Blood indices and circulating tumor cells for predicting metastasis of hepatocellular carcinoma after liver transplantation

Introduction

Hepatocellular carcinoma (HCC) is a devastating malignant tumor, as one of the top common cancer and a primary cause of cancer-related mortality. Among all viable therapeutic options for HCC, liver transplantation (LT) can benefit the long-term survival of HCC patients, with satisfactory results of 1-, 5-, and 10-year survivals [1], for it addresses both HCC and underlying primary liver diseases (the major cause of HCC) at the same time. However, it is widely noticed that tumor may recur after LT, especially for patients with multinodular and large tumors [2]. Different criteria (e.g., Milan criteria) were established to distinguish who can benefit from LT, but there is insufficiency in metastasis prediction. Metastasis causes more than 90% of cancer mortality. Therefore, post-LT monitoring of metastasis merits attention in clinical practice. Many novel techniques have been proposed to estimate the recurrence and metastasis of HCC [3], and the forefront methods take advantages of the circulating tumor cell (CTC). Generally, presence of CTC implies metastasis to extra-hepatic organs. Several studies have claimed that the quantification of CTCs is an emerging tool for diagnosing, stratifying, and assessing extra-hepatic metastasis in HCC patients [4]. But for post-LT monitoring in HCC patients, the clinical value of CTC, as well as CTC subtypes, remains unclear. We hypothesized that CTC levels and subtypes are intimately associated with metastasis status of HCC patients, and this study was designed to test that available hematological indices including CTC data can provide sensitive and accurate prediction of post-LT metastasis.

Materials and methods

Results

Prediction of two-year metastasis

Based on above analysis, three variables had the closest relationship with two-year metastasis: mCTC45d, the AFP level at the fourth month (AFP4M), and the PIVKA-Ⅱ level at the third month (PIVKA-Ⅱ3M). Using the logistic regression (two-ends), we named a variable MP (metastasis predictor), and it was calculated by above three factors using the logistic regression. As the ROC curve shown (Figure 1), the AUC of MP was 0.858, with a cutoff value of 0.328 (sensitivity=0.75, specificity=0.97).

Figure 1:

Receiver operating characteristic (ROC) curve for efficacy analysis of risk factors in post-LT metastasis in two years.

The variable MP is calculated by the two-ends logistic regression model, based on variables including mesenchymal CTC level at the 45th day (mCTC45d), the AFP level at the fourth month (AFP4M), and the PIVKA-Ⅱ level at the third month (PIVKA-II3M).

Discussion

There have been a few widely known predictive factors for survival and tumor progression in HCC patients with LT, such as cumulative tumor size [5], histological grade of differentiation [6], vascular invasion [7], lymph node involvement [8], higher TNM stages [9], and high dosage of cyclosporine [10]. In particular, microvascular invasion, macrovascular invasion, tumor number, and total tumor diameter were the strongest prognostic factors in Chinese patients [11]. In the aspect of postoperative monitoring, different hematological indices have also been investigated (e.g., platelet counts [12], CD133 and CD90 [3]). Moreover, some prophylactic treatments (e.g., Licartin administration, Sorafenib treatment) are helpful for blocking HCC recurrence or metastasis [13]. Theoretically, CTCs may serve as a liquid biopsy for metastatic tumors, with adequate and unique advantages, that the collection of peripheral blood is easy, rapid, lowly invasive, cost-effective, and feasible for serial real-time monitoring. Currently, the index CTC (especially mCTC) has seldom used in monitoring metastasis or recurrence after LT. We here for the first time found that combination of CTC/AFP/PIVKA-Ⅱ can effectively indicate the later metastasis. In particular, hematological indexes from day 45 to 4 months are useful references in prediction of LT metastasis.

However, we failed to find clear links between post-LT metastasis and traditional factors, such as Milan criteria, Child-Pugh-Turcotte classification, lymph node involvement, cirrhosis background or age [14]. Also, prophylactic treatments after surgery did not show significant benefits in our work. This may be due to a limited sample size. Overall, the positive results we discovered are consistent with most published research. Similar to the study of Zavaglia [6], we also noticed microvascular invasion classification influenced the postoperative metastasis. Nevertheless, this variable has a relatively lower association (p=0.045) in comparison with postoperative hematological indices. AFP above 100 or 200 ng/mL indicates poorer outcomes, as reported by other independent studies [15, 16]. We also noticed that AFP monitoring helps reflect the postoperative progression of HCC. Especially, the value at the time point of the fourth month is more valuable. PIVKA-II is an important biomarker for HCC surveillance in conjunction with AFP [17]. AFP and PIVKA-II are associate with giant lymph node metastasis in hepatoid gastric cancer [18]. Some study even claimed that the AFP level, PIVKA-II level, portal vein invasion were the most important prognostic factors for HCC patients [19]. Ultra-high PIVKA-II level is correlated with hepatic portal lymph node metastasis [20]. Our work firstly demonstrated that the PIVKA-II level at postoperative three month is elevated in the metastasis group after LT, which emphasizes its value of metastasis monitoring.

Still, there are some limitations in this study. For the scarcity of HCC patients with LT treatment, the sample size was small, and no significant results were found regarding the progression free survival and overall survival. Besides, also for the limited sample size, we did not observe a regular trend of changes in hematological indices. Some indices were significantly changed at one time point (e.g., the third month) but turned negative at another (e.g., the fourth month) [21], [22], [23], [24], [25]. Although the significantly changed time point may be a key stage of post-LT metastasis (e.g., at round day 45), the clear trend, changing curves, and burst-out time point of metastasis are still to be explored. Moreover, we found there was a relationship between the mesenchymal CTC level at postoperative 1.5 M and postoperative metastasis; however, no difference was found at later period (4 M). This may be due to that the level at 1.5 M is more sensitive before a fully developed metastasis. But the sample size in the metastasis group was small, so the mesenchymal CTC level at 1.5 M was even higher than 4 M. This association (at 1.5 M) needs more data to confirm.

In conclusion, microvascular invasion, AFP level at the third or fourth month, PIVKA-Ⅱ level at the third month, and mesenchymal CTC level at day 45 were associated with post-LT metastasis. Using the logistic regression based on above variables, the two-year metastasis can be predicted with satisfactory sensitivity and accuracy. We recommend that in the clinical practice of preventing/monitoring metastasis, AFP level at the third or fourth month, PIVKA-Ⅱ level at the third month, and mesenchymal CTC level at day 45 can be paid special attention.