Accuracy of Multiphasic Helical Computed Tomography and Intraoperative Sonography in Patients Undergoing Orthotopic Liver Transplantation for Hepatoma: What is the Truth?

Abstract

Objective

To determine the real value of liver imaging in cirrhosis by macro- and histomorphologic examination of the entire organ after orthotopic liver transplantation for hepatocellular carcinoma (HCC).

Summary Background Data

In comparative studies, a virtual sensitivity of up to 94% is described for helical computed tomography in HCC staging. The tumor detection rate of intraoperative ultrasonography (IOUS) is reported to be almost 100%.

Methods

This prospective observational study comprised 23 patients with HCC in cirrhosis admitted for orthotopic liver transplantation. Results of preoperative triphasic helical computed tomography (HCT) and IOUS were correlated with histopathologic results after 3-mm-slicing of the explanted liver.

Results

Overall, 179 liver segments were examined by HCT, IOUS, and MHM. Fifty-two malignant lesions and 10 dysplastic nodules were revealed by MHM. Using HCT, 13 HCCs could not be identified in 8 patients and 15 results were falsely positive in 10 patients. The detection rate of dysplastic nodes was 40% for HCT and 60% for IOUS. IOUS missed four HCCs in four patients and had six false-positive results in six patients. In a segment-based analysis, the overall accuracy of IOUS was significantly higher for IOUS (95.5%) versus HCT (89.9%). In the lesion-by-lesion analysis, the sensitivity was 92.3% for IOUS and 75.0% for HCT, with a significant difference.

Conclusions

Correlation of explanted liver pathologic results offers precise evaluation of imaging modalities. The data of this histopathologically based study confirm that IOUS is significantly superior in staging HCC in cirrhosis versus CT, even after technical refinements through enhanced multiphasic high-velocity helical scanning.

Orthotopic liver transplantation (OLT) has been shown to be an effective treatment of hepatocellular carcinoma (HCC) in certain stages. ¹ However, in patients with advanced tumors, the oncologic outcome after surgery is disappointing. ² In early stages OLT seems to be the treatment of choice, especially in patients with cirrhosis. ^1,2 In the light of limited donor resources, resection still is important. Palliative procedures such as chemoembolization, cryotherapy, thermoablation, and ethanol injection are frequently used in advanced disease ^3–5 because the surgical death rate is enhanced in these situations. ² Above all, adequate patient selection based on accurate staging is a crucial precondition for an appropriate treatment decision.

In terms of preoperative staging, multiphasic helical computed tomography (HCT) has become the standard tool for the detection of nodular lesions in liver cirrhosis. ^6,7 Most radiologic studies dealing with the value of several staging modalities based their gold standard either on intraoperative examination and the findings after partial resection or on follow-up studies with or without biopsy confirmation. ^8–12 Besides detection of hepatic metastases, intraoperative ultrasound (IOUS) evolved into an indispensable tool in hepatic resection or intraoperative intervention for HCC. ^13–16 Due to the lack of a more precise method, the diagnostic value of IOUS was frequently assessed by examination of resected liver segments or by follow-up studies of the liver remnant after hepatic resection. ^17–19 Radiologic studies on liver imaging in patients scheduled for OLT are rare and generally are limited by a long interval between imaging and OLT. ^20–24

To our knowledge, until now there has been no report dealing with the evaluation of triphasic HCT performed within 3 weeks before OLT and IOUS in staging HCC in transplant patients. The present study is an attempt to reveal the true value of multiphasic HCT and IOUS of the liver in patients with HCC in cirrhosis. Lesion-by-lesion analysis and segment-by-segment analysis were performed to evaluate sensitivity, specificity, predictive value of a positive value, and predictive value of a negative result as well as the overall accuracy of these methods.

METHODS

Between October 1998 and January 2000, 23 patients (3 women, 20 men; age range 35.8–67.8 years; mean age 57.2 years) consecutively underwent hepatectomy and liver transplantation for HCC at a single university center. In all except one patient with hemophilia, HCC was confirmed before surgery by percutaneous biopsy. Patients were scheduled for OLT only when extrahepatic tumor manifestation was excluded. Triphasic HCT of the liver was routinely performed 2 to 12 hours before OLT to exclude advanced disease. During surgery, the liver was scanned by ultrasound imaging before hepatectomy. Pathologic examination followed the day after OLT. The data were collected prospectively.

HCT examinations were performed on a Somatom Plus 4 scanner (Siemens, Erlangen, Germany) with a collimation of 5 mm at a table feed of 7.5 mm/s and breath command (rec. increment 4 mm). The helical liver scans were obtained before intravenous administration of 150 mL of the nonionic contrast material Jopamidol (Ultravist 300 mg/mL; Schering, Berlin, Germany) (flow 5 mL/s, bolus-triggered), at the arterial phase (20 seconds after injection of the contrast agent) and at the portal venous phase (50 seconds after injection). Generally the examination was performed between patient admission and OLT. In five instances prior scans (performed <3 weeks before surgery) were evaluated because immediate preoperative scanning was not available. HCT scans were interpreted separately by two experienced independent radiologists (R.P., P.P.) who were unaware of the histopathologic results.

All patients underwent IOUS with a B&K Panther 2002 ADI unit (B&K Medical, Gentofte, Denmark). For real-time B-mode scanning, usually 5-MHz curved-array probes or linear probes with 5 or 7.5 MHz were used as appropriate (B&K Medical). After a right subcostal incision with extension to the left or to the xiphoid, dissection of the falciform and coronary ligament, and surgical exploration, livers were scanned in longitudinal and axial orientation. The two surgeons who performed IOUS (J.Z., C.S.) were not aware of the results of preoperative examination.

The explanted livers were fixed in formalin. Pathologic examination was performed by an experienced pathologist (F.W.) after horizontal sectioning of the entire liver in 3-mm slices. Nodular lesions were identified macroscopically and were further processed for conventional hematoxylin and eosin staining and microscopic examination. In particular, nodular lesions were histopathologically classified as regenerating nodules, dysplastic nodules, or HCC. ²⁵

Lesions were prospectively documented with regard to size, segment-related localization according to the Couinaud classification, ²⁶ and histologic classification. ²⁵ Results from HCT and IOUS were compared with pathologic findings. A lesion-by-lesion analysis was worked out. Satellite lesions situated at the very proximity (within 1 cm) of malignant tumors were not classified as particular lesions. For comparison of sensitivity and specificity, predictive value of a positive or negative result, and overall accuracy of imaging modalities, a segment-by-segment analysis was also performed. Segments affected by malignancy were scored as positive. The significance of differences between methods was established using the chi-square test. P < 0.05 was interpreted as statistically significant.

RESULTS

All the patients had HCC, in two in combination with primary cholangiocellular carcinoma. Sixteen patients were pretreated interventionally, in 12 by ethanol instillation and in 3 by transarterial embolization. All but 2 patients had liver cirrhosis (hepatitis C, n = 7; alcoholic cirrhosis, n = 6; hepatitis B, n = 4; primary biliary cirrhosis, n = 1; hemochromatosis, n = 1; cryptogenic cirrhosis, n = 2). Three patients had already undergone hepatic resection before OLT. Overall, 179 liver segments were observed.

By pathology, 52 malignant lesions were detected (50 HCC, 2 cholangiocellular carcinoma). The median diameter of malignant lesions was 20 mm (range 4–100). Sixteen lesions (30.8%) had a diameter of 10 mm or less. Sixteen patients (69.6%) had at least two malignant liver lesions. Pathologic T-staging and grading is shown in Table 1. Only 14 HCC lesions (26.9%) were detectable by palpation and/or inspection after celiotomy. In seven patients, 10 dysplastic nodules with a median diameter of 8 mm (range 4–15) were revealed pathologically after explantation.

Table 1. DISTRIBUTION OF T STAGES AND GRADING

HCC, hepatocellular carcinoma; CHC, cholangiocellular carcinoma; Gx, grading not possible due to tumor necrosis after ethanol injection.

Lesion-by-Lesion Analysis

HCT detected 54 lesions suspected of being malignant; 15 of them were falsely positive in 10 patients (43.5%). Nine patients (39.1%) were correctly staged by HCT. In eight patients (34.8%), 13 HCC manifestations were not detected by HCT; in five patients (21.7%), these undetected lesions were located at segments not affected by another malignant focus. The median diameter of malignant lesions missed by HCT was 10 mm (range 5–20); 7 of them were smaller than 11 mm. IOUS detected 54 lesions suspected of being malignant. Two of them were hemangioma with an atypical sonographic appearance and four were regenerating nodules. Overall, these false-positive lesions were seen in six patients (26.1%). In four patients (17.4%) a lesion was missed by IOUS; three of them were smaller than 11 mm. In three cases (13%) these lesions were located in segments not affected by other lesions. IOUS correctly staged 14 patients (60.9%).

In 13 patients (56.5%) IOUS added information to the results of HCT regarding the number of malignant lesions. In two of these cases IOUS disproved a lesion falsely detected by HCT. IOUS showed a significantly higher sensitivity and predictive value of a positive test compared with HCT. Results of the lesion-by-lesion analysis are highlighted in Table 2.

Table 2. LESION-BY-LESION ANALYSIS

HCT, helical computed tomography; IOUS, intraoperative ultrasound.

*P = .017.

†P = .029.

HCT detected 40% of pathologically identified dysplastic nodules, whereas IOUS revealed 60%.

Segment-by-Segment Analysis

At pathologic examination, 52 malignant lesions involved 57 segments. Malignant segment involvement was correctly described by HCT in 87.7% and by IOUS in 94.7% of affected segments (Tables 3, 4). When examined by HCT, 11 segments were falsely interpreted to be involved by malignancy. By IOUS, five segments were falsely thought to bear malignant lesions. IOUS was shown to have a significantly higher overall accuracy than HCT (P = .042). Sensitivity, specificity, predictive value of a negative test, and predictive value of a positive test was higher for IOUS than HCT, but without a significant difference.

Table 3. SEGMENT-BY-SEGMENT ANALYSIS

HCT, helical computed tomography; IOUS, intraoperative ultrasound.

Table 4. RESULTS AND ACCURACY OF HCT AND IOUS (SEGMENT-BY-SEGMENT ANALYSIS)

HCT, helical computed tomography; IOUS, intraoperative ultrasound.

DISCUSSION

Careful staging of hepatic tumors is the basis of appropriate selection and is crucial for a tailored treatment. Generally, in solid tumors, outcome is closely related to the stage of the disease. Mainly patients with early-stage disease seem to benefit substantially from surgery. ^1,2 Among the available modern radiologic modalities, HCT is widely established as offering a sensitivity comparable with that of CT during arterial portography and during hepatic arteriography and with contrast-enhanced magnetic resonance imaging. ^9,10 By performing multiphasic HCT, the sensitivity for detecting HCC in cirrhosis is reported to reach 80% to 94% in nontransplant studies. ^9,11 In these papers, the gold standard commonly is based on the results of intraoperative examination, pathologic investigation of the resected specimen, and follow-up tests, respectively.

To evaluate the real tumor extension, we examined explanted livers macromorphologically and pathologically after 3-mm sectioning. This technique led to the detection of even very small HCC manifestations. Comparing the results of imaging modalities with the histopathologic data provides a clearly more precise assessment of the value of these methods than the usually performed comparison of radiologic modalities with a gold standard or the pathologic assessment of only a part of the liver after partial hepatic resection. ^20–24

Recently some authors have reported tumor detection rates of 37% to 71% by multiphasic HCT in transplant patients, ^7,20–23 a sensitivity of 58% for CT with iodized oil ^24,27 and a sensitivity of 85% for CT during arterial portography. ²⁴ Krinksy et al ²⁸ observed disappointing results of gadolinium-enhanced magnetic resonance imaging compared with pathologic examination of explanted livers (HCC tumor detection rate 55%).

We believe that the main limitation of previously reported studies in transplant patients is the long interval between imaging and OLT. 20,22,24,28 Further, in some groups liver sections were comparably thick (up to 10 mm), which may limit the detection of small lesions.

To our knowledge, this is the first study assessing the value of multiphasic HCT performed immediately before OLT and of IOUS in patients undergoing OLT for HCC in liver cirrhosis. We chose thin (3-mm) slicing of the explanted organs to detect even very small HCC lesions.

Unfortunately, the accuracy of HCT was disappointing: this technique provided a completely correct result in only 9 of 23 patients. Tumor-related sensitivity was 75%. This is in contrast to a previously reported sensitivity of up to 94%, ¹¹ when assessed in a nontransplant population, but might more accurately reflect the routine clinical experience of preoperative staging.

The value of IOUS in the detection of colorectal liver metastases is well established. ^15–19 The sensitivity of IOUS is reported to be up to almost 100%. ¹⁷ Less is known about the efficacy of IOUS in tumor detection in cirrhotic livers. Tumor detection rates of 95% to 97% have been published. ^29,30 Due to the pathologic macroarchitecture of the liver, IOUS is very demanding in cirrhosis.

The specificity of IOUS seems to be impaired by regenerating or dysplastic nodules showing sonographic patterns difficult to distinguish from HCC. This is confirmed by our data and experience. In almost every fifth patient, a regenerating nodule was sonographically interpreted as a malignant lesion by IOUS.

However, in three quarters of patients the surgeon was not able to identify the malignant liver tumor due to cirrhosis. Thus, in patients who have received any kind of local tumor therapy as resection or interventional intraoperative treatment of HCC, IOUS is crucial in localization and guidance, especially in patients with cirrhosis. Because of the limited specificity of IOUS, diagnostic evidence should be obtained from sonographically guided biopsies before therapeutic intervention to avoid treatment of false-positive nodules.

In addition to preoperative findings, IOUS correctly revealed “new” malignant lesions in about a quarter of patients. This remarkably confirms the superior diagnostic value of IOUS even in cirrhosis.

As a limitation to our study, we could not perform magnetic resonance imaging or CT during arterial portography and hepatic arteriography routinely because of logistic circumstances and limited access. Previously, sensitivity and specificity of contrast-enhanced magnetic resonance imaging was reported to be far higher than 90% in detecting HCC. ¹⁰ On the other hand, gadolinium-enhanced magnetic resonance results were discouraging when correlated with pathologic examination after OLT. ²⁸ The sensitivity of HCT during arterial portography and hepatic arteriography was shown to be 91% to 96% in a nontransplant setting, ^10,11 but some authors stress the high false-positive rates of up to 35% using this technique. ^11,31

In conclusion, accurate staging of malignancy situated in the cirrhotic liver remains difficult. Unfortunately, the accuracy of triphasic HCT is not acceptable at present. It seems that the results of nontransplant studies dealing with the accuracy of several imaging techniques must be seen in a new light. Perhaps technical improvements in ferumoxide-enhanced magnetic resonance imaging or multidetector-helical CT will raise the accuracy of noninvasive imaging of HCC in cirrhosis. ³² The value of new diagnostic modalities should be evaluated by studies with complete pathologic correlation whenever possible.

Our data show that IOUS is highly sensitive even in patients with cirrhosis. IOUS is indispensable, especially when resection or cryotherapy is planned, but its specificity is limited in our hands. If a suspect lesion that potentially demands adaptation of the treatment is primarily detected by IOUS, evidence should be gained by taking sonographically guided biopsies and subsequent frozen section. If noninvasive methods fail to substantially improve the quality of HCC detection in liver cirrhosis, the role of laparoscopic IOUS ^33–35 in staging HCC should be considered.

Acknowledgments

The authors thank Marlene Thiem, Martin Bodingbauer, Christoph Mittermayer, and Emanuel Sporn for their assistance in this study.