Impact of positron emission tomography on strategy in liver resection for primary and secondary liver tumors

Abstract

Purpose

Outcome of patients with metastatic disease mainly depends on accurate preoperative tumor staging. ¹⁸[F]fluorodeoxyglucose positron emission tomography (18F-PET) has been proven to be a valuable diagnostic tool in a number of different tumors but its direct influence on liver surgery has not been thoroughly investigated.

Materials and methods

Between July 1999 and March 2000, 50 consecutive patients with 174 suspected liver lesions were admitted to the University Hospital Jena. All 50 patients underwent abdominal ultrasound, CT-scan, and 18-FDG positron emission tomography scanning. In 23 patients the diagnostic work-up was completed by MRI scan.

Results

Altogether there were a total of 174 histologically proven intrahepatic lesions, nine of which were benign. The sensitivity, specificity, and positive predictive value of PET for all hepatic lesions was 82%, 25%, and 96% compared with 63%, 50%, and 96% for abdominal ultrasound, 71%, 50%, and 97% for CT-scan, and 83%, 57%, and 97% for MRI-scan. In 23 of 50 patients 24 extrahepatic lesions were identified. In these patients the sensitivity and specificity of PET—compared to abdominal ultrasound, CT-scan, and MRI-scan for all extrahepatic lesions—was 63% and 60%, 29% and 25%, 47% and 50% and 40% and 50%, respectively. The findings on PET scan had a direct impact on operative management in nine patients (18%).

Conclusions

Our series demonstrates good sensitivity and specificity for the detection of primary and secondary liver lesions which is superior to ultrasound and CT scan but not to MRI scan. The main value of PET scan consists in the detection of extrahepatic tumor (64%). Due to better detection of extrahepatic tumor, FDG-PET is a very useful addition to the currently used anatomically-based images in all cases of advanced tumor spread with high risk of extrahepatic tumor.

Introduction

Surgical therapy often offers the only possible chance for cure in patients with liver metastases or primary liver tumors. Planning of liver resection in these situations mainly relies on accurate determination of the number and location of the lesions within the liver. Preoperative cancer staging allows selection of patients for curative hepatic resection and avoidance of unnecessary surgery. In the case of secondary liver lesions, extrahepatic tumor manifestation may be a contraindication to any further surgical intervention.

Abdominal ultrasound (US), computed tomography (CT), and magnetic resonance imaging (MRI) can provide useful information about the size, location, and type of a lesion. However, differentiation between malignant and benign lesions can be difficult, particularly distinguishing scar tissue from tumor recurrence in previously resected patients. Positron emission tomography (PET) using ¹⁸[F]fluorodeoxyglucose (FDG-PET) is a functional imaging method for staging malignancies that detects increased glucose utilization in cancer cells. The technique allows the whole body to be imaged for metastatic spread. During recent years PET has been transformed from a research tool into an important imaging modality for clinical oncology [10, 18].

The present study analyses the power of FDG-PET to differentiate benign from malignant lesions, and to predict resectability of liver lesions in comparison with abdominal ultrasound (US), computed tomography (CT), and magnetic resonance imaging (MRI). In the second stage of this study we focused our interest on the ability of PET to detect extrahepatic tumor manifestation and its direct influence on choice of surgical procedure.

Materials and methods

Between July 1999 and March 2000, 50 consecutive patients (26 men, 24 women) with 174 suspected liver lesions were admitted to our clinic. All 50 patients underwent abdominal ultrasound and CT-scan. The diagnostic work-up was completed by MRI scan in the case of multiple liver tumors or if the US and CT findings were equivocal (23 patients). Abdominal ultrasound was carried out in our surgical department by two experienced ultrasonologists (3.5 mHz curved array scan, Logic 500-Kranzbühler Germany).

For computed tomography (CT) a General Electrics Lightspeed Multidetector scanner was used. CT images were obtained through liver and abdomen with contiguous 7.5-mm slices using oral (30 ml Peritrast/1,000 ml tea) and intravenous contrast (Ultravist 300).

For magnetic resonance tomography (MRI) a Siemens Vision 1.5T-scanner with body array was used. The patients underwent two examinations: on the first day-examination, unenhanced coronary and transversal HASTE-sequences (TR 4.4 ms; TE 64.0 ms, 6-mm slice thickness) and dynamic transversal T1-flash 2D-double-echo sequences (TR 152.0; TE 2.7 ms), with 6-mm slice thickness before and after bolus injection of a gadolinium chelate (0.1 mmol/kg body weight Magnevist), were obtained with the breath-hold technique.

PET was performing using an ECAT EXACT 47 (Siemens-CTI/Knoxville, Tenn., USA) with the whole-body mode implemented as standard software. In this mode, the scanner acquires 24 two-dimensional sections over an axial field of view of 15.5 cm. Up to seven increments were programmed. Patients fasted for at least 6 h before undergoing PET. Static emission scanning was performed 60–90 min after injection of 300–400 MBq ¹⁸[F]fluorodeoxyglucose (FDG). The acquisition time was 10 min (8- min emission, 2-min transmission) per bed position. Images were reconstructed using an iterative reconstruction algorithm (OSEM: two iterations, eight subsets). The reconstructed scans were documented in black and white by using a colour laser printer with a digital interface (Tektronix 560). The PET scans were interpreted by two nuclear medicine physicians who were blinded to the results of the other imaging methods. Interpretation of PET scans was performed visually rather than by quantitative analysis. A lesion was defined as a focus of increased FDG uptake above the intensity of the background as long it was outside area, which can show FDG accumulation under normal conditions. The final diagnosis of suspicious lesions was confirmed histologically in all patients.

Sensitivity (SE), specificity (SP), and positive predictive value (PPV) for intrahepatic lesions was calculated for abdominal ultrasound, CT-, MRI-, and PET-scan. Sensitivity and specificity were derived based on the final histological findings. When miliary hepatic metastases or multifocal hepatic tumors occurred (more than ten lesions) only ten lesions were included for the calculation. Histological results in 40 patients were achieved by liver resection or transplantation. In six patients the operation consisted of exploratory laparotomy only. Four patients underwent percutaneous fine needle aspiration.

In cases of extrahepatic tumor, the basis for calculation was not the number of single metastases but number of affected organs (due to multiple bone metastases, multiple pulmonary metastases or miliary peritoneal tumor deposits).

Results

Among 15 patients with primary liver tumor, there were ten hepatocellular carcinomas, one adenoma, and four cholangiocarcinomas. Liver metastases originated from colorectal cancer in 24 instances, breast cancer in three, and from two cases of gall bladder and ovarian cancer, respectively. A single liver metastasis from pharyngeal, thyroid gland cancer, and a leiomyosarcoma were also treated.

Hepatic lesions

Altogether there were a total of 174 histologically proven intrahepatic lesions, nine of which were benign - one adenoma, one gall bladder empyema, and seven regenerative nodules in cirrhotic liver. The sensitivity, specificity, and positive predictive value of PET for all hepatic lesions was 82%, 25%, and 96% compared with 63%, 50%, and 96% for abdominal ultrasound, 71%, 50%, and 97% for CT-scan, and 83%, 57%, and 97% for MRI-scan (Table 1).

Table 1.

Sensitivity, specificity, and positive predictive value of PET. (TP true positive, FN false negative, TN true negative, FP false positive, SEN sensitivity, SPE specificity)

Modality	TP	FN	TN	FP	SEN %	SPE %	PPV %
174 Intrahepatic lesions in 50 patients
Ultrasound	105	61	4	4	63	50	96
CT	118	48	4	4	71	50	97
MRIa	94	19	4	3	83	57	97
PET	137	29	2	6	82	25	96
74 Intrahepatic lesions from colorectal cancer in 24 patients
Ultrasound	63	10	1	0	86		100
CT	64	9	0	1	88		98
MRIa	42	4	0	0	91		100
PET	69	4	0	1	94		99
34 Intrahepatic lesions from hepatocellular carcinoma in 10 patients
Ultrasound	19	12	2	4	61	33	76
CT	25	6	3	3	81	50	89
MRIa	24	3	4	2	89	67	92
PET	16	15	1	5	52	17	76
24 Extrahepatic tumor manifestation in 23 patients
Ultrasounda	4	10	1	3	29	25	25
CT [CE1] **	8	9	3	3	47	50	49
MRI [CE2] ***	2	3	1	1	40	50	22
PET	12	7	3	2	63	60	62

^a23 patients

PET missed 29 intrahepatic lesions. Fifteen were lesions from multifocal hepatocellular carcinoma, four were metastases from colorectal carcinoma (all ≤ 0.5 cm in size), six were metastases (<1 cm in size) from breast cancer, three were small metastases from gall bladder cancer (<0.5 cm), and, one was a metastasis (2 cm in size) from an ovarian cancer.

We found a total of six false positive intrahepatic lesions by PET scan (1× penetrating gall bladder empyema and 5× regenerative nodules in cirrhotic liver. Another two regenerative nodules were identified by isotope scan to have benign histology.

Sensitivity of PET was superior to abdominal ultrasound, CT-scan, and MRI-scan in detecting liver metastases from colorectal cancer (Table 1). On PET-scan there were four false negative hepatic lesions in two patients. One patient with colorectal cancer had two more metastases of left liver (0.5 cm) which had not been seen on PET-scan. The other patient with multiple liver metastases was prepared for hemihepatectomy; however, intraoperative ultrasound of the liver revealed two further metastases on the left side (<1 cm each).

The false positive result on PET was in a 44-year-old man following right-sided hemicolectomy for colonic cancer (pT₃N₀M₀) 1 year before. A lesion suspicious for a liver metastasis on PET scan proved to be gall bladder empyema at operation.

Thirty-four hepatic lesions were identified in ten patients with hepatocellular carcinoma (HCC). Sensitivity and specificity for PET in these patients was lower than for CT-scan and MRI-scan (Table 1).

A total of seven regenerative nodules were identified of which five were reported as having appearances consistent with tumor on PET scan. In the remaining two cases, FDG-PET correctly identified the lesion as being benign.

Extrahepatic tumor

In 23 of 50 patients, 24 extrahepatic lesions were identified (Table 1). In 12 patients, tumor was seen on FDG-PET scan that was either confirmed by operation or by additional conventional imaging of that region. Sensitivity and specificity of PET compared to abdominal ultrasound, CT-scan, and MRI-scan for all extrahepatic lesions was 63% and 60%, 29% and 25%, 47% and 50%, and 40% and 50%, respectively (Table 1).

Seven extrahepatic tumors were missed on FDG-PET (false negative). Only in one of these patients was tumor identifiable on CT-scan. This patient had multiple pulmonary metastases and the patient was not offered liver resection. The remaining six missed extrahepatic tumors were found during operation. Four of these six extrahepatic tumors that were resected at the time of liver resection The other two patients had unexpected multiple peritoneal metastases from colorectal cancer. In both cases the operation was limited to exploratory laparotomy only (Table 2).

Table 2.

Missed extrahepatic lesions on 18 FDG-PET

Primary tumor	Location of extrahepatic tumor	Maximum diameter of largest metastases
Colorectal cancer	Miliary peritoneal metastases	<0.5 cm
	Multiple pulmonary metastases	<0.5 cm
	10 Peritoneal metastases	1 cm
Gall bladder cancer	2 Lymph nodes liver hilus	1 cm
Ovarian cancer	5 Peritoneal metastases	3 cm
Breast cancer	4 Omental metastases	<1 cm
Renal cell carcinoma	Kidney	2 cm

In the presence of suspected extrahepatic tumor by other imaging modalities, true negative PET saved two patients from unnecessary extension of operation. The two patients had three tumors with true negative FDG-PET results which were confirmed intraoperatively. In one patient the operation revealed a haemangioma of the spleen and a benign tumor of kidney. The other patient had a bulging scar at the anastomosis following hemicolectomy which was misinterpreted as recurrence of colorectal cancer on CT-scan and endoluminal ultrasound (Table 3).

Table 3.

direct influence of PET on operation management

Previous history	Diagnosis	Initial intent	FDG-PET	Performed therapy
	(Before FDG-PET)		Founding	(After FDG-PET)
True positive FDG-PET result
SB, 68 years. Colorectal cancer, 5/99 right hemicolectomy, T3N1G1V0M0, 12/99 liver resection (atypical) segment 2	Liver metastases segment 3	Liver resection segment 3	Additional liver metastases segment 4a	Left hemihepatectomy
GW, 66 years. Colorectal cancer, 9/98 left hemicolectomy T3N1G2V0M0	Liver metastases segment 2,3 and 4	Left hemihepatectomy	Paraaortic lymphnode metastases	Left hemihepatectomy + resection of metastases
KG, 57 years. Colorectal cancer, 2/98 right hemicolectomy, T3N1M0, 1/99 left hemihepatectomy	Liver metastases segment 2	Liver resection segment 2/3	Paraaortic lymphnodes	Exploratory laparotomy
SG, 65 years. Colorectal cancer, 8/99 right hemicolectomy, T3N2G1 V0M0, 1/00 liver resection (atypical) segment 2	Liver metastases segment 5, 6 and 8 (ct segment 3 scar tissue)	Portal embolisation right+ right hemihepatectomy	Metastases liver segment 3	Chemotherapy
RB, 44 years. Colorectal cancer, 2/99 left hemicolectomy, T3N0G2V0M0	Axillary lymphnode	Lymphnode extirpation	Multiple liver metastases	Chemotherapy
MI, 57 years	Multilocal hepatocellular carcinoma	Liver resection	Retinal metastases	Chemotherapy
LB, 54 years. Colorectal cancer, 4/99 right hemicolectomy, T4N1M0	Liver metastases segment 6 and 7	Right hepatectomy	Metastases pelvis and lumbar spine	Chemotherapy
True negative FDGPET result
CB, 61 years.	Hepatocellular carcinoma segment 6 ct: tumor left kidney+spleen	Chemotherapy	No tumor in kidney and splen	Right hemihepatectomy
RH, 63 years. Colorectal cancer, 1996 right hemihepatectomy T3N1M0	Liver metastases segment 4+local recurrence at anastomosis	Liver resection + local resection	Liver metastases + no local recurrence	Left hemihepatectomy

False positive results occurred in two patients with enhanced extrahepatic glucose metabolism on PET scan. One patient had an inflamed lymph node on the left side of neck, the other suffered from silicosis which was described as pulmonary metastases on FDG-PET scan.

Influence on operation strategy

The findings on PET scan had a direct impact on operative management in nine patients (18%).

• True positive FDG-PET. In four patients a positive FDG-PET scan prevented unnecessary operations. Three of these, who all had a history of colorectal cancer, had unexpected additional tumor. In the fourth patient PET discovered retina metastases in a case of multilocular hepatocellular carcinoma. The radicalness of the planned liver resection was extended in two patients with liver metastases in colorectal cancer. One patient with extrahepatic disease on FDG-PET (aortic lymphnodes involved by tumor) was operated on, but was found to be unresectable.

• True negative FDG-PET. Three true negative FDG-PET findings (three tumors) in two patients had an direct impact on operative strategy (see extrahepatic tumor).

Discussion

Accurate preoperative evaluation in patients with primary and secondary liver tumors is essential for surgical management. Differentiation between malignant tumors and benign lesions is of critical importance. The standard non-invasive imaging modalities used in pre-operative staging are abdominal ultrasound, CT scan, and MRI scan. The excellent sensitivity of these modalities in detection of liver metastases or primary liver tumor has been well documented [3, 52, 56, 57, 58, 59, 60]. However, detection of extrahepatic lymph nodes, and differentiation of tumour recurrence from fibrosis and postoperative scar formation remains a diagnostic challenge.

PET is a new, important imaging technique in the diagnosis of malignancy and has the potential to greatly enhance preoperative staging of oncological patients. Sensitivity and specificity of FDG-PET scan have been shown to be greater than 90% for discriminating benign and malignant lesions in the liver. PET scan can also identify new lesions which are not recognised by standard imaging modalities [11, 12, 13, 14]. In contrast to conventional imaging techniques, FDG-PET allows evaluation of cell metabolism. It has been well established that increased glucose metabolism occurs in many tumors due to increased number of glucose transporter proteins and promotion of glycolysis in tumor cells by increased levels of hexokinase and phosphofructokinase [16, 17].

PET has been proven to be a valuable diagnostic tool in a number of different tumors but its direct influence on liver surgery has not been thoroughly investigated. Many studies have shown a significant increase of FDG in malignant liver lesions, whereas benign lesions have been shown to lack increased FDG uptake [19]. FDG-PET has been reported to be of use in imaging liver metastases from carcinomas of the lung, breast, ovary, colon, melanoma, and lymphoma [2, 3, 6, 20, 24]

About 40–50% of patients with colorectal cancer develop liver metastases, but in only 10% are these metastases are confined to the liver [4, 53, 54, 25]. Approximately 50% of patients with colorectal carcinoma die, mostly due to hepatic metastases [26, 27]. In a small group of these patients curative surgery can improve 5-year survival rate up to 40% [22, 24, 55], but this is applicable in only 10–20% of all patients [26]. Computed tomography as standard for evaluation of liver metastases is helpful but of limited use in the differentiation of postsurgical changes from tumor or evaluation of isodense hepatic lesions [15].

In the group as a whole, sensitivity of MRI was superior to sensitivity of FDG-PET (83% vs 75%). This results mainly from the low sensitivity of FDG-PET in hepatocellular carcinoma (Table 1).

Our results demonstrate that FDG-PET has a higher sensitivity in detection of liver metastases from colorectal carcinoma than MRI-scan, CT-scan or abdominal ultrasound. These findings confirm observations made by other authors [4, 50]. Several authors have reported decreased FDG accumulation during or after chemotherapy or chemoembolization and evaluated this effect as a means of monitoring response to treatment in colorectal liver metastases [25, 28, 19, 35, 36].

Some authors who exclude lesions with a size <1 cm found even higher sensitivity and specificity for FDG-PET in colorectal liver metastases [28]. The standard uptake value (SUV) cut-off level for differentiation of malignant from benign lesions is approximately 3.5 (mCi/weight in kg) for metastases >1 cm [3, 28] but differs between several tumors. We chose to focus on visual interpretation of FDG-PET.

Among patients with colorectal liver metastases, we had only four false negative lesions which were all between 0.5 and 1 cm in diameter. False negative lesions can be due to partial volume averaging, leading to the underestimation of the uptake in small lesions [3].

In the colorectal liver metastases group, we found only one false positive FDG-PET result in a patient with a previous history of colonic carcinoma (CEA negative) who had appearances consistent with a liver metastasis on ultrasound, CT-, and MRI-scan. This patient had no evidence on scan of any local inflammatory change. The patient underwent en bloc gall bladder extirpation including lymph node dissection of the hepato-gastric ligament and segment 5 resection. Subsequent histological examination revealed gall bladder empyema. Inflammatory lesions can also exhibit 18-FDG uptake, presumably due to activated macrophages, and may be mistaken for malignancies [34, 38].

Increased FDG-accumulation has been described in the literature with metastases from gall bladder, breast [40, 41], ovarian [42, 43, 44, 45], pharyngeal [39], and thyroid cancer [34], as well as different types of leiomyosarcoma [46, 47, 48, 49], but we had very limited numbers of these lesions during this study period.

As regards primary liver tumours, PET scan was less accurate in imaging hepatocellular carcinoma than MRI. Extrahepatic tumor manifestation is a contraindication to transplantation or extended liver resection in these patients, and accurate staging is essential to allow appropriate selection of treatment. The findings in this study of a sensitivity for FDG-PET in HCC patients of 52% compared with 89% for MRI scan (Table 1) are similar to results reported in other series. This reduced sensitivity may be due to similarity in glucose uptake between differentiated hepatocellular carcinoma cells and normal hepatocyte. Another factor may be the increased glucose-6-phosphatase activity in hepatocellular carcinoma. Differences in FDG uptake compared with surrounding normal liver tissue are only apparent with moderately or poorly differentiated HCC >5 cm or with grossly elevated AFP levels [21, 22, 23, 37]. False positive results were due to a variety of findings including local fibrosis, hepatic cholestasis, and a hepatic adenoma. The latter two instances have previously been reported as causes of false positive results [33].

By contrast, in all four patients with cholangiocarcinoma FDG-PET showed enhanced 18-FDG uptake with an increase in SUV. All four lesions were correctly identified as malignant tumors. Other authors confirm these finding [30].

A major advantage of FDG-PET over MRI-scan, CT-scan, and abdominal ultrasound was its sensitivity and specificity in the detection of unsuspected extrahepatic disease. This frequently has a major influence on surgical strategy (Table 1). The specificity of FDG-PET was also clearly superior to conventional imaging in this regard.

The false negative results in patients with previous colorectal primaries all occurred in patients with multiple small metastases in which the lesion size was below the limits of resolution of PET scan. Another three patients with a history of gallbladder cancer, ovarian cancer, and breast cancer had extensive intraperitoneal tumor spread. In one case only the missed metastasis, from an ovarian primary, was 3 cm in diameter.

Two false positive FDG-PET scans occurred in patients with a history of colorectal cancer. In the first case, PET scan showed a suspicious cervical lymph node after upper respiratory infection. In the second patient, who also had a previous history of gall bladder cancer, increased FDG-uptake was interpreted as bilateral pulmonary metastases. The patient was, however, known to suffer from silicosis and pulmonary metastases were excluded by thoracic CT-scan.

The additional information afforded by PET consequently had an influence on operative strategy in nine patients (18%). In seven patients FDG-PET discovered additional liver tumor or extrahepatic tumor spread. In another two patients suspected tumor could be excluded on FDG-PET (Table 3). In other words, the findings on PET directly changed the operative procedure in three patients and saved four patients from unnecessary operation.

These results in the whole group are similar to results described previously in the literature. Lai et al [6] found additional information with FDG-PET to conventional radiological imaging in 32% of patients with colorectal liver metastases. Operation management was influenced in 29% of patients. A further report by Beets et al. demonstrated that PET scan showed additional deposits of recurrent colrectal cancer that were undetected by CT-scan in 40% of patients [40].

Conclusion

Our series demonstrates good sensitivity and specificity for the detection of primary and secondary liver lesions which is superior to ultrasound and CT scan but not to MRI scan. The main value of PET consists in the detection of extrahepatic tumor (64%). Due to better detection of extrahepatic tumor FDG-PET is a very useful addition to the currently used anatomically-based images in all cases of advanced tumor spread with high risk of extrahepatic tumor. The findings on PET scan had a direct impact on operative management in nine patients (18%).

In the case of hepatocellular carcinoma, FDG-PET has a poor sensitivity and therefore adds little to preoperative management. FDG-PET is not able to evaluate anatomical resectability of liver tumors due to lack of visualisation of hepatic vasculature [6]. Because of false positive hepatic and extrahepatic FDG accumulation a preoperative fine needle biopsy is advisable in cases where tumour is detected solely by FDG-PET to avoid unnecessary resection.