Diagnostic accuracy of ultrasonography in adults with obstructive jaundice

Obstructive jaundice, otherwise called surgical jaundice, is a common condition globally^(1–3). Common causes of obstruction of the extrahepatic biliary tree in adult patients include choledocholithiasis, chronic pancreatitis, as well as neoplasms of the pancreas, gallbladder, biliary tract, or the ampulla of Vater. Other less prevalent causes include metastasis to the porta hepatis, hepatic tumour adjacent to the hilum, perihepatic lymphadenopathy, sclerosing cholangitis and other forms of cholangitis^(3,4).

Many patients present with advanced surgical jaundice requiring various examinations to determine the cause and level of obstruction. The array of invasive and noninvasive radiological techniques commonly employed for investigating hepatobiliary lesions includes computed tomography (CT), percutaneous transhepatic cholangiography (PTC), endoscopic ultrasound (EUS), endoscopic retrograde cholangiopancreatography (ERCP), helical CT cholangiography, Magnetic Resonance Cholangiopancreatography (MRCP), radionuclide imaging, and ultrasonography^(1,5–8). These examinations are useful to varying degrees for assessing both the cause and site of obstruction. However, ultrasonography is the least invasive initial imaging modality for the evaluation of jaundiced patients^(4–6,9,10).

In resource-poor settings, ultrasound is often the only readily available method, with many of the other modalities being too expensive or unavailable to many patients. Furthermore, abdominal ultrasound is considered to be the first initial imaging procedure in obstructive jaundice^(1,5,11,12). Ultrasonography could also be used for some imaging-guided hepatobiliary interventions.

This study aims to determine the sensitivity and specificity of ultrasound in detecting the causes of obstructive jaundice in our locality, where more sophisticated imaging modalities may not be readily available on a routine basis.

This was a hospital-based, cross-sectional study of eighty consecutive adult patients referred for abdominal ultrasound in the Radiology Department of our hospital on account of clinical and biochemical features of obstructive jaundice. The study was approved by the Ethics and Research Committee of the institution (Approval number: IRB/IEC/0004553), and was conducted in accordance with the Declaration of Helsinki (2013).

This multidisciplinary study involved the departments of Radiology, General Surgery and Histopathology. All patients were referred to the Radiology Department from the Surgery Department for sonographic evaluation. They were then referred back to the Surgery Department, where they had surgery. Later, histopathological examinations of surgical biopsy specimens were performed where applicable. The surgical and/or histopathology reports were ultimately correlated with the ultrasonographic diagnosis.

All adult patients with clinical and biochemical (liver function test) features of obstructive jaundice were enrolled consecutively. Written informed consent was obtained from all participants after the study had been adequately explained to them. The following patients were excluded from the study: children with jaundice (<16 years old) and adult patients with features of pre-surgical jaundice or in whom liver function test (LFT) did not show features of obstruction. All the patients underwent sonographic examinations before the treatment/intervention was instituted. They were all scanned by the first author who was a 5^th-year radiology resident, under close supervision of two consultant radiologists with 17–28 years’ experience.

A Mindray real-time ultrasound scanner model DC-6 (Shenzhen Mindray Bio-medical Electronics, Nanshan, Shenzhen, China) with Doppler facilities and a curvilinear transducer (frequency = 2.5–5.0 MHz) was used.

The patients were examined after an overnight fast of 8 hours, with the stomach distended with water to remove excess gas, and to create an acoustic window to view the pancreas adequately.

With the patient in the supine position, a coupling gel was applied to the exposed abdomen in the right upper quadrant. Scanning was done in both longitudinal and transverse planes. The patient was then placed in the left posterior oblique or left lateral decubitus position in order to scan the intra and extra-hepatic ductal systems. The common bile duct (CBD) was imaged by placing the transducer below the right costal margin in the region of the midclavicular line, with the patient in the oblique position. When visualized, the CBD was traced from the hepatic hilum to the retro-pancreatic portion and down to the papilla. It was imaged along its entire length, and its diameter was measured at three levels viz: at the porta hepatis, close to the liver, and distally near the head of the pancreas⁽¹³⁾. The diameters of the common hepatic duct and the common bile duct were measured, with 6 mm taken as the upper limit of normal for both. Above 6 mm was considered dilated (1 mm per decade was added to this upper limit in patients who were older than 60 years). The intra-hepatic ducts were also evaluated. The hepatic hilum and gall bladder were assessed with the patient in the left lateral decubitus position. We used color flow Doppler to differentiate tumefactive sludge from the gallbladder mass. Doppler examination was also used to assess vascularization of the gallbladder wall and cancer or formation of a twinkling artifact in a choledocholithiasis.

Where relevant, scanning was performed with the subject sitting erect in order to assess whether gallstones were mobile. The gallbladder wall thickness was regarded as normal when it was ≤3 mm, including the measurement of its anterior wall⁽¹³⁾.

The spleen, stomach, bowel loops, and para-aortic regions were also scanned to exclude pathologies. Ascites was noted, when present. Sonographic diagnosis of the wide variety of underlying etiologies was based on published, well-established criteria. Sonographic findings and diagnoses were finally compared to operative findings and/or histopathological reports.

The data were analysed using IBM SPSS Statistics software version 22.0 for windows (IBM Corp., Armonk, N.Y., USA). The test of normality was performed with the Kolmogorov Smirnov’s test. Chi-square was used to compare categorical data, while analysis of variance (ANOVA) was used to compare continuous data between groups, with Scheffe/Bonferroni post-hoc analysis done for intergroup differences. Continuous variables were presented as mean ± SD (standard deviation), while categorical variables were expressed as percentages and frequencies. Ultrasonographic diagnoses were compared to the final surgical and/or histopathologic diagnoses to calculate the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of ultrasound. Statistical significance was set at p ≤0.05.

Eighty cases of obstructive jaundice were included in this study. The age range was 16 to 82 years, with a mean of 51.06 ± 14.95 years. The peak age group was the sixth decade with 23 (28.8%) patients. There were nearly twice as many females as males, with 28 (35%) males and 52 (65%) females, giving a male to female ratio of 1:1.9. (Tab. 1).

On ultrasound, pancreatic carcinoma 23 (28.8%), choledocholithiasis 15 (18.7%), and gallbladder carcinoma 13 (16.3%) were the most prevalent diagnoses. Less frequent findings included periampullary carcinoma 8 (10.0%), cholangiocarcinoma 4 (5.0%), and metastatic lesions. Other findings are shown in Tab. 2 and Fig. 1, Fig. 2, Fig. 3. A total of 75 patients (93.7%) had dilatation of the extrahepatic ducts. Forty (50%) patients had dilatation of the extra-hepatic biliary tree up to the distal CBD, while dilatation up to the proximal CBD was seen in 20 (25%) patients. Common hepatic ductal dilatation was seen in 15 patients (18.8%). Only 5 (6.3%) of the patients had no significant ductal dilatation on ultrasound. The CBD diameters ranged from 3.0 mm to 32.0 mm, with a mean of 13.7 ± 7.3 mm, while the common hepatic duct diameters ranged from 4.5 mm to 28.0 mm, with a mean of 12.8 ± 4.8 mm. When visualized, the pancreatic duct diameter ranged between 2.9 mm and 13.0 mm, with a mean of 5.1 ± 2.6 mm (Tab. 3A).

Fig. 1.

Fig. 2.

Fig. 3.

A comparison of the mean CBD diameters with respect to the level of obstruction using analysis of variance (ANOVA) revealed a statistically significant difference (p <0.001; Tab. 3B). Posthoc tests (Bonferroni and Scheffe) showed that the statistically significant difference between the three groups was mainly due to a higher CBD diameter in the intra-pancreatic obstruction group than in the other two groups (p <0.001). The difference in the mean CBD diameter between the hilar and supra-pancreatic obstruction groups was not statistically significant (p = 0.12). A similar pattern was observed between the groups when the mean common hepatic duct diameters were compared.

The preoperative ultrasonographic diagnoses were correlated with the definitive diagnoses at surgery and histopathological reports of operative biopsies for malignant conditions (Tab. 4). The overall sensitivity of ultrasound in detecting the etiology of obstructive jaundice was 76.6%, while the specificity was 98%. For specific disease entities, the values of sensitivity and specificity were 81.0% and 94.6%, respectively, for pancreatic carcinoma, 70.7% and 93.7% for choledocholithiasis, and 80% and 98% for gallbladder carcinoma. The other hepatobiliary pathologies are shown in Tab. 4.

Intraoperative dissection of the hepatopancreatobiliary tree can be challenging for surgeons^(1,3,14), and precise preoperative diagnosis of the exact site of obstruction is beneficial for planning surgical interventions⁽³⁾. In resource-rich settings, this is usually accomplished by a combination of different imaging and endoscopic modalities, some of which may not be readily available for surgeons in resource-limited settings.

This study investigated the diagnostic accuracy of ultrasonography for reaching a definitive diagnosis in patients with obstruction of the extra-hepatic biliary tree. We observed a good correlation between the definitive diagnosis and sonographic diagnosis, with an overall sensitivity of 76.6% and specificity of 98%. These findings are similar to the study by Kumar et al.⁽¹⁵⁾ which concluded that ultrasonography as a single radiological investigation was sufficient in the evaluation of most patients with obstructive jaundice, with a sensitivity of 84% in detecting its cause. In contrast, Dewbury et al.⁽¹⁶⁾ obtained a lower sensitivity of ultrasound for detecting the cause of obstructive jaundice, with a value of 58%. This disparity could be attributed to a higher resolution of the modern ultrasound system we used, and possibly a lower level of experience at the time their study was carried out in 1979.

The most frequent level of obstruction in our study was at the intra-pancreatic common bile duct, as seen in 40 (50%) of the patients, followed by the supra-pancreatic common bile duct in 20 (25%) patients. Fifteen (18.8%) patients had obstruction at the level of the porta hepatis, while there was no sonographic evidence of obstruction in 5 (6.3%) patients. Similar findings were noted by Upadhyaya et al.⁽¹⁷⁾ in India, who noted that 27% of their patients had obstruction at the level of the porta hepatis, 34% at the supra-pancreatic level, and 38% at the intra-pancreatic level. However, Ghimire et al.⁽¹⁸⁾ in Nepal reported that their patients had obstruction at the hilar level in 38% of cases, and supra-pancreatic duct obstruction in 33% of cases, while 28% of their patients had obstruction at the intra-pancreatic level. Differences in the etiology of the obstruction may account for this variation.

The mean diameters of the common bile duct (CBD) and common hepatic duct (CHD) in intra-pancreatic bile duct obstruction were statistically significantly higher than ductal diameters recorded in obstructions at other levels. This implies that a more distal obstruction of the biliary tree is associated with a significantly worse proximal ductal dilatation compared to other levels of obstruction. This may account for a relatively higher sensitivity of ultrasound in detecting lower duct obstruction in our study compared with that of Ghimire et al.⁽¹⁸⁾

Our sensitivity and specificity values in diagnosing pancreatic carcinoma on ultrasound were 81% and 94.6%, respectively. This is quite remarkable, when compared with the findings reported by Admassie et al.⁽⁹⁾, who obtained sensitivity and specificity values of 50% and 90%, respectively. The higher sensitivity noted in our study may be due to greater dilatation of the common bile and hepatic ducts (i.e. worse severity), higher resolution of the ultrasound equipment, and the technique of examining patients in the anterior oblique position with the stomach fully distended with water. D’Onofrio et al.⁽¹⁹⁾ reported a sensitivity of 88%, a specificity of 97%, and an accuracy of 96% in their study in which they employed contrast agents to enhance sonographic images.

The sensitivity, specificity, and accuracy of ultrasound in diagnosing gallbladder carcinoma in this study were 80%, 90%, and 94%, respectively. This is lower than the 100% sensitivity level reported by Upadhyaya et al.⁽¹⁷⁾ We attribute this to an additional unrecognized sonographic feature of advanced gallbladder carcinoma that we encountered: non-visualization of the gall bladder along with an echogenic structure casting posterior acoustic shadows adjacent to the porta hepatis. This was initially misdiagnosed as chronic cholecystitis with choledocholithiasis. At surgery, however, the findings were that of advanced infiltrating gallbladder carcinoma, walled off by bowel loops. This sonographic feature had been documented before in a report by Alatise et al.⁽²⁰⁾, in which patients with gallbladder carcinoma were similarly misdiagnosed at ultrasound as having cholecystitis with cholelithiasis. Gohil et al.⁽²¹⁾ had earlier alluded to this possibility, noting that cholelithiasis and cholecystitis were close differentials of gallbladder carcinoma.

The sensitivity, specificity, and accuracy of ultrasound in diagnosing choledocholithiasis were 70.7%, 93.7% and 88.8%, respectively. We obtained a higher level of accuracy than Uphadyaya et al.⁽¹⁷⁾, who recorded the value of 63.1%. Improved resolution of the equipment used in the present study and increased diagnostic efforts to determine the cause of jaundice may account for this difference. In this study, 3 patients presenting with choledocholithiasis had a normal biliary tree – probably because the condition was still early, hence ductal obstruction had not yet occurred. These cases were misdiagnosed as non-obstructive jaundice, similarly to the experiences of Upadhyaya et al.⁽¹⁷⁾

The sensitivity of ultrasound in diagnosing cholangiocarcinoma in our study was 100%, which is at variance with the study by Robledo et al.⁽²²⁾ reporting that the detection of bile duct cancer varies from 21% to 90%, with the distal common bile duct having the lowest percentage. Both proximal and distal bile duct cancers were encountered in this study, and the sensitivity and accuracy of ultrasonography were not impaired despite this. The higher sensitivity we encountered may be attributable to the fact that all extra-hepatic bile duct masses seen in this study were polypoidal intra-luminal tumors. This was also the finding of Anand et al.⁽²³⁾, who reported an accuracy of 100% for polypoidal intra-luminal tumors, 13% for sclerosing tumors, and 29% for exophytic tumors.

The non-dilatation of the biliary ducts in 5 of our patients probably accounted for the misdiagnosis of 2 cases of cholangitis and 3 cases of choledocholithiasis on ultrasound. They were mistakenly diagnosed as cholecystitis and cholelithiasis, respectively. Sample et al.⁽²⁴⁾ had a similar experience, reporting that 10% of their patients had non-dilated biliary trees due to sclerosing cholangitis which they misdiagnosed as non-obstructive jaundice. These cases emphasize the point that in the absence of ductal dilatation on ultrasonography, the presence of a clinical history of obstructive jaundice with biochemical findings also indicating an obstructive etiology should prompt further diagnostic investigations using other imaging modalities⁽²⁵⁾.

In conclusion, ultrasonography is a helpful tool for evaluating patients with features of surgical jaundice in resource-poor settings. It can demonstrate both benign and malignant causes of obstructive jaundice. The most common causes of obstructive jaundice in our study were choledocholithiasis and carcinoma of the head of the pancreas for benign and malignant causes, respectively. The level of sensitivity obtained in the study is quite comparable to that reported in some previous studies for CT, MRCP, PTC, and ERCP^(3,15,26,27). We believe that the meticulous, protocol-driven^(11,12) use of ultrasound for the diagnosis of obstructive jaundice would further help to improve surgical management.