Increased frequency of Bregs in the peripheral blood and ascites of patients with GC and PM
We analyzed Breg frequency using flow cytometry in peripheral blood and ascites samples from healthy control subjects and patients with early GC, advanced GC, and PM (n = 10 for each group). The Breg frequency is represented by the ratio of the number of CD19+CD24hiCD27+ B cells to the number of CD19+ B cells. Abdominal lavage fluid without ascites was used for patients with early and advanced GC without PM. The frequency of Bregs in the peripheral blood of PM patients was higher than that of healthy control subjects and early GC patients, and was higher than in advanced GC patients, although there was no significant difference (control vs. PM; 10.1 ± 1.6% vs. 19.4 ± 3.9%, p = 0.004, early GC vs. peritoneal metastasis; 11.2 ± 2.0% vs. 19.4 ± 3.9%, p = 0.0143) (Fig. 1a). The frequency of Bregs in ascites of PM patients was also higher than that of early GC and advanced GC patients (early GC vs. PM; 2.5 ± 2.4% vs. 17.6 ± 9.0%, p = 0.02, advanced GC vs. peritoneal metastasis; 5.5 ± 2.3% vs. 17.6 ± 9.0%, p = 0.02) (Fig. 1b).
Breg frequency is elevated in PM tissue and is associated with CD8 + T cell and M2 macrophage infiltration
In double immunostaining, CD19+IL-10+ cells were defined as Bregs, and CD19+ cells as B cells. The frequency of Bregs, represented by the ratio of the number of double-positive cells to the number of CD19+ cells, was evaluated in the primary tumor, PM, and lymph node metastases in paired patients with PM. Breg frequency increased in PM samples compared with their primary tumor (p < 0.001). The frequency of Bregs in PM tissues was also higher than that in the lymph nodes, liver, and ovarian metastatic tissues (p < 0.001) (Fig. 2a, b).
In addition, we evaluated the association between Breg frequency and CD8+ T cell and CD163+ cell (M2 macrophage) infiltration in PM (Fig. 2c). The median value determined the cut-off point for the Breg high /low frequency groups, which was set at 17.2%. The high-frequency group had 29 patients and the low-frequency group had 31 patients. On the other hand, the CD8+ T cell count in the Breg high frequency group was lower than that in the Breg low frequency group (p < 0.001). In contrast, the M2 macrophage count in the Breg high frequency group was higher than that in the Breg low frequency group (p < 0.001). As Bregs in the cancer microenvironment may affect patient prognosis, we investigated the relationship between Breg frequency and clinicopathological factors in patients with PM.
Bregs infiltration is an independent prognostic factor in patients with PM of GC
We analyzed the relationships between Breg frequency in PM and clinicopathological factors in patients with PM, and found no differences. The median value determined the cut-off point for the Breg high /low frequency groups in primary tumor, which was set at 10.2%. There was a tendency for the Breg frequency of peritoneal tumors to correlate with the Breg frequency of the primary tumor, but the difference was not significant (Table 1). The median survival time (MST) of patients in the Breg high frequency group was 10.8 months, and their OS was significantly lower than that of patients in the Breg low frequency group, whose MST was 25.2 months. (p = 0.005) (Fig. 3). The univariate analysis revealed differences in age, performance status (PS), history of chemotherapy, carcinoembryonic antigen (CEA) level, and Breg frequency. In the multivariate analysis, Breg frequency and chemotherapy were identified as independent prognostic factors (hazard ratio (HR): 2.314, 95% confidence interval (CI): 1.298–4.124, p = 0.004; HR: 0.158, 95% CI: 0.034–0.738, p = 0.02, respectively), as shown in Table 2.
Table 1
Relationship between Breg frequency and clinicopathological factors in patients with peritoneal metastasis
Factors | | Number | Bregs frequency | p value |
| | | Low (n = 31) | High (n = 29) | |
Age, years | > 70 | 20 | 9 | 11 | 0.465 |
| ≦ 70 | 40 | 22 | 18 | |
Sex | Male | 26 | 14 | 12 | 0.768 |
| Female | 34 | 17 | 17 | |
Performance status | 0 | 50 | 26 | 24 | 0.590 |
| 1, 2 | 10 | 5 | 5 | |
BMI | ≦ 19.8 | 29 | 15 | 15 | 0.796 |
| >19.8 | 29 | 16 | 14 | |
Histology | differentiated | 13 | 7 | 6 | 0.859 |
| undifferentiated | 47 | 24 | 23 | |
First-onset case /recurrent case | first-onset | 53 | 26 | 27 | 0.241 |
| recurrence | 7 | 5 | 2 | |
Size of PM | Bowel resection | 17 | 9 | 8 | 0.901 |
| Biopsy or excision alone | 43 | 22 | 21 | |
Location of resected PM | Visceral peritoneum | 47 | 24 | 23 | 0.859 |
| Parietal peritoneum | 13 | 7 | 6 | |
Breg frequency in primary lesion | >10.2 | 27 | 10 | 17 | 0.060 |
| ≦ 10.2 | 33 | 21 | 12 | |
Chemotherapy | Yes | 57 | 31 | 26 | 0.066 |
| No | 3 | 0 | 3 | |
Intraperitoneal chemotherapy | Yes | 32 | 17 | 15 | 0.809 |
| No | 28 | 14 | 14 | |
P stage | P1a | 9 | 5 | 4 | 0.544 |
| P1b, P1c | 51 | 26 | 25 | |
CEA | >5 | 13 | 4 | 9 | 0.088 |
| ≦ 5 | 47 | 27 | 20 | |
CA19-9 | >37 | 14 | 6 | 8 | 0.451 |
| ≦ 37 | 46 | 25 | 21 | |
CA125 | >35 | 20 | 7 | 13 | 0.068 |
| ≦ 35 | 40 | 24 | 16 | |
Platelet | ≦ 223×103 | 20 | 7 | 13 | 0.068 |
| >223×103 | 40 | 24 | 16 | |
CRP | >1.0 | 9 | 4 | 5 | 0.638 |
| ≦ 1.0 | 51 | 27 | 24 | |
PNI | ≦ 40 | 26 | 11 | 15 | 0.205 |
| >40 | 34 | 20 | 14 | |
Chemotherapy | Yes | 57 | 31 | 26 | 0.066 |
| No | 3 | 0 | 3 | |
Intraperitoneal chemotherapy | Yes | 32 | 17 | 15 | 0.809 |
| No | 28 | 14 | 14 | |
P stage | P1a | 9 | 5 | 4 | 0.544 |
| P1b, P1c | 51 | 26 | 25 | |
CEA | >5 | 13 | 4 | 9 | 0.088 |
| ≦ 5 | 47 | 27 | 20 | |
CA19-9 | >37 | 14 | 6 | 8 | 0.451 |
| ≦ 37 | 46 | 25 | 21 | |
CA125 | >35 | 20 | 7 | 13 | 0.068 |
| ≦ 35 | 40 | 24 | 16 | |
Breg regulatory B cell, PM peritoneal metastasis, CRP C-reactive protein, PNI prognostic nutritional index, CEA carcinoembryonic, CA carbohydrate antigen |
Table 2
Univariate and multivariate analysis for prognostic factors in patients with peritoneal metastasis
Variable | | Number | Univariate analysis | | Multivariate analysis |
| | | HR | 95% CI | p value | | HR | 95% CI | p value |
Age, years | > 70/≦70 | 20/40 | 1.824 | 1.329–5.309 | 0.037 | | 1.741 | 0.661–2.573 | 0.09 |
Sex | Male/Female | 26/34 | 1.204 | 0.712–2.037 | 0.4826 | | | | |
Performance status | 0 /1, 2 | 50/10 | 2.471 | 0.945–6.457 | 0.0059 | | 0.794 | 0.323–1.955 | 0.616 |
BMI | ≦ 19.8/>19.8 | 29/29 | 0.949 | 0.561–1.604 | 0.8387 | | | | |
Histology | differentiated/undifferentiated | 13/47 | 0.908 | 0.480–1.717 | 0.7509 | | | | |
first-onset case/ recurrent case | first-onset /recurrence case | 53/7 | 0.961 | 0.440–2.097 | 0.9211 | | | | |
Size of PM | Bowel resection/ Biopsy or excision alone | 17/43 | 1.067 | 0.608–1.873 | 0.8204 | | | | |
Location of resected PM | Visceral peritoneum/ Parietal peritoneum | 47/13 | 1.150 | 0.603–2.191 | 0.6561 | | | | |
Chemotherapy | Yes/No | 57/3 | 0.084 | 0.002–3.808 | < 0.001 | | 0.158 | 0.034–0.738 | 0.019 |
Intraperitoneal chemotherapy | Yes/No | 32/28 | 0.809 | 0.473–1.382 | 0.4147 | | | | |
P stage | P1a/P1b, P1c | 9/51 | 1.399 | 0.693–2.826 | 0.3972 | | | | |
CEA | >5/≦5 | 13/47 | 1.974 | 0.913–4.272 | 0.026 | | 0.971 | 0.436–2.165 | 0.943 |
CA19-9 | >37/≦37 | 14/46 | 1.509 | 0.729–3.123 | 0.1932 | | | | |
CA125 | >35/≦35 | 20/40 | 1.480 | 0.822–2.665 | 0.1494 | | | | |
Platelet | ≦ 223×103/>223×103 | 20/40 | 1.734 | 0.916–3.282 | 0.447 | | | | |
CRP | >1.0/≦1.0 | 9/51 | 1.505 | 0.658–3.444 | 0.250 | | | | |
PNI | ≦ 40/>40 | 26/34 | 1.185 | 0.695–2.021 | 0.515 | | | | |
Breg frequency in primary lesion | >10.2/≦10.2 | 27/33 | 1.605 | 0.963–2.675 | 0.068 | | | | |
Breg frequency in PM | high/low | 29/31 | 2.144 | 1.222–3.763 | 0.002 | | 2.314 | 1.298–4.124 | 0.004 |
PM peritoneal metastasis, CRP C-reactive protein, PNI prognostic nutritional index, |
CEA carcinoembryonic, CA carbohydrate antigen, Breg regulatory B cell |
Bregs are frequently localized in ascites and peritoneal tumors in an immunocompetent mouse model of PM
We evaluated Breg infiltration in the ascites, spleen, subcutaneous tumors, and peritoneal tumors in an immunocompetent mouse model of PM using flow cytometry. The total number of cells in the abdominal lavage of control mice was significantly lower than that in the ascites of the PM model. The number of Bregs was significantly higher in the PM model than in the control mice (p < 0.001), although there was no difference in the Bregs ratio (Fig. 4a). We then simultaneously created peritoneal and subcutaneous tumors in mice and compared Breg frequency, represented by the ratio of the number of CD19+CD1dhiCD5+ cells to the number of CD19+ B cells, in both tumors; Breg frequency in peritoneal tumors was more pronounced than that in subcutaneous tumors (6.3 ± 2.5% vs. 1.4 ± 1.3%, p = 0.003) (Fig. 4b).
Analysis of Bregs and immune cells in a B cell-specific Pten-deficient mouse model of PM
We compared a PM model using B cell-specific Pten-deficient mice (Cd19Cre་/−PtenloxP/loxP mice) with a PM model using control mice (Cd19Cre་/− mice) and found an increased ascites volume and peritoneal tumor weight in B cell-specific Pten-deficient mice (Fig. 5a). The Breg frequency in the ascites and peritoneal tumor in the B cell-specific Pten-deficient mice were significantly higher than that in control mice (p = 0.001) (Fig. 5b). CD8+ T cell infiltration in ascites and peritoneal tumors was decreased in B cell-specific Pten-deficient mice (p = 0.005 and p = 0.03, respectively), while M2 macrophage infiltration increased compared to that in control mice (p = 0.005 and p = 0.003, respectively) (Fig. 5c).
Bregs promote M2 macrophage polarization and inhibit CD8 + T cell proliferation through the production of IL-10
The PI3K inhibitor, wortmannin attenuated the proliferation of YTN16 cells in a dose-dependent manner, with no significant effect on growth at a dose of 0.5µM as shown in Fig. 6a. Based on these results, 0.5µM wortmannin was used in subsequent in vitro experiments.
We evaluated the concentration of IL-10 in the supernatants of Bregs and other B cell subsets, excluding Bregs cultured with or without wortmannin. Bregs produced significantly more IL-10 than other B cell subsets, which were reduced by the addition of wortmannin (Fig. 6b). We examined the effect of Bregs on M2 macrophage polarization by co-culturing BMDMs collected from femurs with Bregs for 72 h in the presence of LPS, PMA, and ionomycin, followed by analysis with flow cytometry. We found that the Bregs co-culture group showed increased M2 macrophage differentiation compared to the other group, which was suppressed by 0.5µM wartmannin. (Fig. 6c). CD8+ T cells labeled with CFSE were cultured with Bregs and other B cells, and CD8+ T cell proliferation was analyzed using flow cytometry. CD8+ T cells co-cultured with Bregs showed a decrease in the suppression of proliferation, while wartmannin attenuated the effects of Bregs on CD8+ T cells. (Fig. 6d).
The PI3K inhibitor wortmannin modifies tumor immunity via regulation of Bregs, suppressing peritoneal tumor growth
We used an immunocompetent mouse model of PM to analyze the effect of wortmannin, on Bregs and other immune cell infiltrates in the spleen, ascites, peritoneal tumors, and on peritoneal tumor growth. Wortmannin was administered intraperitoneally at the same dose (0.5 [mg/kg]/day, daily); no body weight loss was observed compared with the controls without wortmannin (n = 6 for each group). The wortmannin-treated group showed a significant reduction in ascites volume and peritoneal tumor weight compared with the control group (Fig. 7a). In the wortmannin-treated group, Breg frequency in ascites and peritoneal tumors was significantly lower than that in control mice (2.8 ± 1.5% vs. 8.5 ± 2.5%, p = 0.002, 3.5 ± 1.3% vs. 6.5 ± 2.8%, p = 0.03, respectively) (Fig. 7b). Wortmannin treatment significantly increased CD8+ T cell infiltration in the spleen, ascites, and peritoneal tumors (p = 0.02, p = 0.04, and p = 0.002, respectively) and significantly decreased M2 macrophage infiltration, which was particularly pronounced in peritoneal tumors (p < 0.001). (Fig. 7c).