Foreign body (FB) ingestion is a common clinical problem. Most ingested FBs pass through the gastrointestinal (GI) tract uneventfully within 1 week [1], and GI perforation is rare, occurring in less than 1% of patients [2, 3]. Fish bones are the most commonly ingested objects and the most common cause of FB perforation of the GI tract. FB perforation of the GI tract has diverse clinical manifestations, and the correct preoperative diagnosis is seldom made. Nonmetallic FBs, such as fish bones, are rarely detected on radiographs [4]. CT may be useful in the correct preoperative diagnosis of FB perforation [5-9]. We report our experience with CT in the preoperative diagnosis of fish bone perforation of the GI tract. To our knowledge, this series is the largest to date addressing the role of CT in the diagnosis of fish bone perforation of the GI tract.

The medical records of all patients admitted to the department of surgery, Singapore General Hospital, between January 1996 and December 2003 were retrospectively reviewed for surgical cases of FB perforation of the GI tract. Informed consent was waived by the institutional review board. Only patients with intraabdominal perforation were included. Perforations proximal to the stomach and distal to the rectum were excluded. The records of 22 patients with FB perforation were found, and 15 of the patients had perforation caused by fish bones. Seven of the patients with fish bone perforation underwent preoperative CT and are the focus of this study. All seven patients underwent preoperative erect chest radiography, and four underwent abdominal radiography. The patients underwent helical CT with oral and IV contrast media performed with either a four-slice MX8000 helical scanner (Philips Medical Systems) with slices 7.5 mm thick or an Xpress/SX helical scanner (Toshiba) with slices 7 mm thick. Patient 3 underwent repeat unenhanced CT for confirmation of the presence of a fish bone. Patients 2 and 6 have been described in previous case reports [9, 10]. The images of six patients were reevaluated retrospectively by a senior radiologist. The films of patient 1 were not available for review.

Demographic data, clinical features, and radiologic findings are summarized in Table 1. The seven patients had a median age of 46 years (range, 32-72 years), and four of the patients were women. All of the patients had symptoms with a median duration of 7 days (range, 2-14 days). The patients presented with intraabdominal abscess (n = 5) or localized peritonitis (n = 2). All the patients were unaware of having ingested an FB preoperatively, and three of the seven patients wore dentures. None of the patients had biochemical or radiologic evidence of pancreatitis. All patients underwent laparotomy for management of perforation. The sites of perforation included the stomach (n = 3), duodenum (n = 2), jejunum (n = 1), and transverse colon (n = 1). Two patients had postoperative complications including wound infection and intraabdominal abscess.

According to the original CT reports, a correct preoperative diagnosis of FB perforation was made in five of seven cases (patients 1-5) (Figs. 1 and 2). Patient 3 was thought to have FB perforation, but the diagnosis was not made definitively with the initial contrast-enhanced CT scans (the images were not available for review). Unenhanced CT was repeated, and the findings confirmed those on the initial scans (Fig. 3). When the CT scans of patients 6 and 7 were reviewed retrospectively by a senior radiologist, the offending fish bone was successfully identified in both patients (Figs. 4 and 5). In patient 6, a linear density traversing a pancreatic mass seen during the original evaluation was thought to be a blood vessel, and the patient underwent radical surgery for presumed locally advanced pancreatic cystadenocarcinoma. The final histologic examination of the surgical specimen showed an abscess of the pancreas secondary to fish bone perforation with no evidence of malignancy. This case has been reported and discussed in detail previously [10]. In all cases, the fish bone was visualized on CT as a linear or circumlinear calcified lesion in the abdominal cavity with adjacent areas of inflammation or abscess formation. Pneumoperitoneum was not detected on any of the CT scans. None of the erect chest radiographs showed pneumoperitoneum, and none of the abdominal radiographs (obtained for four patients) showed the presence of an FB even when reviewed retrospectively.

FB perforation occurs in all segments of the GI tract, although it tends to occur in regions of acute angulation, such as the ileocecal and rectosigmoid junctions [2, 11]. FBs may also perforate through a hernia sac, Meckel's diverticulum, or the appendix [12]. FB perforation of the GI tract has a wide spectrum of clinical presentations, which can be acute or chronic. Patients occasionally present with unusual or even bizarre clinical manifestations, including hemorrhage, bowel obstruction, and even ureteric colic [2, 12]. With these varied and nonspecific clinical presentations, it is not surprising that FB perforation is seldom diagnosed preoperatively [13].

Numerous risk factors for ingested FB perforation of the GI tract have been identified in the literature. Prisoners, psychiatric patients, practitioners of selected professions (e.g., carpenters and dressmakers), alcohol and drug abusers, persons who eat rapidly, and persons in the extremes of life are more susceptible to FB ingestion and its complications than are others [5, 11]. The wearing of dentures is a well-described risk factor for FB ingestion because it eliminates tactile sensation on the palatal surface. This palatal sensory feedback provides a protective mechanism for identifying small, sharp, and hard-textured items in a food bolus [2, 14]. Dental factors have been reported in as many as 80% of cases of FB ingestion [14]. In our series, three of the seven patients wore dentures. In essence, persons at risk have reduced time, awareness, and capability of forming a food bolus during the mastication phase of swallowing.

Nonmetallic FBs, especially fish bones and other bone fragments, pose a unique problem in the diagnosis of FB perforation. The number of occasions on which these objects are swallowed are numerous and underreported [12]. Fish bone ingestion is especially frequent in cultures (e.g., Chinese) in which unfilleted fish is a culinary delicacy; the bones are frequently ingested accidentally and forgotten [15]. This problem is compounded because there may be a time lag of months or even years between ingestion and the onset of symptoms [15]. In contrast, accidental ingestion of nondietary FBs is a more dramatic event and impresses itself vividly on the patient's memory [12]. Objects are also intentionally ingested by children, prisoners, and the mentally infirm [11]. The inability to obtain a history of FB ingestion and its wide spectrum of nonspecific clinical presentations makes dietary FB perforation extremely difficult to diagnose. In this study, none of the patients remembered a recent incident in which they accidentally ingested a fish bone.

Radiography is unreliable in the diagnosis of fish bone perforation [5, 10]. This problem has been illustrated in studies of fish bone ingestion showing that the degree of radiopacity of the bone depends on the species of fish [16, 17]. In contrast, chicken bones are almost always radiopaque. Even when fish bones are sufficiently radiopaque to be visualized on radiographs, large soft-tissue masses and fluid can obscure the minimal calcium content of the bone, particularly in altered or obese patients [5, 18]. Another reason for not identifying fish bones on radiographs is use of the peak kilovoltage setting. Subtle calcifications are more easily identified on low-kilovoltage (70 kV) supine films. In contrast, use of 90 kV makes it more difficult to see the offending FB. Results of a prospective study with 358 patients who had swallowed fish bones revealed that radiography had a sensitivity of only 32% [4]. Another difficulty is that the presence of free gas under the diaphragm is almost never seen in FB perforation of the GI tract [5]. Because the perforation is caused by impaction and progressive erosion of the FB through the intestinal wall, the site of perforation becomes covered by fibrin, omentum, or adjacent loops of bowel. This limits the passage of large amounts of intraluminal air into the peritoneal cavity [11]. Our study showed that free intraperitoneal air is a poor radiologic sign. None of the patients had fish bones visualized on radiographs even at retrospective review of the images.

CT has been helpful in the detection of nonmetallic FB perforation [7, 19]. However, descriptions of the use of CT for FB perforation have been limited to case reports [6-9, 19] and a single case series [5]. Coulier et al. [5] reported the use of CT in the diagnosis of seven cases of nonmetallic FB perforation, including three cases of fish bone perforation. The region of perforation can be identified on CT scans as a thickened intestinal segment, localized pneumoperitoneum, regional fatty infiltration, or associated intestinal obstruction. However, none of these findings are specific, and the definitive diagnosis is made by identification of the calcified FB. Fish bone perforation typically appears on CT scans as a linear calcified lesion surrounded by an area of inflammation, as shown in our study.

Despite its superiority over radiography in the diagnosis of fish bone perforation, CT has potential limitations in the detection of intraabdominal fish bones. In our study, the sensitivity of CT in the detection of intraabdominal fish bones was 71.4% (5/7) for initial reports but improved to 100% (7/7) on retrospective review of CT scans. The main limitation of CT in the detection of FBs in this study was lack of observer awareness. This study showed that without a high index of suspicion, an FB can be missed (patient 7) or mistaken for another structure, such as a blood vessel (patient 6) [10]. Another potential limitation of CT is scanning thickness. Use of thinner CT slices allows reviewers to better trace structures such as blood vessels and differentiate them from calcified FBs. It also limits the chance that an FB is present between slices when movement artifacts are present, especially with use of older scanners with slow scanning times. The CT scans used in our study were obtained with standard 7- to 7.5-mm slices. This thickness was reliable in the detection of fish bones in our cohort of patients. Coulier et al. [5] emphasized the importance of the thickness of CT slices in the detection of FBs. In their series, FBs were identified preoperatively with CT in all seven patients. In that study, single-detector helical CT with 3-mm or 1.5-mm slices and MDCT with 1.25-mm or 0.65-mm slices were used and the images examined with multiplanar reconstructions and cine mode on workstations. Despite the superiority of such technology, this strategy may be overkill. It is not practical for most institutions to use such fine-cut CT scans with 3D reconstruction to examine all patients presenting with an acute abdomen. Nonetheless, it would not be unreasonable for institutions with single-detector equipment to rescan the abscess region in thinner sections to identify a subtle FB. The orientation of an FB with respect to an axial CT scan also can affect the perception of the viewer. Coronal reconstruction would be especially useful in overcoming this limitation.

The use of oral and IV contrast material during CT can cause difficulty in identifying fish bones. Oral contrast media can obscure fish bones in the intestinal lumen, as in patient 7, causing them to be missed. This problem can be ameliorated with the use of 16-MDCT, in which only water is used to distend the stomach and bowel loops. Extraluminal fish bones, as in patient 6, can be mistaken for blood vessels if IV contrast media are used. In general, however, fish bones appear more attenuated and can be appreciated with careful windowing of CT images. Unenhanced CT scans should be repeated, as in the case of patient 3, if the diagnosis is strongly suspected but cannot be confirmed with the initial contrast-enhanced CT scan.

In conclusion, this study showed the utility of CT in the detection of fish bone perforation of the GI tract and the superiority of CT over clinical history and radiography. However, the accuracy of CT is limited by lack of observer awareness, and a high index of suspicion must be maintained for the correct diagnosis of fish bone perforation.

Address correspondence to B. K. P. Goh ([email protected]).