AFP Response to Locoregional Therapy Can Stratify the Risk of Tumor Recurrence in HCC Patients after Living Donor Liver Transplantation

Simple Summary We evaluated the effect of AFP response to locoregional therapy (LRT) on the outcomes of hepatocellular carcinoma patients after living donor liver transplantation. The enrolled patients were divided into four groups according to LRT and AFP response to LRT. The nonresponse group had the highest 5-year cumulative recurrence rates whereas the complete-response group (patients with abnormal AFP before LRT and with normal AFP after LRT) had the lowest 5-year cumulative recurrence rates among the four groups. The 5-year cumulative recurrence rates of the partial-response group (AFP response was not back to the normal level) was comparable to the control group. AFP response to LRT can be used to stratify the risk of HCC recurrence after LDLT and also to clarify the efficacy of LRT. A better prognosis can be expected if a partial AFP response of over 15% is achieved. Abstract (1) Background: Alpha-fetoprotein (AFP) has been incorporated into the selection criteria of liver transplantation and been used to predict the outcome of hepatocellular carcinoma (HCC) recurrence. Locoregional therapy (LRT) is recommended for bridging or downstaging in HCC patients listed for liver transplantation. The aim of this study was to evaluate the effect of the AFP response to LRT on the outcomes of hepatocellular carcinoma patients after living donor liver transplantation (LDLT). (2) Methods: This retrospective study included 370 HCC LDLT recipients with pretransplant LRT from 2000 to 2016. The patients were divided into four groups according to AFP response to LRT. (3) Results: The nonresponse group had the worst 5-year cumulative recurrence rates whereas the complete-response group (patients with abnormal AFP before LRT and with normal AFP after LRT) had the best 5-year cumulative recurrence rate among the four groups. The 5-year cumulative recurrence rate of the partial-response group (AFP response was over 15% lower) was comparable to the control group. (4) Conclusions: AFP response to LRT can be used to stratify the risk of HCC recurrence after LDLT. If a partial AFP response of over 15% declineis achieved, a comparable result to the control can be expected.


Introduction
Hepatocellular carcinoma (HCC) is the sixth most commonly diagnosed cancer worldwide and the third most common cause of cancer-related deaths [1]. Liver transplantation (LT) is the best radical treatment option because it can remove the cancer as well as the entire pre-cancerous cirrhotic liver [2]. The Milan and University of California San Francisco

Patient Grouping
The serum AFP level was measured through radioimmunoassay (normal value: <20 ng/mL). One hundred and seventy-seven (47.8%) patients whose serum AFP levels were within the normal reference range (<20 ng/mL) before and after LRT were assigned to the control group. One hundred and ninety-three (52.2%) patients with abnormal serum AFP levels (>20 ng/mL) before LRT were stratified into the following groups according to the AFP response to LRT: complete AFP response (CR), partial AFP response (PR), and no AFP response (NR). The definition of CR was that AFP levels returned to the normal reference range (<20 ng/mL) after LRT. In patients with persistent abnormal AFP levels after LRT, the AFP decline was calculated by subtracting the pretransplant AFP level from the pre-LRT-maximal AFP level and dividing the result by the pre-LRT (or maximal) AFP level. The cutoff value of AFP decline between the PR and NR groups for tumor recurrence was determined by c-statistics using a receiver operating characteristic (ROC) curve.

Immunosuppression and Follow-Up Protocol
Basiliximab (Simulect; Novartis Pharma AG, Switzerland) was intravenously administered (20 mg) twice. Steroid therapy consisted of intraoperative intravenous methylprednisolone (500 mg), followed by a 20 mg/day dosage, which was tapered down and withdrawn after 3 months if no acute cellular rejection occurred. Patients with stable vital signs and renal function were given tacrolimus (Prograf; Fujisawa, Killorglin, Ireland) at a dose to maintain trough levels at 5-10 ng/mL during the first week after LDLT. Mycophenolate mofetil (CellCept; Roche, Ponce, Puerto Rico) was continuously administered at 0.5-1 g/day. Patients diagnosed with unfavorable tumor histology (such as poorly differentiated HCC, combined hepatocellular cholangiocarcinoma, or presence of macrovascular or microvascular invasion) or who were initially beyond UCSF criteria were given sirolimus (Rapamune; Pfizer, New York, NY, USA) at a dose that was maintained at 4-12 ng/mL. No adjuvant therapy was given to prevent recurrence. During the follow-up, the dosage of immunosuppressants was intentionally minimized if liver function was normal or stable. Follow-up visits in the outpatient clinic were scheduled on a monthly basis for the first year after transplantation and then every 3 months thereafter. Tumor recurrence was assessed by monthly measurements of AFP and liver ultrasound every 3 months. Computed tomography or magnetic resonance imaging were arranged when they were clinically indicated.

Statistical Analysis
Continuous variables were expressed as the mean ± standard deviation (SD) or as the median (interquartile range, IQR) if the data were not normally distributed. Student's t-test or the Mann-Whitney U test was used for comparisons between groups as appropriate. Analysis of variance or the Kruskal-Wallis test was used for comparisons among groups as appropriate. Categorical variables were expressed as percentages and were compared using the chi-square test or Fisher's exact test. The cumulative recurrence rate was calculated, and survival analysis was performed with the Kaplan-Meier method, and survival was compared between groups using the log-rank test. A Cox proportional hazards regression model (forward stepwise selection) was used to assess predictors of tumor recurrence and overall survival. Univariable and multivariable hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs) were calculated. Statistical significance was set at p-value < 0.05. ROC curve analysis was performed using a nonparametric method. All statistical analyses were performed using SPSS for Windows, version 22.0 (IBM, Armonk, NJ, USA).

Ethics Statement
The current study was approved by the Chang Gung Medical Foundation Institutional Review Board (approval number: 201701632A3). All methods were performed in accordance with the approved guidelines. Written informed consent was waived by the Chang Gung Medical Foundation Institutional Review Board due to the retrospective design of this study.

Patient Characteristics and Their Association with Recurrence
The 370 enrolled HCC patients were 295 men and 75 women, with a median age of 55.4 years. The median follow-up after LDLT was 85.9 months (IQR, 55.5-115.0 months). HCC recurred in 47 (12.7%) patients at a median of 15.2 months after LDLT (IQR, 7.1-30.7 months), of which 26 (55.3%) were extrahepatic metastases, and 4 (8.5%) were early extrahepatic metastases which was defined as within 6 months after transplantation. The causes of mortality were HCC-(n = 35, 9.5%), non-HCC-(n = 32, 8.2%), and surgery-related (n = 7, 1.9%). The patients' demographic characteristics and histopathological results of the explanted liver are shown in Table 1. Most patients were infected with hepatitis B and C viruses. Compared with the nonrecurrence group, the recurrence group had higher pretransplant AFP levels, more frequently underwent LRT, had higher tumor burdens, and had higher rates of microvascular invasion and viable tumors.

Cutoff Value of AFP Response to LRT
In patients with persistent abnormal AFP levels after LRT, the median AFP decline was 71.2% (IQR: 21.5-91.5%). To further stratify the risk of tumor recurrence with AFP decline, a ROC curve was used to determine the cutoff point between the PR and NR groups ( Figure 1A). AFP decline demonstrated acceptable discrimination for predicting tumor recurrence, with an area under the ROC curve of 0.641 (p = 0.05). With an optimal cutoff value of 15%, the sensitivity and specificity of predicting tumor recurrence were 65.0% and 67.4%, respectively. The scatter plot in Figure 1B shows the distribution of pre-LRT (or maximal) and pretransplant AFP levels in the LRT groups. Figure 2 shows the recurrence rates of the four groups based on pre-LRT AFP and AFP response. Of the 177 (11.9%) patients with normal pre-LRT serum AFP levels (control group), 21 experienced recurrence after LDLT. Of the patients with abnormal pre-LRT serum AFP levels, 84 had AFP levels that returned to normal (CR group) after LRT. Their recurrence rate (4/84, 4.8%) was significantly lower (p = 0.001) than those with persistent abnormal AFP levels (22/109, 20.2%). In patients with an AFP decline of more than 15% (PR group), 8 of 67 (11.9%) patients experienced recurrence after LDLT. In patients with an AFP decline less than 15% (NR group), their recurrence rate was significantly higher (p = 0.006) than the PR group. was 71.2% (IQR: 21.5-91.5%). To further stratify the risk of tumor recurrence with AFP decline, a ROC curve was used to determine the cutoff point between the PR and NR groups ( Figure 1A). AFP decline demonstrated acceptable discrimination for predicting tumor recurrence, with an area under the ROC curve of 0.641 (p = 0.05). With an optimal cutoff value of 15%, the sensitivity and specificity of predicting tumor recurrence were 65.0% and 67.4%, respectively. The scatter plot in Figure 1B shows the distribution of pre-LRT (or maximal) and pretransplant AFP levels in the LRT groups.   Figure 2 shows the recurrence rates of the four groups based on pre-LRT AFP and AFP response. Of the 177 (11.9%) patients with normal pre-LRT serum AFP levels (control group), 21 experienced recurrence after LDLT. Of the patients with abnormal pre-LRT serum AFP levels, 84 had AFP levels that returned to normal (CR group) after LRT. Their recurrence rate (4/84, 4.8%) was significantly lower (p = 0.001) than those with persistent abnormal AFP levels (22/109, 20.2%). In patients with an AFP decline of more than 15% (PR group), 8 of 67 (11.9%) patients experienced recurrence after LDLT. In patients with an AFP decline less than 15% (NR group), their recurrence rate was significantly higher (p = 0.006) than the PR group.  Table 2 shows the clinicopathologic characteristics of the four groups. The AFP-response groups had significant differences in microvascular invasion and American Joint  Table 2 shows the clinicopathologic characteristics of the four groups. The AFP-response groups had significant differences in microvascular invasion and American Joint Committee on Cancer (AJCC) T stage. The frequency of complete tumor necrosis was higher in the CR group; however, the AFP and pathologic response groups had only borderline significance. Regarding the correlation between the AFP-response groups and LRT, we found that the PR group had the highest percentage (41.8%) of pre-LRT-maximal AFP levels and the NR group had the lowest percentage (14.3%). The NR group also had the highest recurrence (p < 0.001) and mortality rates (p = 0.001) among the three groups. Consequently, the medium of recurrence months was significantly less than in the CR group (p = 0.004) ( Table 2).

Cumulative Recurrence Rate and Overall Survival
The 5-year cumulative recurrence rates in the control, CR, PR, and NR groups were 11.7%, 4.9%, 10.7%, and 33.4%, respectively ( Figure 3A). The CR group had excellent outcomes, while the NR group had the worst outcomes. The PR group obtained similar outcomes to the control group. The 5-year overall survival rates of the control, CR, PR, and NR groups were 81.5%, 94.0%, 88.0%, and 70.7%, respectively ( Figure 3B). The CR group had excellent outcomes, while the NR group had the worst outcomes.  Univariable analysis demonstrated that NR was an unfavorable factor (HR 3.213, 95% CI 1.633-6.322, p = 0.001) for tumor recurrence (Table 3) and overall survival (HR 1.873, 95% CI 1.044-3.361, p = 0.035). CR was a favorable factor (HR 0.496, 95% CI 0.247-0.995, p = 0.049) for overall survival (Table 4). In multivariable analyses, the groups, according to LRT number, AFP response to LRT, and largest tumor size, independently predicted tumor recurrence; moreover, AFP response to LRT and AJCC T stage predicted overall survival (Tables 3 and 4). Univariable analysis demonstrated that NR was an unfavorable factor (HR 3.213, 95% CI 1.633-6.322, p = 0.001) for tumor recurrence (  (Table 4). In multivariable analyses, the groups, according to LRT number, AFP response to LRT, and largest tumor size, independently predicted tumor recurrence; moreover, AFP response to LRT and AJCC T stage predicted overall survival (Tables 3 and 4).

Discussion
This study suggested that AFP response to LRT can be used to stratify the tumor recurrence risk after LDLT in HCC patients with abnormal pre-LRT AFP levels. Patients with AFP levels that returned to normal after LRT had excellent outcomes after LDLT. Patients with an AFP decline of more than 15% after LRT achieved comparable outcomes to those in the non-LRT and normal AFP groups. In contrast, patients with an AFP decline of less than 15% had the worst outcomes after LDLT. The pretransplant selection criteria for HCC patients could be refined with the consideration of AFP response to LRT.
High AFP was usually associated with a poor prognosis after LT. High AFP before transplant was a critical condition. Merani et al. [32] demonstrated that patients with an AFP level of >400 ng/mL at the time of listing who were downstaged to AFP ≤ 400 ng/mL had significantly better intent-to-treat survival than patients failing to show a reduction in AFP to <400 ng/mL. They concluded that downstaging HCC patients with high AFP was feasible and leads to similar survival to that of patients with persistently low AFP levels.
Bhat et al. [24] evaluated 35 HCC patients beyond the Milan criteria who received carboplatin-based TACE before LT and found that the percentage decrease in AFP was a significant predictor for survival and that patients with an AFP decrease exceeding 50% had significantly better median survival rates. Mehta et al. [33] had a similar finding focusing on a high level of AFP (>1000 ng/mL). Our study stratified patients into three groups according to their post-LRT AFP levels. Patients with AFP levels returning to normal were classified as the CR group and had excellent outcomes. The percentage decrease in AFP was only used in patients with persistent abnormal AFP levels after LRT, and the cutoff value of a 15% AFP decline was determined with a ROC curve. Noticeably, there were 25 patients who had high levels of AFP (>1000 ng/mL) before LRT. Four of them belonged to the NR group and the rest were either in the PR or CR group. The mortality and recurrence rates of the NR group were 38.1% and 33.3%, respectively, and of the control group were 21.5% and 11.9%, respectively.
Pretransplant LRT for HCC has been shown to be a strategy to improve long-term survival [34]. The concept of necrotizing therapy has been applied in our center, in which LRT was used as a neoadjuvant therapy to achieve complete or major tumor necrosis [15]. In our cohort, none of the 69 patients with pathological complete tumor necrosis experienced recurrence; however, it could only be recognized in the post-transplant explanted liver. Our results demonstrated that patients with a complete AFP response to LRT (abnormal AFP levels returning to normal levels after LRT) had the lowest 5-year cumulative recurrent rate (4.9%) and excellent 5-year overall survival (94.0%). The CR group could be used as a predictor for excellent outcomes before transplantation.
Our study demonstrated that patients with no AFP response to LRT (NR group) had the highest 5-year cumulative recurrence rate (33.4%) and worst 5-year overall survival (70.7%). AFP was a comprehensive biomarker of tumor behavior. AFP promotes the growth, proliferation, and metastasis of HCC and prevents apoptosis and the escape of HCC from immune surveillance [35]. A high AFP level has been shown to be associated with poorer outcomes [36,37]. If the serum AFP level fails to decrease after LRT, it might be due to technical or anatomic issues, aggressive tumor biology, or undetected tumors present in other parts of the liver or outside the liver. In this study, our patients had a whole-body bone scan and brain MRI to exclude the possibility of extrahepatic metastasis. Only one patient in the NR group had extrahepatic metastasis. Our data showed that the NR group more frequently had microvascular invasion. Therefore, NR indicated that tumors had a more aggressive behavior, and NR could be used as a predictor for tumor recurrence and poor outcomes.
Our study also showed that patients in the PR group with a partial AFP response to LRT had significantly better outcomes than those in the NR group. Compared with patients with normal AFP levels (control group), there were no significant differences in the 5-year cumulative recurrent rate and overall survival. In patients with cirrhotic liver and impaired liver function, comprehensive LRT is intolerable. One cycle of LRT resulting in a 15% decrease in the AFP level might be able to predict promising outcomes after LT.
In some previous studies, suboptimal results with high recurrence rates were seen in patients who had received LT with an AFP level of ≥400 ng/mL [38,39]. We further investigated the details of three AFP-response groups with LRT. Notably, only 14.3% of maximal AFP in the NR group was higher than 400 ng/mL. The PR group had the highest proportion of maximal AFP among the three groups. This result indicated that a high level of AFP was not a main indicator of recurrence in this study. AFP response could be another indicator to predict the risk of recurrence. If AFP did not respond to the LRT, the prognosis was worse than for PR even if the AFP was still low. Nevertheless, if AFP could partially respond to LRT, even decreasing by only 15% to the maximal AFP, it could improve the outcome after LT. Recent studies suggested that the change in AFP while on the waiting list could best predict survival, with cutoff values of 200, 50, or 15 ng/mL/month in patients without downstaging or bridging LRT [40][41][42]. However, in our center, downstaging or bridging LRT was an important method to help patients fitting the national criteria. This study also showed that the AFP response could be used to distinguish the efficacy of LRT. The dynamic change in AFP was indicative of the tumor's biological behavior, which was complicated. In patients with pretransplant LRT, a single AFP level could not be used for evaluating tumor aggressiveness. Therefore, AFP response to LRT was an alternative method for evaluating tumor aggressiveness and could be used to stratify the risk of tumor recurrence in HCC patients after LDLT. In this retrospective study, we simplified two points of AFP level to evaluate the prognosis. It was not only more accurate than a single point but also simpler than dynamic AFP levels. In a medical application, if LRT produces a complete or partial response before LDLT in abnormal AFP patients, a better prognosis can be expected. Although the survival rate in the no AFP response group was higher than that of most patients receiving hepatectomy or ablation, obviously the inferior survival compared with the AFP-response groups will change the post-transplant treatment policy. For instance, post-transplant management such as immunosuppressant modification and the indication of adjuvant therapy by target therapy (study on-going) to reduce recurrence are completely different. By the pretransplant AFP response from our study, we can stratify the patients for different post-transplant protocols [43].
In real-world practice, patients with high AFP (>1000 ng/mL) would be normally exclude by some centers. In our LRT groups, there were 11 patients (13.4%) in the CR group, 10 (14.9%) in the PR group, and 4 (9.5%) in the NR group whose maximal AFP was beyond 1000 ng/mL. The proportion in each group was similar, but the outcomes were very different.
The present study had some limitations. First, this study was a single-center retrospective study with a small sample size. Second, there was a lack of data on HCC-related dropouts from the waiting list and patient selection bias. In our center, infiltrative tumors or those accompanied by sky-high AFP as contraindication criteria for a downstaging procedure are excluded in the first place. Third, the pretransplant LRT method was not standardized. The choice of TACE, RFA, PEI, or their combined use depended on the patient's clinical condition.

Conclusions
This study showed that AFP response to LRT was a predictor for tumor recurrence in HCC after LDLT. It was an easy and feasible method to stratify the risk of tumor recurrence. The CR group had better outcomes than the control group of patients with normal AFP levels. The NR group had the worst outcomes, whereas the outcomes of the PR group were comparable to those of the control group. These results might encourage patients who planned to undergo LT but could not reach a complete AFP response or had a high level of AFP in the first place. The patients who had high AFP (>400 ng/mL) might consider undergoing LRT and could still have the opportunity to get a better outcome. If they still do not have a partial AFP response after LRT, the surgeon can monitor the condition of these patients to treat early when recurrence occurs.