Comparison of the Results of Abdominal Ultrasonography and Exploratory Laparotomy in the Dog and Cat

Introduction

Abdominal ultrasonography is generally considered to be a useful modality for investigation of a wide range of clinical signs in small animals; however, studies are lacking that correlate the results of ultrasonography with the results of laparotomy. Studies that compare abdominal ultrasonography and laparotomy in humans have concentrated on women with ovarian cancer or patients with abdominal trauma.^1–8 Although ultrasonography is considered useful in many cases and is a sensitive method for detecting certain lesions, it lacks sensitivity for peritoneal lesions (other than fluid) and lesions <2 cm in humans.¹ In a study assessing the accuracy of ultrasonography for the evaluation of penetrating torso trauma, it was found that ultrasonographic results may not contribute to patient management or adequately predict the need for surgical intervention.⁴

Various studies in animals have described ultrasonographic findings in relation to specific conditions or presenting signs.^9–16 For example, use of a focused assessment in a post-trauma protocol was found to be useful for detection of peritoneal hemorrhage in 100 dogs with blunt abdominal trauma.⁹ Peritoneal fluid and masses in the peritoneum were also found in all 14 cats examined with carcinomatosis.¹⁰ In a series of 12 dogs that had ultrasonography of the liver and gallbladder, ultrasonographic diagnoses matched the surgical findings in the majority of cases, although the cause of gallbladder obstruction could not be determined in two of the dogs by ultrasonography alone.¹¹ In a series of 14 dogs with hepatic abscesses, masses were found in the livers of all nine dogs by ultrasonography. ¹² In a similar study in cats with hepatic abscesses, liver lesions were found in five of seven cats by abdominal ultrasonography.¹³ Ultrasonography was found to be insensitive in a series of dogs with serious, acute pancreatitis. Similar limitations may be associated with diagnosing pancreatitis in cats, although a recent prospective study found a sensitivity of 80% and a specificity of 88% for ultrasonography in cats with moderate to severe pancreatitis.^14–16

While these prior studies are helpful for assessing the value of ultrasonography in diagnosing particular abdominal conditions, they do not give an overall impression of how well ultrasonography can be used to guide the surgeon. The aims of the present study were to quantify how well preoperative ultrasonographic diagnoses correlated with surgical findings on exploratory laparotomy in cats and dogs, and to identify abdominal lesions most likely to be missed by ultrasonography.

Materials and Methods

This retrospective study was conducted at the Queen Mother Hospital for Animals. The surgical logbook was searched for dogs and cats that had an exploratory laparotomy performed from January 2003 through December 2004. The medical records for these animals were then searched in a reverse chronological order for reports of a preoperative abdominal ultrasound done within 48 hours of the laparotomy, until 100 cases were found that satisfied this criterion. Information extracted from the medical records of these animals included signalment, ultrasonographic findings and diagnosis, and surgical findings. Animals were separated into groups according to body sizes, which included cats, small dogs (<10 kg), medium dogs (10 to 27 kg), and large-breed dogs (>27 kg). Groups were created for the purpose of determining if sensitivity of ultrasonography differed between groups, and the hypothesis was that large dogs may be more difficult to examine.

Abdominal ultrasonography was performed by one of three experienced radiologists or radiology residents working under their direct supervision. Various MHz probes and a single ultrasound machine were used.^a Names of personnel performing the ultrasound were recorded, but they were not used in the present study. A standard ultrasound report form was used for all cases that required an entry of “normal” or “not seen,” or the report included the description of any abnormality for each abdominal organ. Surgical reports were in the form of prose descriptions of the surgeon’s observations and the procedures performed.

Results of abdominal ultrasonography and the surgical findings were independently classified according to the affected anatomical location of the lesion(s) and the gross pathological features. Surgical findings were classified as primary or secondary on the basis of the surgeon’s recorded observations. The primary lesions were defined as the main surgical finding, while secondary lesions were categorized as any other related surgical findings. For example, in an animal with a ruptured gallbladder and peritonitis, the gallbladder lesion was ranked primary and the peritonitis was a secondary finding. In other instances, the surgeon’s classification did not always correspond to the pathophysiology; for example, in a dog with paraprostatic cyst, the cyst may be considered the primary surgical finding, despite the presence of an underlying prostatic hyperplasia. Histological results were not considered in the classification of lesions as primary or secondary. Any findings on ultrasound or at surgery that appeared by the surgeon to be unrelated to the primary lesion were also recorded.

The sensitivity of ultrasonography for the three most common surgical findings was calculated by dividing the number of animals that had the abnormalities mentioned in the ultrasound report, by the total number of affected animals. Fisher’s exact test was used to compare proportions of animals of different body size, lesions affecting different anatomical locations, and the different gross pathological features. Differences with a P value <0.05 were considered significant.

Results

Nineteen cats and 81 dogs were included in the study. The majority of the dogs were medium-sized breeds, and the type of cat varied [Table 1]. The genders of the dogs included males (n=21), castrated males (n=22), females (n=11), and spayed females (n=27). The cats included males (n=1), castrated males (n=10), and spayed females (n=8). Dogs ranged in age from 2 months to 13 years (mean 7.2 years). Cats ranged in age from 6 months to 14 years and 1 month (mean 5.1 years). The distribution of lesions based on anatomical location and type of gross pathology as seen at surgery are listed in Tables 3 and 4. Body size groupings included 19 cats, 17 small dogs, 47 medium-sized dogs, and 17 large-breed dogs.

One hundred primary lesions and 67 secondary lesions were seen at surgery. The primary surgical finding was not discovered on ultrasonography in 24 animals, and secondary surgical findings were not identified in an additional 11 animals. The three most common primary surgical lesions were peritonitis (n=28), splenic/liver mass(es) (n=30), and intestinal obstruction (n=14). The sensitivities of ultrasonography for detection of these conditions were 89%, 63%, and 64%, respectively. In one dog with a history of vomiting, the ultrasound report described gastric wall thickening and heterogeneous echogenicity of the pancreas and/or peripancreatic fat; however, no corresponding abnormalities were found surgically. Of the 15 animals (12 dogs, three cats) with secondary lesions missed during ultrasonography, four (three dogs, one cat) also had their primary lesion missed. A total of 36 animals had discordant ultrasonographic and surgical findings of their primary and secondary lesions.

Lesions were missed by ultrasonography in all anatomical regions of the abdomen, although a trend was seen for missed lesions to be located in the gastrointestinal tract (P=0.08) [Figure 1]. Lesions missed by ultrasonography represented each of the gross pathological features listed in Table 4, except for urinary calculus and intussusception. Ulcerations, perforations, and ruptures of the gastrointestinal tract were missed most frequently (P=0.04) among the gross pathological features, and there was a trend for organomegaly (P=0.06) to be missed [Figure 2]. Intestinal foreign bodies were missed in five animals, but the discrepancy was not statistically significant. No apparent relationship was detected between the body size of the animal and the sensitivity of ultrasonography for the primary lesion.

Findings on ultrasound that were unrelated to the surgical lesions included prostatic cysts (n=3), renal lesions (n=5), bladder lesions (n=2), splenic/liver lesions (n=2), and a gallbladder polyp (n=1). Findings at surgery that were unrelated to the primary or secondary lesions included spleen or liver nodules (n=10), adhesions (n=1), uterine cyst (n=1), renal cyst (n=1), and thrombus in a mesenteric artery (n=1). In one animal, only one kidney was found at surgery despite an ultrasound report stating that two kidneys were present.

Discussion

One of the key outcomes of an abdominal ultrasound is a determination of the need for surgical intervention, either for diagnostic or therapeutic purposes. With respect to exploratory laparotomy, the surgical focus is on gross abnormalities (i.e., those that are visible with the naked eye) rather than histological abnormalities. Since one purpose of this study was to assess how well abdominal ultrasonography guides the surgeon, the decision was made to base the accuracy on the surgeon’s observations rather than on pathological results. This study found good agreement between ultrasonographic and surgical findings in 64% of the animals. This rate of correlation supported use of ultrasound in potential surgical candidates; however, the fact that the primary surgical lesion was missed in 25% of the cases emphasized the limitations of ultrasonography.

The finding that ulceration, perforation, or rupture was missed most frequently, and the trend for missed lesions to affect the gastrointestinal tract, suggested that the results of ultrasonography be interpreted with caution in animals with clinical signs compatible with these diagnoses. The tendency to use ultrasound to examine the gastrointestinal tract as part of a screening test for a wide variety of abdominal conditions should probably be reviewed. Other studies have identified a lack of sensitivity of ultrasound for gastrointestinal perforation.^17,18 In a series of 19 animals with gastrointestinal perforation, perforation was listed as a differential diagnosis by the ultrasonographer in 14 (74%) of the cases.¹⁷ In another retrospective study of spontaneous gastroduodenal perforation in 23 animals, abdominal ultrasonography was performed in 11 animals but was diagnostic for perforation in only one dog.¹⁸ Peritoneal fluid is one of the principal ultrasonographic findings of peritonitis or perforation of the gastrointestinal tract. Peritonitis or perforation may be missed in animals with adhesions at a perforation site and in those with minimal peritoneal fluid, or in animals that have an ultrasound performed before peritoneal fluid accumulates in sufficient amounts to be recognized.

Any delay between ultrasonography and laparotomy may increase the possibility of disparity between their findings in animals with progressive conditions. For this reason, only animals in which a preoperative ultrasound was done within 48 hours of laparotomy were included in the study. Reducing this period to ≤24 hours might improve the agreement between ultrasonographic and surgical findings. Repeated ultrasound scans have been recommended as a means of maximizing the detection of peritoneal fluid. For example, in humans with abdominal trauma, a second ultrasound done within 24 hours of the initial ultrasound increased the sensitivity for abdominal injury or intra-abdominal fluid from 31% to 72%.¹⁹

The sensitivity of ultrasonography for detection of gross peritonitis was 89% in this study. Of the 28 cases in the study with a surgical report of peritonitis, four of five with adhesions were detected ultrasonographically. This result was similar to that reported in a study of dogs that had ultrasonography performed of gastropexy sites, in which adhesions were found in all of the dogs.²⁰ The higher sensitivity in that study may be related to the fact that the investigators were concentrating on looking for adhesions in a single predictable location. In animals with spontaneous disease, adhesions vary in location and may be small, which would make them more difficult to detect.

Of the 14 animals with intestinal obstruction in this study, five were missed on ultrasonography, resulting in a sensitivity of 64%. In a prospective study that used specific ultrasonographic criteria to determine if a small intestinal obstruction was present, the obstruction was correctly diagnosed in 11 of 13 dogs and was correctly excluded in 29 of 31 nonobstructed cases (i.e., the sensitivity of the positive predictive value was 85%, and specificity of the negative predictive value was 94%).²¹ The higher sensitivities in these studies may reflect the relatively small number of cases in each study, as well as the sites and completeness of the obstructions, which can markedly affect the degree of intestinal dilatation and the likelihood of detection.

Thirty animals in this study had hepatic or splenic nodules, and the sensitivity of ultrasonography to detect these lesions was 63%. This modest sensitivity occurred because of the small size of some nodules detected surgically and/or a lack of contrast (i.e., difference in their echogenicity) compared to the surrounding parenchyma.

One hypothesis of the study was that large dogs are more difficult to examine thoroughly by ultrasonography, so the sensitivity would be lower in large dogs than in small dogs or cats; however, there was no apparent relationship between body size and sensitivity for the primary lesion. Factors that may have played a role in the outcome of the study included the experience of the ultrasonographer and the quality of the ultrasound machine and probe used; however, the sensitivities of the different examiners were not evaluated. The heterogeneous caseload in this study would have tended to undermine comparisons between ultrasonographers, as it would be difficult to make direct comparisons with such a variety of ultrasound findings. Ideally, a comparison of the sensitivities of different ultrasonographers would be based on independent, paired examinations of a series of animals.

The study may have been biased in favor of more complex or challenging cases by using the entry “exploratory laparotomy” as one of the selection criteria, because it may have excluded animals with relatively uncomplicated conditions (e.g., cystotomy for urolith removal, ovariohysterectomy for pyometra). Only animals recorded as having exploratory laparotomy were chosen based on the assumption that they were more likely to have had a complete examination of the abdomen at the time of surgery. The surgeon’s knowledge of the ultrasound report prior to surgery may also have influenced the surgical findings. Clearly, advantages would be associated with repeating this study using a prospective methodology, in which both ultrasonographer and surgeon independently complete a common form of report; however, masking of the surgeon to the results of the ultrasound would complicate the management of many clinical cases.

Conclusion

This study showed good overall agreement between abdominal ultrasound findings and exploratory laparotomy; however, major discrepancies occurred in 25 of 100 animals studied. The anatomical location of the primary lesion was not significantly associated with missed lesions; however, a significant association was found between ulceration or perforation of the gastrointestinal tract and lesions missed on ultrasonography. Based on the results of this study, ultrasonography is a useful presurgical diagnostic tool. Prospective studies would be beneficial to further determine the accuracy of preoperative ultrasonography.