Decreased granzyme B⁺CD19⁺B cells are associated with tumor progression following liver transplantation

Abstract

Lymphocytes play an important role in antitumor immunity following organ transplantation. However, the function of granzyme B⁺CD19⁺B cells on the hepatocellular carcinoma cells from liver transplant recipients remains largely unknown; we aimed to analyze the function and elucidate the mechanisms behind it. Blood samples and clinical data from liver transplant recipients and healthy controls at Beijing Chaoyang Hospital as well as from a validation cohort were collected and analyzed. In this study, we found decreased granzyme B⁺CD19⁺B cells were correlated with early hepatocellular carcinoma recurrence and could further identify liver transplant recipients with poor tumor differentiation, microvascular invasion, increased total tumor diameter, and tumor beyond Milan criteria. Notably, granzyme B⁺CD19⁺B cells directly inhibited the proliferation, migration, and invasion of hepatocellular carcinoma cells. Upon activation regulatory B cells from liver transplant recipients with hepatocellular carcinoma recurrence displayed a CD5⁺CD38⁺CD27⁺CD138⁺CD19⁺ granzyme B⁺ phenotype, but the increased expression of CD5, CD38, and CD138, and the decreased protein level and transcriptional level requiring JAK/STAT signaling. In an independent validation cohort, liver transplant recipients with decreased granzyme B⁺CD19⁺B cells had not only early hepatocellular carcinoma cell recurrence but also shorter survival. Our study provides comprehensive data from liver transplant recipients with hepatocellular carcinoma, indicating a critical role of granzyme B⁺CD19⁺B cells in preventing cancer progression. Our findings warrant further investigations for the design of future immunotherapies leading to immune responses and improved patient survival.

Introduction

Liver transplantation offers the highest chance of cure for hepatocellular carcinoma (HCC) among all other methods as it treats not only the tumor but also the underlying liver disease [1,2]. However, liver transplant recipients (LTR) have been plagued by HCC recurrence following transplantation, which ranges between 8%-20% [3]. Unlike HCC recurrence after surgical resection, extrahepatic metastases are more seen following transplantation, particularly in the lungs and bones [4-6]. Studies have shown that post-transplant HCC recurrence is a result of the growth of occult metastases indicating that circulating tumor cells are the major route of HCC recurrence [7,8]. Therefore, peripheral immune responses play a key role in HCC progression requiring further elucidation [9,10].

B cells were once considered to act as antigen-presenting cells and provide co-stimulatory signals for T cell activation in addition to antibody production. In 2006, Jahrsdörfer B et al. first described that B-chronic lymphocytic leukemia (B-CLL) cells were characterized by inducing apoptosis of untreated B-CLL cells through granzyme B (GrB) production [11]. Subsequently, in patients with operationally tolerant kidney grafts, B cells exerted a suppressive function on CD4⁺CD25^- effector T cell by secreting GrB [12]. We also confirmed GrB⁺CD19⁺B cells could maintain allospecific tolerance and enhance viral control in renal transplant recipients in a previous study [13]. Furthermore, studies reported that B cells infiltrating the tumor correlated with an increased GrB expression in addition to with reduced tumor viability suggesting B cells possessing tumor-killing potential by producing GrB [14,15].

However, there is little data on the relation between GrB⁺CD19⁺B cells and LTR survival or HCC recurrence following liver transplantation. Here, we examined the role of GrB⁺CD19⁺B cells from LTR with and without HCC recurrence as well as from healthy controls (HC) and LTR with hepatic failure (HF) and further confirmed our findings in a validation cohort.

Materials and methods

Patients and samples

A total of 102 liver transplant recipients with HF and with pathologically confirmed HCC were enrolled in this study as well as 41 HC. Characteristics are summarized in Table 1. Another validation cohort was further enrolled between January 2016 and December 2020. There were neither ABO incompatible donors nor living donors. The blood samples were taken at least 6 months following liver transplantation. Basiliximab was administered for immunosuppressive induction. LTR were given a low dose of calcineurin inhibitors. Steroids were not used or gradually withdrawn within one month. Mycophenolate mofetil was adjusted according to the white blood cell count. Sirolimus was used one month after surgery.

Table 1.

Characteristics of liver transplant recipients and healthy controls

Parameters	Liver transplant recipients		Healthy controls (n=41)	P
	Malignant (n=57)	Benign (n=45)
Age	51.26±7.78	53.27±7.02	50.76±7.72	0.253
Sex (male)	49	34	29	0.169
Maintenance				0.562
Tacrolimus	47	39
Cyclosporin A	10	6
Recurrence (cases)	22
Tumor-free survival (months)	22.09±13.06	47.09±26.74		0.000

Inclusion criteria: patients with pathologically confirmed hepatic carcinoma and hepatitis related cirrhosis; patients undergoing a first liver transplantation; no distant metastasis; patients with complete clinical and pathological data. Exclusion criteria: patients with HCC receiving anticancer therapy before sampling; patients with concurrent autoimmune disease; patients with any kind of infections, acute rejection and graft-versus-host disease; patients with diseases of hematopoietic and lymphoid systems and human immunodeficiency virus; patients undergoing combined organ transplantation.

The study was approved by the Institutional Review Board of Beijing Chaoyang Hospital (No.2016-2-19-38) in accordance with the Helsinki declaration of 1975, as revised in 1983. Written informed consent was obtained from all participants.

Flow cytometry, polymerase chain reaction, western blot and tumor cell cultivation

According to the instructions of the manufacturers, flow cytometry, polymerase chain reaction (Table S1), Western blot and tumor cell cultivation were performed, respectively (Supplementary Information).

Statistical analysis

Data analyses were carried out by using SPSS 19.0 computer software (IBM Corp., Armonk, NY, USA) and GraphPad Prism 5 software (GraphPad Software Inc., La Jolla, CA, USA). Values were expressed as mean ± standard deviation. The Kolmogorov-Smirnov test was used for the normal distribution of continuous variables and one-way analysis of variance for the differences of multi-groups. The independent samples t-test was employed for quantitative variables. The Mann-Whitney U test was selected due to non-normal distribution. The Chi-square or Fisher’s exact test was used to compare nominal variables. Receiver operating characteristic curve (ROC) analysis and comparison of the area under the curve (AUC) was performed. The cut-off value for positive parameters was further determined by optimal sensitivity and specificity on ROC curve analysis. Multivariate Cox analysis was employed to determine the predictors. Relative risk was expressed as an odds ratio (OR) with a 95% confidence interval (CI). A P-value <0.05 was considered statistically significant.

Results

Decreased GrB⁺CD19⁺B cells are correlated with early HCC recurrence

First, we detected the GrB expression in CD19⁺B cells from blood samples of LTR under varying conditions. Based on the previous study [13], we found CD19⁺B cells started to produce GrB upon activation. A combination of anti-BCR and IL-21 is the most potent stimulus to induce GrB production (Figure S1A). Then, we found there was no significant difference among the groups with respect to GrA production or perforin production in resting CD19⁺B cells (P>0.05, Figure S1B, S1C). Moreover, both GrA production and perforin production remained similar when CD19⁺B cells were stimulated with anti-BCR and IL-21 (P>0.05, Figure S1D-F).

Subsequently, we wanted to compare the percentages of GrB⁺CD19⁺B cells in HC and LTR. GrB was weakly expressed in a small portion of resting B cells in both groups (P>0.05, Figure 1A). After stimulation, we found the frequency of GrB expression was higher in HC (P<0.05, Figure 1B). We further checked the percentages of GrB⁺CD19⁺B cells between LTR with HF and with HCC. Of note, LTR with HCC showed a lower proportion of GrB⁺CD19⁺B cells in activated B cells (P<0.05, Figure 1D) although there was no difference in resting B cells (P>0.05, Figure 1C). Finally, we analyzed the data from patients with HCC recurrence and without HCC recurrence. LTR with and without HCC recurrence had similar GrB production in resting B cells (P>0.05, Figure 1E). More importantly, upon activation LTR with HCC recurrence had a remarkably lower proportion of GrB⁺CD19⁺B cells (P<0.05, Figure 1F, 1G).

Figure 1.

Decreased GrB⁺CD19⁺B cells are correlated with early HCC recurrence. Comparison of the expression of GrB in resting (P>0.05, A) and activated (P<0.01, B) CD19⁺B cells from HC and LTR, respectively; Comparison of the expression of GrB in resting (P>0.05, C) and activated (P<0.05, D) CD19⁺B cells from LTR with HF and with HCC, respectively; Comparison of the expression of GrB in resting (P>0.05, E) and activated (P<0.01, F) CD19⁺B cells from LTR with and without HCC recurrence, respectively; Representative flow cytometry dot plots of the expression of GrB in activated CD19⁺B cells from HC and LTR with HF and with HCC, respectively (G). Comparison of the percentages of GrB⁺CD19⁺B cells between tumor within and beyond Milan criteria (P<0.01, H); Kaplan-Meier curve analysis of tumor-free survival between high and low percentages of GrB⁺CD19⁺B cells using the mean value of HCC group as the cut-off (P<0.01, I). PBMC were stimulated with IgG+IgM (5.4 μg/ml) in the presence of IL-21 (50 ng/ml) for 20 h and Brefeldin A (1 μg/ml) was added for the last 4 h of incubation. GrB expression in B cells was detected by flow cytometry. Bars represent mean and standard deviation. GrB, granzyme B; HC, healthy controls; LTR, liver transplant recipients; HF, hepatic failure; HCC, hepatocellular carcinoma; PBMC, peripheral blood mononuclear cells.

After further analysis, a higher proportion of GrB⁺CD19⁺B cells were observed when patients with HCC met the Milan criteria, which is the golden criteria for liver transplantation (P<0.01, Figure 1H). To assess the effect of the percentages of GrB⁺CD19⁺B cells on tumor-free survival, we divided the patients into two groups (high frequency >25.46% and low frequency <25.46%) using the mean value of the percentages of GrB⁺CD19⁺B cells calculated according to the patients with and without HCC recurrence. Surprisingly, nearly 2.3 times more patients with low frequency than those with high frequency showed early recurrence (58.6% vs. 17.9%; Figure 1I). Collectively, these data indicate that decreased percentages of GrB⁺CD19⁺B cells correlate with early recurrence in LTR.

Decreased GrB⁺CD19⁺B cells can identify LTR with progressive HCC

Given that our data demonstrated a strong association between HCC recurrence and decreased GrB⁺CD19⁺B cell production, we wondered whether clinical-pathological parameters, which differentiated patients with HCC recurrence from without HCC recurrence, could have an impact on GrB expression. First, we compared the clinical parameters to check their effect. As patients with HCC had a stable preoperative liver function, the model for end-stage liver disease score was similar between LTR with and without HCC recurrence (P>0.05, Figure S2A). Although high exposure to them can increase the risk of HCC recurrence [16], calcineurin inhibitors are still our major choice due to their effective suppression. Therefore, neither the type of calcineurin inhibitors nor the tacrolimus trough level affected HCC recurrence (P>0.05, Figure S2B, S2C). HCC is commonly associated with chronic hepatitis and liver cirrhosis in China, which prevailed in both groups. Accordingly, they failed to have an impact on HCC recurrence (P>0.05, Figure S2D, S2E). Notably, preoperative transcatheter arterial chemoembolization and hepatitis recurrence which was reported to correlate with HCC recurrence [17,18] showed a negative association as a result of a partial pathologic response and tenofovir disoproxil fumarate treatment [19], respectively (P>0.05, Figure S2F, S2G). Interestingly, levels of postoperative AFP remained similar between LTR with and without HCC recurrence (P>0.05, Figure S2H). As the patients were followed up at regular intervals, therefore, HCC recurrence was detected at an early stage.

Then, we analyzed the pathological parameters, which play a key role in HCC recurrence. We found there was no significant difference between LTR with and without HCC recurrence with respect to macrovascular invasion, capsule invasion, or lesions ≥4 (P>0.05, Figure S3A-C). However, LTR with poor tumor differentiation, microvascular invasion, HCC beyond Milan criteria, and larger total tumor diameter had a higher rate of HCC recurrence (P<0.05, Figure S3D-G). Subsequently, we checked the associations between the positive parameters and the percentages of GrB⁺CD19⁺B cells. Notably, the percentages of GrB⁺CD19⁺B cells correlated with tumor differentiation, microvascular invasion, total tumor diameter (Figure 2A), and tumor beyond Milan criteria (P<0.05). More importantly, using poor tumor differentiation, microvascular invasion, total tumor diameter >9 cm, and HCC beyond Milan criteria as the cut-offs to regroup the LTR, respectively, we found LTR meeting these conditions all had lower proportions of GrB⁺CD19⁺B cells (P<0.05, Figures 2B-D, 1H).

Figure 2.

Decreased GrB⁺CD19⁺B cells can identify LTR with progressive HCC. Correlations between the percentage of GrB⁺CD19⁺B cells and total tumor diameter (P<0.01, A); The percentages of GrB⁺CD19⁺B cells between LTR with poor and non-poor tumor differentiation (P<0.01, B), with and without microvascular invasion (P<0.01, C), with total tumor diameter >9 cm and ≤9 cm (P<0.01, D), respectively; ROC curve analysis of microvascular invasion, the percentage of GrB⁺CD19⁺B cells, tumor differentiation, total tumor diameter and Milan criteria (E); Kaplan-Meier curve analysis of tumor-free survival between LTR with the percentage of GrB⁺CD19⁺B cells >22.45% and <22.45% (P<0.01, F). PBMC were stimulated with IgG+IgM (5.4 μg/ml) in the presence of IL-21 (50 ng/ml) for 20 h and Brefeldin A (1 μg/ml) was added for the last 4 h of incubation. GrB expression in B cells was detected by flow cytometry. Bars represent mean and standard deviation. HCC, hepatocellular carcinoma; GrB, granzyme B; LTR, liver transplant recipients; ROC, receiver operating characteristic; PBMC, peripheral blood mononuclear cells; MVI, microvascular invasion; TD, tumor differentiation; TTD, total tumor diameter; MC, Milan criteria.

To test their predictive values we performed the analysis of AUC-ROC for the positive pathological parameters and the percentages of GrB⁺CD19⁺B cells (Figure 2E), which revealed the latter strongly classified HCC recurrence versus non-recurrence (AUC 0.801, P<0.01) with the cut-off value of 22.45%. The corresponding sensitivity and specificity were 77.3% and 74.3%, respectively. Consequently, LTR with GrB⁺CD19⁺B cells <22.45% had shorter tumor-free survival (P<0.01; Figure 2F). Collectively, these data reveal that decreased GrB⁺CD19⁺B cell production is significantly associated with progressive HCC and has a better predictive value.

GrB⁺CD19⁺B cells can directly suppress HCC progression

B cells were previously regarded as antigen-presenting cells and immunoglobin-producing cells. Recently, studies have shown that B cells can exert a suppressive impact on lymphocytes through GrB secretion, which could be partially reversed by PI-9, namely the GrB inhibitor [11,12]. Furthermore, Shi et al. reported that CD20⁺B cells in the tumor invasive margin possessed tumor-killing potential by producing GrB [14]. These researches and our data prompted us to find out whether GrB⁺CD19⁺B cells can directly suppress HCC progression. Therefore, we chose two HCC cell lines (Hepg2 and Huh7) for co-culture and divided them into three groups including HCC cells alone (group 1), HCC cells + GrB⁺CD19⁺B cells (group 2), and HCC cells + GrB⁺CD19⁺B cells + PI-9 (group 3). We first detected the effect of GrB⁺CD19⁺B cells on HCC cell proliferation. Using CCK8 assays, we found on day 1 the OD values were a little lower in group 1 because group 2 and group 3 had larger cell numbers. On the following days group 1 showed remarkably increased OD values compared with group 2. In contrast, the OD values rose in group 3 after adding PI-9 despite failing to reach significance. All these data meant that CD19⁺B cells significantly restrained HCC cell growth via GrB secretion (P<0.05, Figure 3A, 3B). We also noticed a reduction in OD values of all groups on day 3, which might be a result of limited space and nutrition. Additional colony formation assay further confirmed that GrB⁺CD19⁺B cells could significantly inhibit the proliferation of HCC cells (P<0.05, Figure 3C, 3D). After that, we investigated whether GrB⁺CD19⁺B cells affected HCC cell migration and invasion by taking advantage of scratch wound healing assay and transwell cell invasion assay, respectively. We found that group 1 had the higher healing rate (P<0.05, Figure 3E, 3F) and the invasive cell numbers (P<0.05, Figure 3G, 3H) compared with group 2. The results indicated invasion and migration abilities of HCC cells were greatly decreased after co-culture with GrB⁺CD19⁺B cells. Furthermore, the suppressive function of GrB⁺CD19⁺B cells can be affected in part by PI-9 in accordance with the literature. Collectively, these data indicate that GrB⁺CD19⁺B cells directly inhibits the proliferation, migration, and invasion of HCC cells.

Figure 3.

GrB⁺CD19⁺B cells can directly suppress HCC progression. The effects of GrB⁺CD19⁺B cells on the proliferation (A-D), migration (E, F), and invasion (G, H) in HepG2 and Huh-7 cells as analyzed by CCK-8, colony formation, scratch wound healing, and transwell cell invasion assays, respectively. Bars represent mean and standard deviation. Original magnification ×5 for colony formation assay; original magnification ×40 for scratch wound healing assay; original magnification ×100 for transwell cell invasion assay. Purified CD19⁺B cells (10⁶/ml) were stimulated with IgG+IgM (5.4 μg/ml), CpG OND 2006 (9.6 μg/ml) and IL-21 (50 ng/ml) in the presence or absence of PI-9 (5 μg/ml) for 4 d before coculture. GrB, granzyme B; CCK-8, cell counting kit-8; HCC, hepatocellular carcinoma; OD, optical density; PI-9, proteinase inhibitor-9.

Phenotypic characteristics of GrB⁺CD19⁺B cells from LTR with HCC

In order to investigate the phenotypic features of this cell subset isolated from LTR with HCC, we assessed the surface molecules on the CD19⁺B cell population identified on the basis of the GrB production. CD19⁺B cells were stimulated with IgG+IgM in the presence of IL-21 as either GrB (Figure S4A, S4B) or surface molecules (Figure S4C) could be hardly detected without stimulation. After analysis, we found GrB⁺B cells displayed a CD5⁺ CD38⁺CD27⁺CD138⁺CD19⁺ phenotype upon stimulation (Figure 4A), which were different from that of typical IL-10-producing regulatory B cells [20-22]. Subsequently, we found GrB⁺CD19⁺B cells and IL-10⁺CD19⁺B cells were distinct B cell subsets (Figure 4B), which may explain the different phenotypes of the cell subsets. Then, we compared the frequency of each surface molecule between LTR with and without recurrence. Notably, the proportions of CD5⁺GrB⁺CD19⁺B cells from LTR with HCC recurrence were higher than those from LTR without HCC recurrence in activated and resting GrB⁺CD19⁺B cells, respectively (P<0.05, Figures 4C and S4D). Studies have demonstrated CD5⁺B cells could play an important role in supporting tumor cell growth [23,24], which are consistent with our data. Interestingly, in LTR with HCC recurrence the percentages of CD5⁺GrB⁺CD19⁺B cells were similar to those of CD5^-GrB⁺CD19⁺B cells (P>0.05, Figure 4D). In contrast, the production of CD27⁺GrB⁺CD19⁺B cells remained similar among different populations with or without stimulation (P>0.05, Figures 4E and S4E). Finally, we analyzed the frequencies of CD38⁺GrB⁺CD19⁺B cells and CD138⁺GrB⁺CD19⁺B cells, which were found to be higher in LTR with HCC recurrence upon activation suggesting their potential differentiation toward plasma cells (P<0.05, Figure 4F, 4G; P>0.05, Figure S4F, S4G).

Figure 4.

Phenotypic characteristics of GrB⁺CD19⁺B cells from LTR with HCC. Representative flow cytometry dot plots of the expression of surface molecules on GrB⁺CD19⁺B cells from LTR with HCC recurrence upon stimulation (A); A representative flow cytometry dot plot of GrB⁺CD10⁺CD19⁺B cells from LTR with HCC recurrence upon stimulation (B); Comparison of the percentages of CD5⁺GrB⁺CD19⁺B cells between LTR with and without HCC recurrence upon stimulation (P<0.05, C); Comparison of the percentages of CD5⁺GrB⁺CD19⁺B cells and CD5^-GrB⁺CD19⁺B cells between LTR with and without HCC recurrence upon stimulation (P>0.05, D); Comparison of the percentages of CD27⁺GrB⁺CD19⁺B cells (P>0.05, E), CD38⁺GrB⁺CD19⁺B cells (P<0.05, F) and CD138⁺GrB⁺CD19⁺B cells (P<0.05, G) between LTR with and without HCC recurrence upon stimulation. PBMC were stimulated with IgG+IgM (5.4 μg/ml) in the presence of IL-21 (50 ng/ml) for 20 h and Brefeldin A (1 μg/ml) was added for the last 4 h of incubation. Surface markers expression on B cells was detected by flow cytometry. Bars represent mean and standard deviation. GrB, granzyme B; LTR, liver transplant recipients; HCC, hepatocellular carcinoma; PBMC, peripheral blood mononuclear cells.

GrB⁺CD19⁺B cell production decreases at both the protein level and the transcriptional level requiring JAK/STAT signaling in LTR with HCC recurrence

The expression of GrB in CD19⁺B cells isolated from LTR with HCC was confirmed by PCR and western blot analysis, respectively. Hence, GrB expression was demonstrated for both mRNA and protein levels. Unlike classical cytotoxic cells such as CD3⁺T cells, which store preformed GrB without stimulation (Figure S5A), B cells only express GrB following activation. First, we tested the effect of varying stimuli on GrB production while the resting CD19⁺B cells served as a negative control confirming that GrB production was sharply enhanced upon stimulation with anti-BCR and with IL-21 (Figure S5B). Next, we found that exogenous IL-21 could increase the levels of GrB mRNA (Figure S5C) but higher levels of IL-21 could lead to an increased rate of CD19⁺B cell apoptosis (Figure S5D). In contrast, higher levels of anti-BCR could result in slowly elevated levels of GrB mRNA rather than an increased rate of CD19+B cell apoptosis (Figure S5E). After that, the proper levels of anti-BCR and IL-21 were reached to maximize GrB expression (Figure 5A). We further found LTR with HCC recurrence had lower levels of GrB mRNA compared with LTR without HCC recurrence under the same condition (P<0.05, Figure 5B, 5C).

Figure 5.

GrB⁺CD19⁺B cell production decreases at both the protein level and the transcriptional level requiring JAK/STAT signaling in LTR with HCC recurrence. Comparison of GrB mRNA levels upon activation under varying conditions with resting B cells as a control (A); Representative PCR for GrB in LTR with and without HCC recurrence upon activation with resting B cells as a control (B); Comparison of GrB mRNA levels in LTR with and without HCC recurrence upon activation with resting B cells as a control (P<0.05, C); Western blot analysis of GrB, pJAK1, pJAK3, pSTAT1 and pSTAT3 with β-actin as a control in CD19⁺B cells from LTR with HCC recurrence (D); Comparison of relative band intensity for GrB (E), pJAK1 (F) and pSTAT3 (G) in activated GrB⁺CD19⁺B cells from LTR with and without recurrence with β-actin as a control (P<0.05). CD19⁺B cells were stimulated in the presence or absence of IgG+IgM (5.4 μg/ml) or IL-21 (50 ng/ml) or both for 20 h and Brefeldin A (1 μg/ml) was added for the last 4 h of incubation. GrB mRNA expression was detected by PCR and GrB signal pathway by Western blot, respectively. Bars represent mean and standard deviation. GrB, granzyme B; LTR, liver transplant recipients; HCC, hepatocellular carcinoma.

Induction of a B cell-mediated response depends on the activation of the JAK/STAT pathway and IL-21 signaling is associated with activation of a series of JAK/STAT family members [25-27]. Consequently, we stained CD19⁺B cells for pJAK1, pJAK3, pSTAT1, and pSTAT3 and included GrB and actin as controls. We found the different extent of up-regulation of GrB, pJAK1, and pSTAT3 and no up-regulation of pSTAT1 or pJAK3 (Figure 5D). When we compared protein expression from patients with and without HCC recurrence, the levels of GrB, pJAK1, and pSTAT3 were significantly higher in patients without HCC recurrence (Figure 5E-G). Collectively, these data show that CD19⁺B cells from LTR with HCC recurrence produce less GrB at the protein level and the transcriptional level via JAK1 and STAT3 signaling.

Decreased GrB⁺CD19⁺B cells can predict HCC recurrence in an independent validation cohort

In an attempt to validate the suppressive effect of GrB⁺CD19⁺B cells on tumor progression following liver transplantation, we performed an independent study. Patients lacking the follow-up data were excluded. In this validation cohort, a total of 155 LTR with HCC were enrolled, who accepted liver transplantation and were then followed up at our center. Blood samples were obtained to analyze the lymphocyte subsets. These patients were followed up at regular intervals to observe the HCC recurrence and patient’s survival following liver transplantation. 49 patients had a tumor recurrence during the follow-up period and 33 patients died (Table 2).

Table 2.

Characteristics of liver transplant recipients with HCC from a validation cohort

Parameters	With recurrence (n=49)	Without recurrence (106)	P
Age	53.02±8.05	54.42±8.24	0.322
Sex (male)	45	99	0.988
Maintenance			0.355
Tacrolimus	38	89
Cyclosporin A	11	17
Tumor-free survival (months)	16.69±9.47	28.32±14.58	0.000
Status (alive)	18	102	0.000
Cum survival (months)	26.33±13.20	28.32±14.58	0.416

In this validation cohort, LTR with recurrence showed lower percentages of GrB⁺CD19⁺B cells (P<0.05, Figure 6A). Then, they were regrouped at the cut-off value of 22.45% calculated above, the HCC recurrence rate proved to be higher in the group with the frequency <22.45% (P<0.01, Figure 6B). Again, LTR with poor tumor differentiation, microvascular invasion, HCC beyond Milan criteria, and larger total tumor diameter had a higher rate of HCC recurrence (P<0.05, Figure S6A-D), which all exhibited a positive correlation with the decreased percentages of GrB⁺CD19⁺B cells (P<0.01, Figure S6E-H). Therefore, LTR from the group with the frequency <22.45% had the shorter tumor-free survival and cum survival (P<0.01, Figure 6C, 6D). Subsequent ROC analysis further confirmed that the percentage of GrB⁺CD19⁺B cells was a strong predictor for HCC recurrence with the AUC of 0.808 (P<0.01, Figure 6E). Finally, we asked if the percentage of GrB⁺CD19⁺B cells could independently predict HCC recurrence. Using multiple logistic regression analysis, the percentage of GrB⁺CD19⁺B cells (OR: 1.142; 95% CI: 1.064-1.225; P<0.01) was the only significant independent risk factor (Table 3), while tumor differentiation, microvascular invasion, tumor beyond Milan criteria or total tumor diameter did not reach statistical significance. Collectively, these data indicate that the percentage of GrB⁺CD19⁺B cells has a strong impact not only on the HCC recurrence but also the patient survival.

Figure 6.

Decreased GrB⁺CD19⁺B cells can predict HCC recurrence in an independent validation cohort. Comparison of the percentages of GrB⁺CD19⁺B cells upon activation in LTR with and without HCC recurrence (P<0.05, A); Cases of tumor recurrence between LTR with the frequency of GrB⁺CD19⁺B cells >22.45% and <22.45% (P<0.05, B); Kaplan-Meier curve analysis of tumor-free survival (C) and cum survival (D) between LTR with the frequency of GrB⁺CD19⁺B cells >22.45% and <22.45% (P<0.05); ROC curve analysis of the percentage of GrB⁺CD19⁺B cells, tumor differentiation, microvascular invasion, total tumor diameter and Milan criteria (E). PBMC were stimulated with IgG+IgM (5.4 μg/ml) in the presence of IL-21 (50 ng/ml) for 20 h and Brefeldin A (1 μg/ml) was added for the last 4 h of incubation. GrB expression in B cells was detected by flow cytometry. Bars represent mean and standard deviation. GrB, granzyme B; LTR, liver transplant recipients; HCC, hepatocellular carcinoma; ROC, receiver operating characteristic; PBMC, peripheral blood mononuclear cells.

Table 3.

Multivariate Cox analysis

Parameters	OR	95% CI	P
Poor tumor differentiation	1.508	0.542-4.195	0.431
Microvascular invasion	1.789	0.760-4.213	0.183
Total tumor diameter	0.898	0.726-1.112	0.326
Within Milan criteria	1.325	0.381-4.606	0.658
GrB+CD19+B cell%	1.142	1.064-1.225	0.000

CI, confidence interval; OR, odds ratio; GrB, granzyme B.

Discussion

We demonstrated that decreased GrB⁺CD19⁺B cells are correlated with early HCC recurrence and can identify LTR with pathological proven progressive HCC, suggesting their suppressive function on HCC. We further demonstrated that GrB⁺CD19⁺B cells directly inhibit the proliferation, migration, and invasion of HCC cells. GrB⁺CD19⁺B cells from LTR with HCC recurrence display a CD5⁺CD38⁺CD27⁺CD138⁺ phenotype, but the higher expression of CD5, CD38, and CD138, and the lower protein level and transcriptional level requiring JAK/STAT signaling. To support the role of GrB⁺CD19⁺B cells, we found LTR with decreased GrB⁺CD19⁺B cells have not only early HCC recurrence but also shorter survival in an independent validation cohort.

GrB is a serine protease most commonly produced by natural killer cells and cytotoxic T cells, which can also secrete the pore-forming protein, perforin, to mediate apoptosis in target cells. Recent studies have revealed that B cells can produce GrB upon stimulation [11,28]. We demonstrated that human B cells gained regulatory potential in response to IL-21 provided additional triggering of the BCR and TLRs is present. Due to the immunosuppressive nature, regulatory B cells have been repeatedly reported to be involved in transplantation tolerance [12,13,29]. In contrast, its regulation of antitumor immunity is poorly understood following transplantation although the existence of tumor-infiltrating B cells has been previously described [14,15]. In our study, we found LTR with decreased percentages of GrB⁺CD19⁺B cells showed early HCC recurrence. When we further analyzed the correlations between the frequency of GrB⁺CD19⁺B cells and pathological data, poor tumor differentiation, microvascular invasion, total tumor diameter, and HCC beyond Milan criteria, which in essence reflect the progressive nature of HCC, had a positive association with decreased frequency of GrB⁺CD19⁺B cells. Taken together, these data suggested the potential antitumor function of GrB⁺CD19⁺B cells. Subsequent co-culture experiments confirmed the suppressive role of GrB⁺CD19⁺B cells, which is consistent with the clinical observations. Interestingly, the B cells produce GrB without co-expression of perforin. Therefore, GrB uptake into target cells is independent of perforin, through either mannose-6-phosphate-receptor or fluid phase endocytosis [30,31]. However, the specific mechanism as to how B cells suppress HCC cells via GrB production needs to be investigated further.

In addition, we found GrB⁺CD19⁺B cells were separate from IL-10⁺CD19⁺B cells, which are the classic type of regulatory B cells [32], and only a very small portion of GrB⁺IL-10⁺B cells developed after activation. This indicates that GrB⁺B cells are a separate subset different from IL-10⁺ regulatory B cells. Until today the exact phenotype of regulatory B cells is not clearly defined. Therefore, wide discrepancies exist among varying studies [33]. In this study, we observed that GrB⁺CD19⁺B cells in LTR with HCC recurrence had elevated expression of plasma cell marker, namely CD38 and CD138, meaning the recruited GrB⁺CD19⁺B cells underwent further maturation. Of note, we also detected the elevated expression of CD5 on B cells. The role of over-expression of CD5 on B cells in cancer pathogenesis has been recognized recently [23,24]. Thus, this might be a possible explanation at the molecular level for early tumor recurrence in LTR with decreased GrB⁺CD19⁺B cells. Induction of GrB in B cells involves both transcriptional and translational events and depends on the phosphorylation of JAK1 and STAT3. Accordingly, we noticed a reduction in GrB⁺CD19⁺B cells from LTR with HCC recurrence at both the protein level and the transcriptional level. Therefore, these data support the result that LTR with HCC recurrence exhibit decreased production of GrB⁺CD19⁺B cells.

In this study, we showed comprehensive data from LTR with HCC, indicating a critical role of GrB⁺CD19⁺B cells in preventing cancer progression. Our findings warrant further investigations for the design of future immunotherapies leading to immune responses and improved patient survival.

Disclosure of conflict of interest

None.