Peritumoral pDC abundance in ICC patients correlates with clinicopathologic features
BDCA2-immunopositive pDCs were more abundant in peritumoral tissue than in intratumoral areas (48.3 ± 55.4 cells/spot vs. 39.3 ± 62.8 cells/spot, respectively; p < 0.05) (Figure 1A and B). The number of peritumoral pDCs was significantly positively correlated with carbohydrate antigen 19-9 and gamma-glutamyl transferase levels (p = 0.025 and p = 0.01, respectively), tumor size (p = 0.007), tumor number (p = 0.043), degrees of vascular/bile duct invasion and lymphatic metastasis (p = 0.045 and p = 0.018, respectively), and TNM stage (p = 0.002) (Table 2). By contrast, sex was the only clinical characteristic that correlated with the number of pDCs in intratumoral tissues.
Table 2. Correlation between intratumoral and peritumoral plasmacytoid dendritic cells (pDCs) and clinicopathologic characteristics in ICC (n=359 for intratumoral tissues, 322 for peritumoral tissues)
Clinicopathological indexes
|
Intratumoral pDCs
|
Peritumoral pDCs
|
Low
|
High
|
P
|
Low
|
High
|
P
|
Age(year)
Sex
HBsAg
CA199
AFP (ng/ml)
GGT (U/L)
Liver cirrhosis
Tumor size(cm)
Tumor number
Vascular/bile duct invasion
Lymphatic metastasis
Tumor encapsulation
Tumor differentiation
TNM stage
|
≤50
>50
Female
Male
Negative
Positive
≤36
>36
≤20
>20
≤54
>54
No
yes
≤5
>5
Single
Multiple
absence
present
no
yes
complete
none
I+II
III+IV
I
II+III+IV+IV
|
32
145
85
92
124
53
78
99
157
20
80
97
138
39
74
103
130
47
132
45
152
25
37
140
93
84
90
87
|
42
140
68
114
128
54
71
111
160
22
92
90
135
47
85
97
128
54
150
32
154
28
25
157
88
94
92
90
|
0.242
0.041
0.955
0.331
0.816
0.310
0.400
0.351
0.511
0.070
0.736
0.072
0.427
0.955
|
32
128
75
85
110
50
78
82
145
15
89
71
120
40
84
76
122
38
134
26
145
15
32
128
90
70
97
63
|
36
126
61
101
116
46
59
103
141
21
67
95
127
35
61
101
107
55
121
41
132
30
23
139
79
83
70
92
|
0.625
0.094
0.576
0.025
0.307
0.010
0.471
0.007
0.043
0.045
0.018
0.167
0.179
0.002
|
|
|
|
|
|
|
|
|
Abbreviations: AFP, alpha-fetoprotein; GGT, gamma glutamyl transferase; CA 19-9, carbohydrate antigen 19-9; TNM, tumor-node-metastasis.
Chi-square tests for all analyses.
Peritumoral pDC accumulation in ICC patients is a predictor of poor prognosis
At the time of the final follow-up examination, 68.2% (245/359) of the patients had died, and 49.6% (178/359) had experienced recurrence. The 1-, 3-, and 5-year rates were 65.7%, 42.3%, and 33.8%, respectively, for OS and 34.8%, 51.6%, and 58.7% for cumulative recurrence.
We next classified patients according to the number of peritumoral pDCs: those with ≤28 cells/spot were assigned to the pDCslow group, and those with >28 cells/spot were assigned to the pDCshigh group. Patients in the pDCslow group had significantly higher 1-, 3-, and 5-year OS rates than those in the pDCshigh group (75.9% versus 59.6%, 56.4% versus 32.4%, and 46.4% versus 24.9%, respectively) (Figure 1C). Patients in the pDCshigh group also showed higher cumulative recurrence rates at 1, 3, and 5 years than those in the pDCslow group (41.0% versus 29.0%, 64.7% versus 38.4%, and 69.5% versus 48.2%, respectively) (Figure 1C). Additionally, peritumoral pDCs were associated with OS and cumulative recurrence rates in patients with early-stage (TNM stage I) ICC (n = 182) and normal carbohydrate antigen 19-9 levels (≤36 ng/ml, n = 149) (Supplementary Figure 1). Of note, OS and cumulative recurrence rates were not associated with the number of intratumoral pDCs (Figure 1C).
In the univariate analysis, prolonged TTR and OS were associated with a lower number of peritumoral pDCs in addition to various clinicopathologic factors (Table 3). The multivariate analysis revealed that the abundance of peritumoral pDCs, along with tumor number, lymphatic metastasis, and tumor encapsulation, was an independent factor for OS (p = 0.002, hazard ratio [HR] = 1.55) and TTR (p = 0.01, HR = 1.54) (Table 3).
Table 3. Univariate and multivariate analyses of prognostic factors in ICC (n=359)
Variable
|
TTR
|
OS
|
HR(95%CI)
|
P
|
HR(95%CI)
|
P
|
Univariate analysis
Age, year (≤50 versus >50)
Sex (female versus male)
HBsAg (negative versus positive)
AFP, ng/ml (≤20 versus>20)
CA199(≤36 versus >36)
GGT,U/L (≤54 versus>54)
Liver cirrhosis (no versus yes)
Tumor size,cm(≤5 versus>5)
Tumor number (single versus multiple)
Microvascular/bile duct invasion (no versus yes)
Lymphatic metastasis (no versus yes)
Tumor encapsulation (complete versus none)
Tumor differentiation (I+II versus III+IV)
TNM stage (I versus II+III+IV)
Intra-pDCs (low versus high)
Peri-pDCs (low versus high)
Peri-Tregs (low versus high)
Peri-pDCs and Tregs ( both low vs. both high)
Multivariate analysis
HBsAg (negative versus positive)
CA199(≤36 versus >36)
GGT,U/L (≤54 versus>54)
Tumor size,cm(≤5 versus>5)
Tumor number (single versus multiple)
Microvascular/bile duct invasion (no versus yes)
Lymphatic metastasis (no versus yes)
Tumor encapsulation (complete versus none)
Tumor differentiation (I+II versus III+IV)
Peri-pDCs (low versus high)
Peri-Tregs (low versus high)
Peri-pDCs and Tregs ( both low vs. both high)
|
0.89(0.63-1.28)
1.04(0.77-1.39)
1.09(0.80-1.50)
1.17(0.76-1.79)
1.19(0.88-1.60)
1.35(1.00-1.81)
1.16(0.83-1.62)
1.38(1.02-1.86)
2.48(1.81-3.40)
1.02(0.70-1.48)
2.23(1.52-3.28)
1.91(1.20-3.04)
1.10(0.82-1.48)
1.92(1.42-2.59)
1.16(0.87-1.56)
1.77(1.29-2.44)
1.63(1.19-2.24)
2.22(1.49-3.31)
NA
NA
NA
1.05(0.75-1.46)
1.94(1.36-2.77)
NA
1.72(1.12-2.64)
1.96(1.18-3.25)
NA
1.54(1.11-2.14)
1.59(1.16-2.20)
1.96(1.30-2.94)
|
.551
.809
.577
.483
.258
.050
.396
.035
.000
.937
.000
.006
.519
.000
.314
.000
.003
.000
NA
NA
NA
.786
.000
NA
.014
.009
NA
.010
.004
.001
|
1.29(0.94-1.78)
1.07(0.83-1.38)
0.72(0.54-0.96)
0.84(0.56-1.26)
1.90(1.46-2.48)
1.92(1.48-2.48)
0.83(0.61-1.12)
1.65(1.28-2.14)
2.66(2.03-3.47)
1.48(1.10-1.97)
2.70(1.96-3.72)
1.46(1.01-2.11)
1.43(1.11-1.84)
2.78(2.14-3.61)
1.18(0.91-1.51)
1.87(1.42-2.46)
1.53(1.17-2.01)
2.17(1.55-3.03)
0.70(0.51-0.97)
1.35(1.01-1.82)
1.65(1.24-2.21)
1.13(0.84-1.51)
2.12(1.57-2.86)
1.28(0.91-1.79)
2.09(1.45-3.00)
1.58(1.04-2.42)
1.32(1.00-1.74)
1.55(1.17-2.05)
1.34(1.02-1.77)
1.74(1.23-2.47)
|
.116
.583
.025
.396
.000
.000
.226
.000
.000
.009
.000
.042
.005
.000
.208
.000
.000
.000
.033
.045
.001
.435
.000
.153
.000
.034
.051
.002
.036
.002
|
NOTE: Cox proportional hazards regression model.
Abbreviations: AFP, alpha-fetoprotein; GGT, gamma glutamyl transferase; CA 19-9, carbohydrate antigen 19-9; TNM, tumor-node-metastasis.
HR, hazard ratio; CI, confidential interval.
pDC accumulation is associated with Treg abundance in peritumoral tissues
To examine the association between pDCs and T cell-mediated immune responses, T lymphocytes in peritumoral tissues were immunostained (Figure 2A). The number of Foxp3+ Tregs was significantly positively correlated with the number of peritumoral pDCs (p < 0.001, R = 0.291) (Figure 2B). However, the numbers of CD3+, CD4+, and CD8+ lymphocytes were not correlated (Supplementary Table 1).
Combination of peritumoral pDC and Treg abundance for predicting ICC patient outcomes
As the numbers of pDCs and Tregs were correlated, we evaluated the prognostic value of these factors combined. We compared the prognoses of ICC patients categorized into three groups according to peritumoral cell abundance: pDCslow/Tregslow, pDCslow/Tregshigh or pDCshigh/Tregslow, and pDCshigh/Tregshigh. The 1-, 3-, and 5-year OS rates for the patients in the pDCshigh/Tregshigh group were 60.2%, 29.1%, and 21.3%, respectively, which were significantly lower than for those in the pDCslow/Tregslow group (74.9%, 56.8%, and 50.8%, respectively) (Figure 3A–C). Similarly, the 1-, 3-, and 5-year cumulative recurrence rates for patients in the pDCshigh/Tregshigh group (42.4%, 67.6%, and 73.0%, respectively) were significantly higher than for those in the pDCslow/Tregslow group (26.3%, 35.8%, and 43.5%, respectively).