Figure 1. Top, Illustration of the α-fetoprotein nested reverse-transcriptase polymerase chain reaction (RT-PCR). Arrow indicating the position of the primer sets for PCR (the first cycle) and nested PCR (the second cycle). bp indicates base pair. Bottom, Amplification products of first cycle (upper) and second cycle (lower). M indicates molecular-sized marker; lane 1, negative control; lane 2, positive control; lanes 3 and 4, patients.
Figure 2. Nested reverse-transcriptase polymerase chain reaction tests with α-fetoprotein–specific primers in peripheral blood of patients receiving curative resection of hepatocellular carcinoma. Lane M indicates molecular-sized marker; lane 1, negative control; lane 2, positive control; lanes 3 through 5, patients. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) indicates positive control.
Figure 3. Cumulative recurrence-free survival rates (Kaplan-Meier method) after curative resection of hepatocellular carcinoma (HCC) for patients being positive postoperatively for circulating α-fetoprotein messenger RNA (AFP mRNA) (solid square, n = 19) vs being negative (open square, n = 62) (compared by log-rank test, P<.001), and for patients being preoperatively positive for circulating AFP mRNA (solid circle, n = 22) vs being negative (open circle, n = 59) (P = .28).
Figure 4. The additive predictive role of postoperative circulating α-fetoprotein messenger RNA (AFP mRNA) to cumulative recurrence-free survival rates (Kaplan-Meier method) after curative resection for hepatocellular carcinoma (HCC). Patients positive for postoperative circulating AFP mRNA and positive for having pathologically invasive factors (solid square, n = 17) vs patients negative for postoperative circulating AFP mRNA and positive for having pathologically invasive factors (open square, n = 42) (P<.001). Patients being positive for postoperative circulating AFP mRNA and negative for having pathologically invasive factors (solid circle, n = 2) vs patients negative for postoperative circulating AFP mRNA and negative for having pathologically invasive factors (open circle, n = 20) (P<.001).
Jeng K, Sheen I, Tsai Y. Circulating Messenger RNA of α-FetoproteinA Possible Risk Factor of Recurrence After Resection of Hepatocellular Carcinoma. Arch Surg. 2004;139(10):1055-1060. doi:10.1001/archsurg.139.10.1055
Isolated tumor cells may be associated with micrometastasis. Circulating α-fetoprotein messenger RNA (AFP mRNA) in patients with hepatocellular carcinoma (HCC) has been considered to represent isolated tumor cells. We propose that circulating AFP mRNA may have prognostic value after curative resection of HCC.
A prospective cohort study.
Eighty-one consecutive patients who underwent curative resection of HCC.
A nested reverse-transcriptase polymerase chain reaction (RT-PCR) assay for circulating AFP mRNA before and again 12 weeks after surgery in 81 patients and in a control group. Clinicopathological variables and postoperative course (recurrence and recurrence-free survival) were examined for correlation with the levels of circulating AFP mRNA.
Of the 81 patients, 22 (27.2%) and 19 (23.4%) had AFP mRNA detected preoperatively and postoperatively, respectively. The recurrence-free survival at 1, 2, and 3 years after resection was significantly shorter in the latter patients (52.6%, 15.6%, and 0%, respectively; P<.001) but not in the former (81.8%, 54.5%, and 29.2%; P = .28). In univariate analysis, a significantly higher recurrence rate was found in patients with liver cirrhosis (P = .03), Edmondson-Steiner differentiation grade III or IV (P<.001), incomplete or absent capsule (P = .001), vascular invasion (P<.001), daughter nodules (P = .003), or a positivity for postoperative circulating AFP mRNA (P<.001). Postoperative positivity for circulating AFP mRNA remained a significant risk factor (P = .002; hazard ratio 3.13; 95% confidence interval, 1.52-6.47) in the multivariate analysis.
The detection of circulating AFP mRNA 12 weeks postoperatively is associated with an increased and earlier risk of HCC recurrence.
Quiz Ref IDHepatocelluar carcinoma (HCC) ranks fifth among malignancies worldwide and results in nearly 1 million deaths annually.1Quiz Ref IDIt is the leading cause of death in Taiwan.2,3 It is also one of the most common malignancies worldwide. After curative resection of HCC, tumors recur in about 70% of patients, usually within the first 3 years.2- 4 In addition, the 3-year recurrence after liver transplantation for small HCC is still as high as 25%.5 By the time the recurrence of HCC is detected by the current imaging studies, aggressive treatment is frequently futile either because of extensive recurrence in the liver or distant metastasis. The high recurrence rate after resection or even transplantation is the most important limit of long-term survival. The challenge is to more accurately stage the disease early so that the most effective treatment strategy can be chosen to prolong the disease-free survival.
Using molecular diagnostic tools to detect subclinical levels of residual leukemia (minimal residual disease) has changed the treatment concepts.6 Although there have been several clinicopathological factors reported to be useful in the prediction of earlier HCC recurrence after resection,7,8 there are few studies about whether molecular factors may also be one of the significant predictors in the multivariate analysis. Tumor cells may be shed into the circulation spontaneously or iatrogenically to form metastatic foci. The polymerase chain reaction (PCR) to identify tumor-specific gene transcripts (messenger RNA [mRNA]/circular DNA) is a sensitive method for detecting isolated tumor cells (ITCs) in a small blood sample.9,10 Detection of ITCs is increasingly being investigated as a prognostic indicator.11,12 α-Fetoprotein (AFP) mRNA is generally accepted as a liver-specific and tumor-specific marker.13- 17 Some authors have reported a higher risk of recurrence or metastasis in patients with circulating AFP mRNA,15- 18 but the results are conflicting.19 We designed this study to investigate the contribution of AFP mRNA in the prediction of tumor recurrence after curative resection.
There were 42 men and 39 women (mean ± SD age, 52.2 ± 13.3 years) of 89 consecutive patients with HCC undergoing curative hepatectomy in this prospective observational study (1995-1999). Patients who had had preoperative or postoperative alcohol injection or hepatic arterial chemoembolization, previous hepatectomy, or hospital complications were excluded. The curative surgical procedures included 62 major resections (6 extended right lobectomies, 18 right lobectomies, 14 left lobectomies, and 24 two segmentectomies) and 19 minor resections (11 segmentectomies, 5 subsegmentectomies, and 3 wedge resections). The demographic characteristics included the presence of liver cirrhosis (confirmed by the operative findings and also from the pathologic examination findings for the specimen) in 56; hepatitis B virus infection (hepatitis B surface antigen) in 62; hepatitis C virus infection (anti–hepatitis C virus) in 31; Child-Pugh classification of liver reserve (A vs B) in 70 vs 11; tumor size (≤3, 3-10, and >10 cm) in 25 vs 28 vs 28; cell differentiation grade (Edmondson-Steiner grade I or II vs III or IV) in 42 vs 39; encapsulation (complete vs incomplete or absent) in 28 vs 69; vascular permeation (vascular invasion and/or tumor thrombi within the portal or hepatic vein) in 37; and the presence of daughter nodules in 32.
Peripheral blood (PB) samples were obtained from a forearm vein 1 day prior to surgery (preoperative sample) and 12 weeks after surgery (postoperative sample) from all 81 patients. After discharge, patients with HCC had regular clinical assessment to detect tumor recurrence, including periodic abdominal ultrasonography (every 2-3 months during the first 5 years, then every 4-6 months thereafter) and serum AFP level testing and liver biochemistry tests (every 2 months during the first 2 years, then every 4 months during the following 3 years, and every 6 months thereafter). Abdominal computed tomography was done every 6 months during the first 5 years and then annually. Hepatic arteriography was performed if there was a suspicion of cancer recurrence on ultrasonography, computed tomographic scan, or when testing for serum AFP levels. Chest radiographs to detect pulmonary metastasis and bone scans to detect osseous metastases were undertaken every 6 months. The detection of tumors on any imaging study was defined as clinical recurrence. The recurrence-free interval was defined as the time from resection to the detection of clinical recurrence.
A control group including 30 healthy volunteers without liver disease (15 men, 15 women; mean age, 40 years) and 20 patients with chronic liver disease without evidence of HCC also had a PB sample from a forearm vein tested for AFP mRNA. No AFP mRNA was detected in them.
The sensitivity of our assay was determined using human hepatocytes. The sequence of the sense primers were 5′-TGC AGC CAA AGT GAA GAG GG-3′ (external sense) and 5′-TGC ACA CAA AAA GCC CAC TCC-3′ (internal sense), and those of the antisense primers 5′-CAA GCT GCT TTC TCT TAA TTC-3′(external antisense) and 5′-TAG CGA GCA GCC CAA AGA AGA-3′ (internal antisense). The size of the amplified AFP mRNA products was 315 and 177 base pairs (bp) by external and internal primer pairs, respectively (Figure 1).
The EDTA-treated whole blood was centrifuged and the plasma fraction removed. The cellular fraction was enriched for mononuclear cells or possible tumor cells according to the method described by Oppenheim.20 Total cellular RNA was then extracted using TRIzol Reagent (Life Technologies Inc, Gaithersburg, Md) according to the protocol from the manufacturer. Complementary DNA (cDNA) was synthesized from 5 µg of the mRNA. The reverse transcriptase reaction solution contained 6 µL of 5×first-strand buffer, 10mM dithiothreitol, 125mM each of deoxycytidine triphosphate (dCTP), deoxyadenosine triphosphate (dATP), deoxyguanosine triphosphate (dGTP), and deoxythymidine triphosphate (dTTP), 0.3 µg of random hexamers, and 200 U of Superscriptase II Moloney murine leukemia virus reverse transcriptase (Life Technologies Inc). The RNA solution was incubated at 95°C for 10 minutes, quickly chilled on ice, and then mixed with the reverse-transcriptase reaction solution (total volume, 20 µL), and incubated at 37°C for 60 minutes. The first PCR reaction solution contained 5 µL of the synthesized cDNA solution, 10 µL of 10×polymerase reaction buffer, 500µM each of dCTP, dATP, dGTP, and dTTP, 15 pmol of each external primer (external sense and external antisense), 4 U of Thermus Brockiamus Prozyme DNA polymerase (PROtech Technology Ent Co, Ltd, Taipei, Taiwan), and water. The PCR product was reamplified with internal primers. The final product was electrophoresed on 2% agarose gel and stained with ethidium bromide for the specific band of 101 bp.
We used EDTA-treated water (filtered and vaporized) as a negative control and Hep G2 (hepatoblastoma) cell line as a positive control for AFP mRNA expression. With cDNA derived from the Hep G2 cell line, specific bands for AFP (101 bp) were observed. In contrast, the cell lines that served as negative controls did not yield these bands. It was also impossible to detect free RNA extracted from 5 mL aliquots of blood from normal controls in which Hep G2 cells were suspended. The sensitivity of our assay, determined in a dilution experiment using freshly isolated human hepatocytes (105-101) in 1 mL of whole blood before RNA extraction, was approximately 1 hepatocyte for every 105 peripheral mononuclear cells.
The Kaplan-Meier method and log-rank test were used for the calculation and comparison of survival curves. The Cox proportional hazards model was used for multivariate stepwise analysis to identify factors predictive of recurrence and death. Significance was accepted at P<.05.
Quiz Ref IDIn the preoperative PB sample from patients with HCC, AFP mRNA was detected in 27.2% (22/81) (Figure 2). Nineteen patients (23.4%) had AFP mRNA detected in the postoperative sample. Eleven patients (13.6%) had AFP mRNA detectable in both samples. Thirty-six patients (44.4%) had clinically detectable recurrence during the follow-up period (median, 3 years; range, 2-5 years). All recurrences were intrahepatic when initially detected.
Quiz Ref IDPatients with a positive postoperative detection of AFP mRNA had a significantly lower recurrence-free survival than those negative for this postoperatively (Figure 3). (P<.001). However, no difference was observed between those with and without preoperative positivity for circulating AFP mRNA (P = .28). The cumulative probability of recurrence-free survival at 1, 2, and 3 years after curative resection was 52.6%, 15.6%, and 0%, respectively, for patients positive for postoperative circulating AFP mRNA; 87.1%, 70.8%, and 64.5% for patients negative for postoperative circulating AFP mRNA; 81.8%, 54.5%, and 29.2% for patients positive for preoperative circulating AFP mRNA; and 83%, 61.5%, and 54.6% for patients negative for preoperative circulating AFP mRNA.
We added preoperative and postoperative AFP mRNA to the clinicopathological characteristics to examine whether they were significant independent risk factors of recurrence (Table 1). In the univariate analysis, the presence of AFP mRNA preoperatively did not associate with the risk of recurrence (P = .47). Quiz Ref IDA significantly higher recurrence rate was found in patients with liver cirrhosis (P = .03; hazard ratio [HR], 2.49; 95% confidence interval [CI], 1.09-6.59), Edmondson-Steiner differentiation grade III or IV (P<.001; HR, 4.54; 95% CI, 2.21-9.34), incomplete or absent capsule (P = .001; HR, 4.13; 95% CI, 1.71-9.99), vascular invasion (P<.001; HR, 7.18; 95% CI, 3.30-15.64), daughter nodules (P = .003; HR, 2.64; 95% CI, 1.36-5.13), or positive for circulating AFP mRNA 12 weeks postoperatively (P<.001; HR, 4.68; 95% CI, 2.39-9.18).
In the multivariate analysis, after adjusting for other significant variables (Table 2), postoperative circulating AFP mRNA was still a significant risk factor (P = .002; HR 3.13; 95% CI, 1.52-6.47). Patients who had any one of Edmondson-Steiner grades III or IV, incomplete or absent capsules, vascular permeations, or daughter nodules were classified as having positive invasiveness. Those without any of them were classified as having negative invasiveness. In both groups, patients positive for postoperative circulating AFP mRNA had a significantly shorter recurrence-free survival (Figure 4) (P<.001). In the negative invasiveness group, the cumulative probability of recurrence-free survival at 1, 2, and 3 years after curative resection was 100%, 50%, and 0%, respectively, for patients postoperatively positive for circulating AFP mRNA, and 100%, 93.3%, and 93.3% for patients postoperatively negative for circulating AFP mRNA. In the positive invasiveness group, they were 52.9%, 7.9%, and 0% for patients positive for postoperative circulating AFP mRNA, and 85.4%, 63.4%, and 50.7% for patients negative for postoperative circulating AFP mRNA.
There have been correlative studies exploring the prognostic factors of HCC recurrence after curative resection.7,8 Among those clinicopathological factors, characteristics representing pathological invasiveness are the most consistent indicator of long-term prognosis. Detection of circulating HCC cells might be a real feature representing cancer invasiveness.10,18 We demonstrated that the presence of circulating HCC cells by assaying a positive AFP mRNA in the PB 12 weeks after curative resection of HCC is associated with an increased risk of recurrence and a shorter recurrence-free survival.
Using PCR to detect tumor cell DNA or RNA permits detection of 1 ITC in 105 to 106 normal cells. Studies in animal models with solid tumors have suggested that only 0.01% of circulating cancer cells eventually create a single metastatic deposit.21- 23 However, the absolute number of cells required for metastasis in the human circulation is still unclear.
Reverse-transcriptase polymerase chain reaction (RT-PCR) is a sensitive and versatile method to amplify cDNA generated by reverse transcriptase from mRNA. It is very helpful in the qualification or quantitation of the very small amount of RNA. However, RNA is less robust than DNA. After about 30 cycles of amplification, the yield visualized might not be clear and specific enough. Nested RT-PCR is developed to perform with 1 set of primers (the first round) and then with or without removal of reagents, to reamplify with an internally situated "nested" set of primers (the second round) (Figure 1). This process ensures that all products nonspecifically amplified in the first round will not be amplified in the second round, producing a yield that is specific and enhanced. The increased sensitivity and specificity of nested RT-PCR were readily apparent in patients with minimal residual disease.6
Circulating hepatocytes positive for circulating AFP mRNA have been reported in patients with acute hepatitis or acute exacerbation of chronic hepatitis, cholangiocarcinoma, liver metastases, other nonliver diseases, or even rarely in healthy persons.14,24- 26 However, Komeda et al10 and Okuda et al16 did not detect any AFP gene transcripts in patients with liver metastases or in healthy people. Wong et al27 quantitated the level of AFP mRNA in circulating liver cells and found HCC cells had a much higher level than nonmalignant cells. They concluded that the actual level of expression was useful for diagnosis. All 20 of our controls with chronic liver disease tested negative for AFP expression in PB, leading us to support the belief that the presence of AFP mRNA was specific for patients with HCC. Differences in the exact method of obtaining PB have been postulated to be one of the reasons for the discrepancies between various reports. If blood is drawn from a central line in or just proximal to the right atrium, it likely contains blood that has just left the liver without having passed through any capillary beds that may filter out circulating cells.17 This was why we sampled blood for analysis from the forearm.
A higher risk of HCC recurrence has been reported for patients with preoperative circulating AFP mRNA.16,18 However, only those who also persistently tested postoperatively positive for circulating AFP mRNA had a higher recurrence rate.16,18 Such results are consistent with our study that shows there is no correlation with preoperative circulating AFP mRNA. After adjusting for variables of invasiveness, postoperative circulating AFP mRNA still correlated significantly with recurrence. Wong et al27 further set a cutoff point (710 tumor cells) for those harboring a positive preoperative circulating AFP mRNA and a shorter recurrence-free survival.
Would the PCR techniques (nested RT-PCR vs RT-PCR, qualitative PCR vs quantitative PCR, semiquantitative PCR vs real-time PCR) influence the sensitivity or specificity of positive results? Most reports used nested RT-PCR with a good sensitivity, but the results still had a wide range. Few studies27- 29 used semiquantitative RT-PCT to investigate the dynamic change of circulating HCC cells. Wong et al28 found that surgical manipulation shed off nontumor hepatocytes, and circulating HCC cells increased gradually after operation. Only those with persistently large numbers of circulating HCCs were at risk of earlier recurrence.27,28 The advent of real-time PCR in recent years makes the quantitation of DNA/cDNA products easy work. Further investigation using real-time RT-PCR to add more information about quantity of circulating cancer cells should give us a more complete understanding.
The HCC tissue is surrounded by a vascular space analogous to the hepatic sinusoids. Because of this anatomic structure, tumor cells might easily be released into the sinusoids after detachment from a tumor focus after either spontaneous or iatrogenic detachment from a tumor, and these cells might migrate into the portal or hepatic vein and finally enter the systemic circulation. Whether micrometastasis occurs is determined by the ability of tumor emboli to survive in the circulation, attach firmly to the endothelium of distant organ capillaries, gain entrance to extravascular tissues, evade host defense mechanisms, and then grow into tumor colonies. Preoperative circulating AFP mRNA in PB may represent shedding of cells from primary HCC. However, the shedding process should cease after curative resection of HCC. Isolated tumor cells remaining at that time are probably destroyed immunologically. This may explain why preoperative AFP mRNA correlates poorly with tumor recurrence. The postoperative presence of ITC may therefore represent surviving malignancy that can continue the metastatic process.
Witzigmann et al19 and Lemoine et al24 did not find any correlation between postresection recurrence of HCC and the presence of circulating AFP mRNA, irrespective of whether it was measured before, during, or after surgery. Their patient numbers were small. Our study, along with those of other investigators,14,16,18 found that HCC recurrence was almost always in the liver. We attribute this to the fact that the liver microenvironment30 (eg, endothelial cells, extracellular matrices such as laminin, parenchymal cells, and growth-promoting factors) is best suited to the survival and secondary growth of HCC cells.
The best timing for postoperative blood sampling is still uncertain. Funaki et al14 and Okuda et al16 thought ITCs transiently liberated during surgery would be destroyed within 7 days. There are reports of samples drawn 7 days,18 1 month,28 1 to 15 months,19 and 6 to 15 months24 after resection. In the quantitative analysis of Wong et al28hematogenous dissemination of hepatocytes and tumor cells, hepatocytes were removed from the circulation within 8 weeks after operation, and HCC cells were cleared within 2 to 4 weeks. We thought 12 weeks were adequate to exclude the transient presence of HCC cells due to operative manipulation. The exact timing might influence the predictability for recurrence. Our finding of a correlation with the postoperative sample alone suggests that the AFP mRNA in PB needs to be sampled only once instead of twice.18
Molecular detection of circulating malignant cells is extending the concept of early systemic disease. The gene expression of AFP mRNA in tumor cells is heterogenous. In addition to it, p53 mRNA is another well-known marker for malignant transformation. However, not all HCCs result from the mutation of the p53 gene.31 If only cells that coexpress AFP mRNA and p53 mRNA could be considered circulating HCC cells, the specificity is increasing but the sensitivity is lowering.
We conclude that positivity for circulating AFP mRNA in PB 12 weeks postoperatively is correlated with an earlier recurrence and a shorter survival, independent of pathological invasive factors. This conclusion may therefore be of value in deciding whether to give adjuvant therapy after curative resection. However, until large, quantitative, statistically robust studies are completed with multivariate analysis of an array of potential prognostic factors, neither this nor other molecular or genetic markers should be used routinely to develop treatment recommendations in patients with resected hepatocellular carcinoma.
Correspondence: I-Shyan Sheen, MD, Liver Research Unit, Chang Gung Memorial Hospital, No. 199, Tung-Hwa North Road, Taipei, Taiwan (email@example.com).
Accepted for publication April 8, 2004.