%), and three had mild GER (15%). Conversely, a feline esophagus was detected in only 20 of 168 patients (12%) with GER. Relation Between a Feline Esophagus and Other Esophageal Disorders A hiatal hernia was detected in 16 of the 20 patients (80%) with a feline esophagus and in 103 of the 204 (50%) without a feline esophagus (p = 0.0116). When the hiatal hernias were stratified by size, 11 of the 20 patients (55%) with a feline esophagus had small hiatal hernias, three (15%) had moderate hiatal hernia, and two (10%) had large hiatal hernia. Reflux esophagitis was detected in 10 of the 20 patients (50%) with a feline esophagus (Fig. 4) and in 84 of the 204 (41%) without a feline esophagus (p = 0.4454). A peptic stricture Fig. 2 84-year-old woman with a feline esophagus. Supine right posterior oblique view from double-contrast esophagram shows moderate gastroesophageal reflux with transient transverse folds (arrows) in middle and distal thirds of esophagus that were seen only as barium refluxed from stomach. Small hiatal hernia also is present. was detected in two of the 20 patients (10%) with a feline esophagus and eight of the 204 (4%) without a feline esophagus (p = 0.2091). Esophageal dysmotility was detected in three of the 20 patients (15%) with a feline esophagus and in 31 of the 204 (15%) without a feline esophagus (p = 0.9814). Summary of Results A significant association was observed between the presence of a feline esophagus and GER and between the presence of a feline esophagus and a hiatal hernia but not between the presence of a feline esophagus and reflux esophagitis, a peptic stricture, or esophageal dysmotility. Discussion The feline esophagus has a distinctive appearance at double-contrast esophagography, Fig. 3 42-year-old woman with a feline esophagus. Supine view from double-contrast esophagram shows marked gastroesophageal reflux with transient transverse folds (arrows) in middle and distal thirds of esophagus that were seen only as barium refluxed from stomach. manifested by multiple thin closely spaced transverse folds in the esophagus [3]. In our study, these transverse folds were almost always observed as 1- to 2-mm-wide striations crossing the entire circumference of the middle and distal thirds of the thoracic esophagus as a transient finding at barium esophagography (Figs. 1 3). In a previous study, Williams et al. [5] detected a feline esophagus at barium examinations of 19 of 400 patients (4.8%), whereas in our study, a feline esophagus was detected in 20 of 224 patients (9%). However, the prevalence of a feline esophagus in our study was artificially elevated because of our study design and the high frequency of GER in our patient population. In the work by Williams et al. [5], GER was detected on barium studies of 23 of 44 patients (52%) with a feline esophagus, suggesting an association with reflux disease. In our 974 AJR:194, April 2010
Feline Esophagus and Gastroesophageal Reflux Fig. 4 71-year-old woman with a feline esophagus. Upright left posterior oblique view from doublecontrast esophagram shows transient transverse folds (arrows) in middle third of esophagus that were seen only during swallowing of barium. There also is slight granularity of mucosa from mild reflux esophagitis. investigation, however, all 20 patients with the finding of a feline esophagus on double-contrast esophagrams had GER at fluoroscopy, so our experience indicates that this association is extremely strong and that GER occurs in almost all patients with a feline esophagus on barium studies. There also was a relation between the degree of reflux at fluo roscopy and the detection of a feline esophagus. Patients with marked GER were at least three times as likely as those with mild GER to have a feline esophagus. In 17 of our 20 patients (85%) with a feline esophagus on barium studies, this finding was observed only during reflux of barium from the stomach into the esophagus (Figs. 1 3) and not during swallowing of barium by the patient. These transverse folds are thought [4] to be associated with contraction of the longitudinally oriented muscularis mucosae in the wall of the esophagus. We therefore hypothesize that Fig. 5 59-year-old man with eosinophilic esophagitis and ringed esophagus. Upright left posterior oblique view from double-contrast esophagram shows concentric transverse rings (representative rings [arrows]) associated with mild narrowing in lower third of esophagus. Rings are farther apart than transverse folds of a feline esophagus and are associated with stricture. the muscularis mucosae is more likely to contract during reflux than during swallowing of barium. Whatever the explanation, detection of a feline esophagus on barium studies should strongly suggest that one or more episodes of GER have occurred, even if no GER is observed Not infrequently, residual barium in the esophagus is observed at fluoroscopy during a barium study, and the fluoroscopist is uncertain whether this finding is secondary to retained barium in the esophagus or to unobserved reflux of barium from the stomach. On the basis of our findings, however, detection of a feline esophagus on doublecontrast esophagrams is indirect evidence that GER has occurred because all patients with this finding were found to have GER. When a feline esophagus is detected on barium studies, the patient therefore is extremely likely to have GER whether or not GER is seen It is well recognized that the barium study is an insensitive technique in the detection of GER [6, 7]. Furthermore, a feline esophagus was detected in only 20 of 168 patients (12%) with GER. Conversely, all 20 patients with a feline esophagus had fluoroscopic evidence of GER. A feline esophagus therefore should be recognized as a relatively specific but insensitive sign of GER on barium studies. Because a feline esophagus was not observed on barium studies of patients without GER, we believe that there is a causal relation between GER and the development of a feline esophagus. On the basis of anatomic findings on resected esophageal specimens, it has been postulated that thickening and contraction of the longitudinally oriented muscularis mucosae cause shortening of the esophagus and undulation of the overlying squamous epithelium, leading to the development of transient transverse folds [4]. We also observed a significant association between a feline esophagus and a hiatal hernia but not between a feline esophagus and the development of reflux esophagitis or peptic strictures, perhaps because esophageal injury (i.e., esophagitis and stricture formation) is a multifactorial process related not only to GER per se, but also to esophageal clearance of refluxed acid, the physiologic composition and acidity of the refluxate, and the intrinsic resistance of the esophageal mucosa [8, 9]. Furthermore, we observed a higher frequency of reflux esophagitis and peptic strictures in patients with a feline esophagus than in those without a feline esophagus, and it is possible that our sample sizes lacked sufficient power to show statistically significant differences in these findings. At esophagography, the differential diagnosis of a feline esophagus includes the ringed esophagus of idiopathic eosinophilic esophagitis and fixed transverse folds due to scarring from reflux esophagitis. Idiopathic eosinophilic esophagitis sometime manifests distinctive ringlike indentations, resulting in a so-called ringed esophagus [10]. These ringlike indentations are characterized by multiple concentric rings, almost always associated with esophageal strictures (Fig. 5) [10], whereas the delicate transverse folds of a feline esophagus are more closely spaced and are not associated with strictures. Scarring from reflux esophagitis also may produce fixed transverse folds in the distal esophagus [11]. However, these transverse folds are wider (2 5 mm) and fewer than the transverse folds of a feline esophagus; they do not extend com- AJR:194, April 2010 975
Samadi et al. pletely across the esophagus; they often are observed as a persistent finding at fluoroscopy; and they typically are associated with peptic strictures in the distal esophagus [11]. Thus, it usually is possible to differentiate a feline esophagus from the ringed esophagus of idiopathic eosinophilic esophagitis and fixed transverse folds due to scarring from reflux esophagitis on the basis of the radiographic findings. Our investigation had the inherent limitations of a retrospective study, including selection bias and interpretation bias, because the attending radiologists who reviewed the images might have searched more carefully for a feline esophagus when GER was detected. For this reason, the images were interpreted by consensus of two reviewers, and there were no disagreements regarding the presence or absence of a feline esophagus. It should also be recognized that we arbitrarily created four equal groups of patients with no GER, mild GER, moderate GER, and marked GER, artificially elevating the prevalence of GER in our study sample. Because of the limited availability of clinical data in the computerized information system at our hospital, our study also was limited by a lack of clinical correlation for assessing the prevalence of reflux symptoms in the study groups. Finally, GER often is believed to be caused by transient relaxations of the lower esophageal sphincter rather than a sustained decrease in sphincter tone [12, 13]. As a result, some cases of GER in patients without a feline esophagus might have been missed during the brief period of fluoroscopic observation. We conclude that all patients with a feline esophagus at barium esophagography have associated GER and that these transverse folds are mainly observed during reflux of barium from the stomach rather than during swallowing of barium. When a feline esophagus is detected on barium studies, the patient therefore is extremely likely to have GER whether or not GER is seen References 1. Bremner CG, Shorter RG, Ellis FH. Anatomy of feline esophagus with special reference to its muscular wall and phrenoesophageal membrane. J Surg Res 1970; 10:327 331 2. Goldberg HI, Dodds WJ, Jenis EH. Experimental esophagitis: roentgenographic findings after insufflation of tantalum powder. AJR 1970; 110:288 294 3. Gohel VK, Edell SL, Laufer I, Rhodes WH. Transverse folds in the human esophagus. Radiology 1978; 128:303 308 4. Furth EE, Rubesin SE, Rose D. Feline esophagus. AJR 1995; 164:900 5. Williams SM, Harned RK, Kaplan P, Consigny PM. Work in progress: transverse striations of the esophagus association with gastroesophageal reflux. Radiology 1983; 146:25 27 6. Thompson JK, Koehler RE, Richter JE. Detection of gastroesophageal reflux: value of barium studies compared with 24-hr ph monitoring. AJR 1994; 162:621 626 7. Ott DJ. Gastroesophageal reflux: what is the role of barium studies? AJR 1994; 162:627 629 8. Dodds WJ. The pathogenesis of gastroesophageal reflux disease. AJR 1988; 151:49 56 9. Barlow WJ, Orlando RC. The pathogenesis of heartburn in nonerosive reflux disease: a unifying hypothesis. Gastroenterology 2005; 128:771 778 10. Zimmerman SL, Levine MS, Rubesin SE, et al. Idiopathic eosinophilic esophagitis in adults: the ringed esophagus. Radiology 2005; 236:159 165 11. Levine MS, Goldstein HM. Fixed transverse folds in the esophagus: a sign of reflux esophagitis. AJR 1984; 143:275 278 12. Schoeman MN, Tippett MD, Akkermans LM, et al. Mechanisms of gastroesophageal reflux in ambulant healthy human subjects. Gastroenterology 1995; 108:83 91 13. Mittal RK, Holloway RH, Penagini R, et al. Transient lower esophageal sphincter relaxation. Gastroenterology 1995; 109:601 610 976 AJR:194, April 2010