CXCL9 Overexpression Predicts Better HCC Response to Anti-PD-1 Therapy and Promotes N1 Polarization of Neutrophils

Figures & data

Table 1 Baseline Characteristics of Patients with Different Serum CXCL9 Levels

Characteristic	<ULC	ULC~2×ULC	>2×ULC	P value
n	62	38	23
Age, years, mean ± sd	64.581 ± 8.8068	62.579 ± 9.5171	62.304 ± 10.412	0.457
Tumor burden, cm, n (%)				0.286
<10	35 (28.5%)	21 (17.1%)	17 (13.8%)
>10	27 (22%)	17 (13.8%)	6 (4.9%)
Tumor number, n (%)				0.776
Multiple	19 (15.4%)	10 (8.1%)	8 (6.5%)
Single	43 (35%)	28 (22.8%)	15 (12.2%)
Sex, n (%)				0.153
Male	51 (41.5%)	36 (29.3%)	21 (17.1%)
Female	11 (8.9%)	2 (1.6%)	2 (1.6%)
BCLC stage, n (%)				0.060
B	17 (13.8%)	7 (5.7%)	1 (0.8%)
C	45 (36.6%)	31 (25.2%)	22 (17.9%)
Metastasis, n (%)				0.580
No	45 (36.6%)	24 (19.5%)	15 (12.2%)
Yes	17 (13.8%)	14 (11.4%)	8 (6.5%)
Child‒Pugh class, n (%)				0.733
A	61 (49.6%)	37 (30.1%)	23 (18.7%)
B	1 (0.8%)	1 (0.8%)	0 (0%)
ECOG, n (%)				0.423
1	15 (12.2%)	12 (9.8%)	5 (4.1%)
0	45 (36.6%)	23 (18.7%)	18 (14.6%)
2	2 (1.6%)	3 (2.4%)	0 (0%)
HBsAg, n (%)				0.832
Positive	50 (40.7%)	32 (26%)	18 (14.6%)
Negative	12 (9.8%)	6 (4.9%)	5 (4.1%)
HBV-DNA, n (%)				0.700
Positive	29 (23.6%)	21 (17.1%)	12 (9.8%)
Negative	33 (26.8%)	17 (13.8%)	11 (8.9%)
AFP, ng/mL, n (%)				0.633
20~400	9 (7.3%)	6 (4.9%)	2 (1.6%)
>400	34 (27.6%)	16 (13%)	11 (8.9%)
<20	19 (15.4%)	16 (13%)	10 (8.1%)
CEA, ng/mL, median (IQR)	2.2 (1.6, 3.925)	2.75 (2.2, 4.475)	3.1 (1.7, 4.65)	0.262
TB, μmol/L, median (IQR)	14.6 (10.575, 19.075)	14.35 (10.725, 20.575)	16.1 (11.85, 22.85)	0.539
ALB, g/L, mean ± sd	37.968 ± 4.089	38.342 ± 4.6631	41.217 ± 5.0178	0.011
SII, median (IQR)	791.5 (447.62, 1350.9)	614.33 (431.51, 957.33)	442.93 (305.5, 959)	0.050
PLR, median (IQR)	165.78 (118.94, 230.74)	164.44 (98.306, 198.94)	114 (83.872, 154.2)	0.009
NLR, median (IQR)	5.3667 (3.8409, 7.4875)	4.5528 (3.7056, 6.9886)	4.1429 (2.2873, 5.9621)	0.238
MLR, median (IQR)	1.1143 (0.8264, 1.4794)	0.88167 (0.64479, 1.0795)	0.85556 (0.59509, 1.1511)	0.017

Notes: Categorical variables are summarized as n (%). Continuous variables are summarized as medians (ranges).

Abbreviations: AFP, α-fetoprotein; ALB, albumin; BCLC, Barcelona Clinic Liver Cancer; CEA, carcinoembryonic antigen; ECOG PS, Eastern Cooperative Oncology Group performance status; HBsAg, hepatitis B surface antigen; IQR, interquartile range; MLR, monocyte-to-lymphocyte ratio; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; TB, total bilirubin; ULC, upper limit of the cutoff value.

Figure 1 Patients with higher serum CXCL9 levels show a better treatment response to anti-PD-1 therapy. (A) Volcano plots showing the genes differentially expressed between responders and nonresponders. The most significantly expressed gene, CXCL9, is highlighted in red. (B) Serum CXCL9 levels in patients with PR, SD and PD HCC. (C) ROC curves for the response to anti-PD-1 therapy plotted according to the serum CXCL9 concentration. (D) Sankey analysis of serum CXCL9 levels and tumor response according to Response Evaluation Criteria in Solid Tumors (RECIST) v1.1. (E) PFS of patients with a serum CXCL9 concentration < ULC, 1~2x ULC or >2x ULC.

Abbreviations: PD-1, programmed death-1; PR, partial response; SD, stable disease; PD, progressive disease; HCC, hepatocellular carcinoma; ROC, receiver operating characteristic; AUC, area under curve; ULC, upper limit of the cutoff value.

Table 2 Variables Associated with the Response to Anti-PD-1 Therapy in the Training Cohort

Characteristic	Total(N)	Univariate Analysis		Multivariate Analysis
		Hazard Ratio (95% CI)	P value	Hazard Ratio (95% CI)	P value
Serum CXCL9, pg/mL	123
<ULC	62	Reference		Reference
ULC~2×ULC	38	3.407 (0.879–13.208)	0.076	3.909 (0.971–15.742)	0.055
>2×ULC	23	8.266 (2.365–28.892)	< 0.001	7.902 (2.117–29.494)	0.002
Age, years	123	0.989 (0.947–1.033)	0.627
Tumor burden, cm	123
<10	73	Reference		Reference
>10	50	0.428 (0.179–1.021)	0.056	0.245 (0.087–0.688)	0.008
Tumor number	123
Multiple	37	Reference
Single	86	1.005 (0.450–2.248)	0.989
Sex	123
Male	108	Reference
Female	15	0.914 (0.214–3.898)	0.903
BCLC stage	123
B	25	Reference
C	98	2.000 (0.684–5.848)	0.206
Metastasis	123
No	84	Reference
Yes	39	1.468 (0.607–3.553)	0.394
ECOG PS	123
1	32	Reference
0	86	2.030 (0.698–5.900)	0.194
2	5	0.000 (0.000 - Inf)	0.998
HBsAg	123
Positive	100	Reference
Negative	23	0.732 (0.274–1.956)	0.533
HBV-DNA	123
Positive	62	Reference
Negative	61	0.630 (0.291–1.364)	0.241
AFP, ng/mL	123
<20	45	Reference
20~400	17	0.921 (0.253–3.353)	0.901
>400	61	1.561 (0.695–3.504)	0.281
CEA, ng/mL	123	0.883 (0.730–1.067)	0.198
TB, μmol/L	123	0.994 (0.960–1.029)	0.726
ALB, g/L	123	1.090 (1.005–1.182)	0.038	1.026 (0.932–1.130)	0.598
SII	123	0.998 (0.996–1.000)	0.017	0.999 (0.997–1.002)	0.648
PLR	123	0.985 (0.976–0.995)	0.002	0.987 (0.976–0.999)	0.027
NLR	123	0.899 (0.783–1.031)	0.128
MLR	123	0.466 (0.173–1.252)	0.130

Note: Bold font, P<0.05.

Abbreviations: AFP, α-fetoprotein; ALB, albumin; BCLC, Barcelona Clinic Liver Cancer; CEA, carcinoembryonic antigen; CI, confidence interval; ECOG PS, Eastern Cooperative Oncology Group performance status; HBsAg, hepatitis B surface antigen; MLR, monocyte-to-lymphocyte ratio; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; TB, total bilirubin; ULC, upper limit of the cutoff value.

Table 3 Comparison of the Baseline Characteristics Between the Training and Validation Cohorts

Characteristic	Training Cohort	Validation Cohort	P value
n	123	60
Serum CXCL9, pg/mL, n (%)			0.133
>2×ULC	23 (12.6%)	18 (9.8%)
ULC~2×ULC	38 (20.8%)	20 (10.9%)
<ULC	62 (33.9%)	22 (12%)
Age, years, median (IQR)	65 (57.5, 70)	64 (57, 68)	0.372
Tumor burden, cm, n (%)			0.239
<10	73 (39.9%)	41 (22.4%)
>10	50 (27.3%)	19 (10.4%)
Tumor number, n (%)			0.656
Multiple	37 (20.2%)	20 (10.9%)
Single	86 (47%)	40 (21.9%)
Sex, n (%)			0.162
male	108 (59%)	48 (26.2%)
female	15 (8.2%)	12 (6.6%)
BCLC stage, n (%)			0.473
C	98 (53.6%)	45 (24.6%)
B	25 (13.7%)	15 (8.2%)
Metastasis, n (%)			0.350
Yes	39 (21.3%)	15 (8.2%)
No	84 (45.9%)	45 (24.6%)
Child‒Pugh class, n (%)			0.814
A	121 (66.1%)	60 (32.8%)
B	2 (1.1%)	0 (0%)
ECOG PS, n (%)			0.302
1	33 (18%)	14 (7.7%)
0	86 (47%)	46 (25.1%)
2	4 (2.2%)	0 (0%)
HBsAg, n (%)			0.050
Positive	100 (54.6%)	41 (22.4%)
Negative	23 (12.6%)	19 (10.4%)
HBV-DNA, n (%)			0.185
Positive	62 (33.9%)	24 (13.1%)
Negative	61 (33.3%)	36 (19.7%)
AFP, ng/mL, n (%)			0.072
20~400	17 (9.3%)	11 (6%)
<20	45 (24.6%)	30 (16.4%)
>400	61 (33.3%)	19 (10.4%)
CEA, ng/mL, median (IQR)	2.7 (1.8, 4.35)	2.4 (1.775, 4.825)	0.728
TB, μmol/L, median (IQR)	14.7 (10.75, 20.85)	14.45 (10.475, 21.575)	0.664
ALB, g/L, mean ± sd	38.691 ± 4.5828	40.067 ± 4.8812	0.064
SII, median (IQR)	655.5 (403.1, 1083.8)	755.92 (386.23, 1297.2)	0.650
PLR, median (IQR)	154.55 (107.67, 218.04)	138.88 (90.929, 248.75)	0.418
NLR, median (IQR)	5.1667 (3.75, 6.9773)	4.0455 (2.7612, 7.3143)	0.117
MLR, median (IQR)	0.98571 (0.72111, 1.3667)	0.67222 (0.39417, 1.2)	< 0.001

Notes: Categorical variables are summarized as n (%). Continuous variables are summarized as medians (ranges).

Abbreviations: AFP, α-fetoprotein; ALB, albumin; BCLC, Barcelona Clinic Liver Cancer; CEA, carcinoembryonic antigen; ECOG PS, Eastern Cooperative Oncology Group performance status; HBsAg, hepatitis B surface antigen; IQR, interquartile range; MLR, monocyte-to-lymphocyte ratio; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; TB, total bilirubin; ULC, upper limit of the cutoff value.

Figure 2 Establishment of the nomogram for predicting the response to anti-PD-1 therapy (A) Nomogram for the prediction of the response of HCC to anti-PD-1 antibody in the training cohort. B. ROC curves for the nomogram in the training cohort (B) and external validation cohort (E). Decision curves for the nomogram in the training cohort (C) and external validation cohort (F). Calibration curves for the nomogram in the training cohort (D) and validation cohort (G).

Abbreviations: ROC, receiver operating characteristic; AUC, area under curve; SII, immune-inflammation index; PLR, platelet-to-lymphocyte ratio.

Table 4 Incidence of Response to Anti-PD-1 Therapy in the High-Risk Group and Low-Risk Group in the Training Cohort and Validation Cohort

Cohort	Subgroup	Response	P value
Training cohort	High response group	24(80)	P < 0.001
	Low response group	4(4.3)
Validation cohort	High response group	12(52.2)	P = 0.004
	Low response group	6(16.2)

Note: Variables are summarized as n (%).

Figure 3 Tumor-derived CXCL9 promotes N1 polarization of neutrophils through CXCR3. (A and B) qPCR analysis of the transcription of N1 and N2 markers in neutrophils treated with or without recombinant human CXCL9 and IFN-α. (C) qPCR analysis of the transcription of N1 markers in neutrophils treated with or without AMG487. (D) Schematic diagram of the coculture of neutrophils and patient-derived HCC cells. (E) qPCR analysis of the transcription of N1 markers in neutrophils cocultured with designated patient-derived HCC cells. (F) Representative mIHC images of stained CD206⁺CD16b⁺ cells and iNOS⁺CD16b⁺ cells from the high- and low-CXCL9 groups. *P < 0.05; **P<0.01; ***P<0.001.

Abbreviation: ns, not significant.