Orthotopic liver transplantation (OLT) has become widely accepted as the treatment of choice for early hepatocellular carcinoma (HCC). Among appropriately selected candidates, OLT for HCC provides excellent results, with overall survival rates exceeding 70%.1–3 Mazzaferro et al.1 reported that 48 patients with a single tumor ≤5 cm in diameter or with ≤3 tumors all ≤3 cm in diameter displayed survival rates comparable to those of non-HCC liver transplant recipients. These Milan criteria (MC) are currently widely accepted as an effective way of selecting patients with early-stage HCC for curative OLT.
Conversely, because the MC were originally aimed at predicting good outcomes with low rates of recurrence, rather than poor outcomes with high recurrence rates, a substantial subset exists with the potential for good outcome after OLT among patients with HCC beyond the MC. In fact, many centers have performed OLT using extended criteria beyond the MC, showing acceptable overall and recurrence-free outcomes.3–9 In particular, various rationales support the expansion of criteria for living donor liver transplantation (LDLT). First, in contrast to deceased donor liver transplantation (DDLT), which uses a scarce public resource subject to an equitable allocation system, living donor liver grafts are dedicated to related recipients. The acceptable recurrence rate is not absolute but rather depends on donor organ availability, and the value of LDLT for the patient and donor is judged on the basis of donor safety, risk of recurrence, and probable outcomes without transplantation. In most cases, the donor wants to donate even if the expected benefit for the recipient is not the same as that for non-HCC cases. Second, the waiting time between preoperative tumor staging and transplantation is usually shorter for LDLT than for DDLT. Furthermore, because of recent progress in imaging technology, such as multidetector computed tomography, the accuracy of imaging studies has been increased.10 Given these improvements, the wide safety margins offered by the MC for tumor stage underestimation by preoperative imaging studies can be diminished.11
The issue of concern is thus how to preoperatively identify categories beyond the MC that still offer predictably good outcomes. Although several studies have proposed extended criteria based on tumor number and size as an estimate of tumor burden, these are considered insufficient to stratify patients in such categories, and additional parameters for tumor biological features related to risk of recurrence are necessary.12 Moreover, for patient selection criteria before LDLT, only tumor variables available in preoperative evaluation should be incorporated.
We have previously reported the results of LDLT for 93 patients with HCC and have shown that recurrence rates are significantly higher for patients exceeding the MC than for patients meeting these criteria.13 To determine optimal criteria expanded beyond the MC, the present study further analyzed risk factors for recurrence, including serum tumor markers, in an extended series of 125 patients.
AFP, α-fetoprotein; DDLT, deceased donor liver transplantation; GRWR, graft-to-recipient body weight ratio; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; LDLT, living donor liver transplantation; MC, Milan criteria; MELD, Model for End-Stage Liver Disease; OLT, orthotopic liver transplantation; PBC, primary biliary cirrhosis; PIVKA-II, protein induced by vitamin K absence or antagonist-II.
PATIENTS AND METHODS
Patients and Surgery
Between February 1999 and June 2006, a total of 125 patients (89 men and 36 women) who were diagnosed with HCC by preoperative imaging studies underwent LDLT at Kyoto University. Patients with incidental tumors were excluded from this study. Our institutional selection criteria for LDLT include (1) HCC unsuitable for resection or local ablation therapies and (2) extrahepatic metastasis or macroscopic venous invasion excluded on preoperative imaging. No restrictions were placed on the number or size of tumors.13
Patient profiles and associated clinical characteristics are shown in Table 1. The median age of the recipients was 55 years (range, 22-69 years). The median Model for the End-Stage Liver Disease score was 15 (range, 6-43). A total of 66 patients (53%) were hepatitis C virus antibody–positive, and 42 (34%) were hepatitis B surface antigen–positive, with 5 patients both hepatitis B surface antigen–positive and hepatitis C virus antibody–positive. Of the 125 patients, 94 patients (75%) displayed a history of previous treatment for HCC using various nontransplant methods, including transarterial chemoembolization (n = 82), percutaneous ethanol injection or radiofrequency ablation (n = 66), and hepatic resection (n = 14). In Japan, given the disadvantages of LDLT, such as the risks to the live donor and higher perioperative morbidity and mortality in comparison with other treatment modalities, the argument has been made that, if suitable for resection or ablation therapy, HCC should initially be managed with these conventional methods, with LDLT held in reserve as a second-line option.4, 13 Therefore, these pretransplant treatments were performed not as a bridge to transplant or for downstaging but with intent for curative ablation. Thereafter, when the patients suffered from uncontrolled recurrent HCC, they were referred to our department for LDLT as a second-line treatment. For the remaining 31 patients (25%) who had no history of HCC treatment before LDLT, conventional treatment was not indicated because of advanced liver dysfunction in most cases.
Table 1. Preoperative Profile and Clinical Characteristics
Abbreviations: GRWR, graft-to-recipient body weight ratio; HBV, hepatitis B virus; HCV, hepatitis C virus; MELD, Model for End-Stage Liver Disease; PBC, primary biliary cirrhosis.
An evaluation of the extent of tumor involvement using contrast-enhanced computed tomography was usually performed within 1 month before LDLT. Tumor staging was determined by the counting of only viable and enhancing nodules on computed tomography. For patients with previous treatment, resected tumors or nodules that were judged as nonviable were not counted. A total of 70 patients (56%) met the MC according to pretransplant imaging, whereas 55 patients did not.
Of the 94 patients who had a history of previous nontransplant treatment, records of tumor stage at first diagnosis of HCC were available for 78 patients: 59 patients met the MC, whereas 19 patients did not at that time. The median period between first diagnosis of HCC and LDLT was 23 months (range, 5-168 months) in these 78 patients. At the time of LDLT, the tumor stage was upstaged in 24 patients, downstaged in 6, and unchanged in 48 according to the MC.
LDLT was performed with a right lobe graft for 123 patients and a left lobe graft for 2 patients. Operative procedures for donor and recipient surgery have been described elsewhere.14, 15 The donor group comprised 63 men and 62 women, with a median age of 41 years (range, 19-64 years). The median graft-to-recipient body weight ratio was 1.04 (range, 0.49-1.69). ABO blood type matching was incompatible in 17 cases.
The standard immunosuppression protocol involved tacrolimus and low-dose steroid administration.13, 16 Patients who received ABO blood type–incompatible transplants were treated with preoperative plasma exchange or double filtration plasmapheresis to reduce the anti-A or anti-B antibody titer. During the first 3 postoperative weeks, prostaglandin E1 and additional steroids were administered via the portal vein or hepatic artery.17 As of the end of February 2007, the median follow-up period was 29 months (range, 1-96 months).
Data are presented as median and range. Statistical analysis was performed with the Fisher exact or χ2 test and analysis of variance. Cumulative probability curves of survival or HCC recurrence were calculated with Kaplan-Meier methods, and differences between curves were compared with the log-rank test. Cox's proportional hazard regression analysis was used for multivariate analysis. Any variable identified as significant (P < 0.05) in univariate analysis with the log-rank test was considered a candidate for multivariate analysis. All statistical analyses were performed with the StatView 5 statistical software package (Abacus Concepts, Berkeley, CA).
Overall Patient Survival and Recurrence of HCC
A total of 88 patients remained alive as of the end of February 2007, with 14 patients dying of recurrent HCC and 23 patients dying of tumor-unrelated causes, mainly infectious complications. The overall patient survival rate at 5 years was 68.3%. The survival rates did not differ significantly between the within-MC and over-MC groups at 72.4% and 64.0%, respectively (P = 0.6649).
Postoperative recurrence of HCC occurred in 20 patients and occurred within 2 years after LDLT in 17 (85%) of these patients. The first sites of recurrence comprised the lung (n = 6), bone (n = 5), graft liver (n = 3), abdominal lymph nodes (n = 3), adrenal gland (n = 1), brain (n = 1), and right subphrenic space (n = 1). After the exclusion of deaths without recurrence, the overall cumulative recurrence rate was 21.2% at 5 years. The recurrence rate at 5 years was significantly higher for patients who exceeded the MC than for those who met these criteria (34.3% versus 9.7%, respectively; P = 0.0006).
None of the recipient and donor variables listed in Table 1, including the Model for End-Stage Liver Disease score, underlying hepatitis, history of previous treatment, graft-to-recipient body weight ratio, and ABO blood type compatibility, were significantly associated with postoperative survival or recurrence (data not shown).
Preoperative Tumor Variables and Recurrence Rate
Univariate analysis was performed to evaluate associations between preoperative tumor variables and risk of recurrence. The preoperative tumor variables significantly associated with postoperative recurrence are shown in Table 2. When the recurrence rate was analyzed on the basis of tumor number and size separately, patients with 4-10 tumors exhibited 5-year recurrence rates similar to those of patients with ≤3 tumors (12.8% versus 17.6%). However, the recurrence rate for patients with ≥11 tumors (56.8%) was significantly higher than that for those with fewer tumors (P < 0.0001). In addition, patients with tumors >5 cm in diameter showed a significantly higher recurrence rate than those with smaller tumors. Accordingly, patients were divided into 3 groups: patients who met the MC (n = 70); patients beyond the MC but with ≤10 tumors all ≤5 cm in diameter (n = 30); and patients with ≥11 tumors, any tumor >5 cm in diameter, or both (n = 25). Tumor number and maximal size for each patient in the second and third groups (beyond-MC groups) are shown in Fig. 1. In the second group, 2 patients died of HCC-unrelated causes 1 and 5 months after LDLT, and another 2 patients developed HCC recurrence 3 and 26 months after LDLT. The remaining 26 patients were alive without recurrence at the time of writing, and their median follow-up was 48 months (range, 9-91 months). In the third group, the median tumor number and maximal size were 13 (range, 1-32) and 5.2 cm (range, 0.8-18 cm), respectively. In this group, 5 patients died of HCC-unrelated causes between 1 and 11 months (median, 1 month) after LDLT, and 14 patients developed HCC recurrence between 1 and 30 months (median, 8 months) after LDLT. The remaining 6 patients were alive without recurrence at the time of writing, and their median follow-up was 29 months (range, 16-68 months). A comparison of the recurrence rates among the 3 groups (Fig. 2) showed that the 5-year recurrence rates were similar for the first and second groups (9.7% versus 7.3%) but that the rate for the third group was significantly higher than those for the other groups (73.9%, P < 0.0001). As preoperative tumor markers, we evaluated serum α-fetoprotein (AFP) and protein induced by vitamin K absence or antagonist-II (PIVKA-II) or desgamma-carboxy prothrombin.18, 19 An AFP level >400 ng/mL and PIVKA-II level >400 mAU/mL represented significant predictors of a higher recurrence rate (Table 2).
Table 2. Univariate Analysis of Preoperative Tumor Factors and Recurrence
5-Year Recurrence Rate (%)
Abbreviations: AFP, α-fetoprotein; MC, Milan criteria; PIVKA-II, protein induced by vitamin K absence or antagonist-II.
Among the preoperative tumor variables, including ≥11 tumors, maximal tumor diameter >5 cm, MC, AFP level >400 ng/mL, and PIVKA-II level >400 mAU/mL, multivariate analysis revealed that ≥11 tumors, tumor diameter >5 cm, and PIVKA-II >400 mAU/mL represented independent risk factors for postoperative recurrence (Table 3). We therefore defined new and extended criteria to minimize the risk of tumor recurrence as HCC with ≤10 tumors all ≤5 cm in diameter and PIVKA-II ≤400 mAU/mL. The 5-year recurrence rate for the 78 patients who met the new criteria (4.9%) was significantly lower than that for the 40 patients who exceeded them (60.5%, P < 0.0001; Fig. 3). Similarly, patients who met the new criteria showed a significantly better 5-year survival rate (86.7%) than those who did not (34.4%, P < 0.0001; Fig. 4).
Table 3. Multivariate Analysis of Preoperative Tumor Factors and Recurrence
95% Confidence Interval
Abbreviations: AFP, α-fetoprotein; MC, Milan criteria; PIVKA-II, protein induced by vitamin K absence or antagonist-II.
Tumor number ≥11 nodules
Tumor diameter >5 cm
AFP >400 ng/mL
PIVKA-II >400 mAU/mL
Explant Pathology and Recurrence Rates
The correspondence of preoperative staging with the results of pathological analysis in terms of the 10-5 rule (number ≤10 and maximal size ≤5 cm) was evaluated in postoperative examinations of explanted livers. Foci completely ablated (100% necrosis) by pretransplant treatments were not counted as tumors. On pathological examination, 94 patients met the so-called pathological 10-5 rule, whereas 31 did not. Accordingly, preoperative imaging studies underestimated diagnosis in 10 patients (8%) and overestimated it in 4 patients (3%). The remaining 89% of patients had been correctly classified.
Univariate analysis of pathological findings revealed that the 5-year recurrence rates for the 94 patients who met the pathological 10-5 rule (9.5%) were significantly lower than the rates for the 31 patients who exceeded the pathological 10-5 rule (55.2%, P < 0.0001). Positive microvascular invasion and tumor grade of poor differentiation were also significantly associated with postoperative recurrence (Table 4).
Table 4. Univariate Analysis of Pathological Variables in Explant and Recurrence Rate
5-Year Recurrence Rate (%)
Number and size
n ≤ 10 and ≤5 cm
n ≥ 11 or >5 cm
Well or moderate
Effects of Preoperative PIVKA-II Value
Out of 100 patients who met the 10-5 rule on the basis of preoperative imaging, postoperative recurrence occurred in 6 patients (Table 2). Four of these 6 patients displayed preoperative PIVKA-II >400 mAU/mL. Therefore, patients who met the 10-5 rule but displayed PIVKA-II >400 mAU/mL showed a significantly higher 5-year recurrence rate (n = 16, 33.4%) than those within the 10-5 rule and with PIVKA-II ≤400 mAU/mL (n = 78, 4.9%; P < 0.0001). On the other hand, of the patients who met the 10-5 rule, the recurrence rates did not differ between those with AFP ≤400 ng/mL (n = 87, 8.1%) and those with AFP >400 ng/mL (n = 13, 16.7%; P = 0.1632).
Table 5 shows the relationship between pathological findings and preoperative tumor factors. The positive rate of microvascular invasion (66.7% versus 30.0%, P = 0.0013) and incidence of tumor grade of poor differentiation (40.7% versus 15.5%, P = 0.0113) were significantly higher for patients with PIVKA-II >400 mAU/mL than for those with PIVKA-II ≤400 mAU/mL. The 10-5 rule and AFP value were also significantly associated with these pathological findings.
Table 5. Relationship Between Microvascular Invasion or Tumor Grade (Poor Differentiation) and Preoperative Tumor Factors
Abbreviations: AFP, α-fetoprotein; PIVKA-II, protein induced by vitamin K absence or antagonist-II.
Determined by χ2 analysis.
n ≤ 10 and ≤5 cm
n ≥ 11 or >5 cm
In the present study, the recurrence rate was significantly higher for patients who exceeded the MC than for those within the MC on the basis of preoperative imaging studies, and this is consistent with previous findings.13 However, when over-MC patients were divided into 2 subgroups, patients who exceeded the MC but presented with ≤10 tumors all ≤5 cm displayed recurrence rates comparable to those who met the MC. The cutoff of tumor diameter ≤5 cm seems acceptable, as similar results have been reported in several studies.3, 4, 20, 21 Conversely, the limit of ≤10 tumors was deduced from the present analysis in which none of the 5 patients with 10 tumors developed recurrence whereas 1 of the 3 patients with 11 tumors developed recurrence and another died of HCC-unrelated complications (Fig. 1). Although some studies have reported similar results, that multinodularity (≥4 tumors) is unrelated to recurrence,21, 22 the safety of extending this to ≤10 tumors has not yet been definitively proven, and confirmation in more cases is required. However, the feasibility of the 10-5 rule is supported by the result that patients within the 10-5 rule, as determined by pathological examination of the explanted livers, demonstrated significantly lower 5-year recurrence rates (9.5%) than patients with more advanced tumors (55.2%, P < 0.0001; Table 4).
The limitations of imaging studies may represent a barrier to expanding the criteria for OLT in patients with HCC11 because the rate of underestimation of preoperative tumor staging based on imaging studies is reportedly 20%-40%.1, 8, 23, 24 However, recent progress in imaging technology has increased the accuracy of preoperative diagnosis of HCC.10, 25, 26 At our institute, since multidetector computed tomography was introduced for preoperative tumor staging in 2002, the detectability of HCC nodules has been considerably increased, so the accuracy of diagnosis reaches nearly 90% (data not shown). In this study, the rate of underestimation in terms of the 10-5 rule was only 8%, and 89% of patients were correctly diagnosed by preoperative imaging. These results support findings advocating the expansion of criteria.
To explore extended criteria that are able to reasonably predict risk of recurrence, preoperative markers of biological behavior in addition to tumor number and size are needed. As tumor pathologies such as histological grade and microvascular invasion are important prognostic markers, Cillo et al.27 adopted pretransplant tumor grading and excluded poorly differentiated HCC cases. Although the results were satisfactory, most surgeons hesitate to perform preoperative needle biopsy because of the risk of tumor seeding. Otto et al.28 recently proposed response to transarterial chemoembolization during waiting time as a biological selection criterion in DDLT. This approach is not applicable in LDLT because patients need not wait so long if a living donor is available. In addition, LDLT is offered in most cases when these treatments fail to control HCC or are not indicated because of advanced liver dysfunction.
PIVKA-II (or desgamma-carboxy prothrombin) and AFP represent useful tumor markers for HCC. PIVKA-II is known to be superior to AFP in specificity, and the plasma PIVKA-II level does not correlate with tumor size or AFP level.18 Among various tumor variables, PIVKA-II is reportedly the most significant predisposing factor for development of portal venous invasion in HCC patients.19 In the present study, both markers represented significant predictors of a higher recurrence rate by univariate analysis. However, multivariate analysis revealed only PIVKA-II as an independent risk factor for recurrence. This indicates a significant association between PIVKA-II and recurrence, regardless of tumor size or number. In contrast, there appeared to be some correlation between AFP value and tumor size or number, and we believe that AFP >400 ng/mL cannot predict higher recurrence rates in patients with fewer or smaller tumors (within the 10-5 rule). Furthermore, PIVKA-II and AFP were closely related to pathological tumor features such as microvascular invasion or histological tumor grade. All these results suggest that the preoperative PIVKA-II level offers additional information on tumor biological behavior related to histological findings usually obtained by explant pathology and represents a useful predictor of recurrence in conjunction with tumor size and number. Although the system for preoperative tumor staging is different between DDLT and our LDLT program, it is implied that these effects of tumor markers may also be helpful for allocation policy in DDLT.
In conclusion, this study suggests that the extension of tumor size and number limits is feasible, and by the incorporation of the preoperative PIVKA-II value, the false-positive rate for predicting a better subgroup outside the MC is decreased. The proposed new criteria are expected to serve efficiently as expanded selection criteria for LDLT in patients with HCC.