3.1. Clinical Data. A total of 143 patients' data were collected. The median age of the patients was 56 years (range 18~83 years). The efficacy evaluation of the patients were complete response (CR) 0 cases (0%), partial response (PR) 90 cases (62.9%), stable disease (SD) 45cases (31.5%), progressive disease (PD) 8 cases (5.6%), objective response rate (ORR) was 62.9% and disease control rate (DCR) was 94.4%. The general clinical data of training and validation cohorts were listed in Table 1.
Table 1. The Clinical Characteristics of Patients with mCRC in the training and validation cohort.
Characteristics
|
Training cohort (n = 100)
|
Validation cohort (n = 43)
|
No. of patients
|
%
|
No. of patients
|
%
|
Gender
Male
Female
|
69
31
|
69.0
31.0
|
31
12
|
72.1
27.9
|
Age(year)
<60
≥60
x ± s
|
61
39
56.06 ± 12.30
|
61.0
39.0
|
27
16
54.98 ± 13.30
|
62.8
37.2
|
BMI(kg/m2)
<18.5
≥18.5
|
9
91
|
9.0
91.0
|
6
37
|
14.0
86.0
|
Site of primary tumor
Right colon
Left colon/Rectum
|
11
89
|
11.0
89.0
|
9
34
|
20.9
79.1
|
Resected primary tumor
Yes
No
|
79
21
|
79.0
21.0
|
33
10
|
76.7
23.3
|
Liver metastases
Yes
No
|
76
24
|
76.0
24.0
|
27
16
|
62.8
37.2
|
CEA(ng/mL)
<5
≥5
|
20
80
|
20.0
80.0
|
7
36
|
16.3
83.7
|
CA199(U/mL)
<37
≥37
|
57
43
|
57.0
43.0
|
33
10
|
76.7
23.3
|
CT regimen
FOLFOX/XELOX
FOLFIRI
|
58
42
|
58.0
42.0
|
27
16
|
62.8
37.2
|
First evaluation results
CR
PR
SD
PD
|
0
63
32
5
|
0
63.0
32.0
5.0
|
0
27
13
3
|
0
62.8
30.2
7.0
|
Abbreviations: CEA, carcino-embryonic antigen; CA199, carbohydrate antigen 19-9; CT, chemotherapy; CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease.
3.2. The Cut-off Value for NLR, PLR and SII. X-tile software was used to determine the optimal cut-off value for the NLR, PLR, and SII in the training cohort (cut-off at PFS=10.4 months). The optimal cut-off value of NLR, PLR and SII were calculated as 3.9, 152.2 and 464.3, respectively. Thus, patients were separated into low NLR group (NLR<3.9) (n=81), high NLR group (NLR≥3.9) (n=19), low PLR group (PLR<152.2) (n=57), high PLR group (PLR≥152.2) (n=43), low SII group (SII<464.3) (n=50) and high NLR group (SII<464.3) (n=50) (Figure 1).
3.3. Relationship Between Clinical Features and Baseline Blood Parameters. The Chi-square test revealed the distinction between the clinical features and baseline blood parameters in the training cohort. NLR, PLR and SII showed no differences regarding Gender, age, BMI, site of primary tumor, resected primary tumor, liver metastases, CEA, CA199, CT regimen and first evaluation results groups (P>0.05). However, NLR, PLR and SII revealed a significant distinction between different PFS groups (PFS<10.4 months vs PFS ≥10.4 months) (P≤0.001) (Table 2).
Table 2. Correlation between Hematological Parameters and Clinical Characteristics of Patients with mCRC by Chi-square Test.
Characteristics
|
NLR
|
|
PLR
|
|
SII
|
|
<3.9
|
≥3.9
|
P
|
<152.2
|
≥152.2
|
P
|
<464.3
|
≥464.3
|
P
|
Gender
Male
Female
|
56
25
|
13
6
|
0.952
|
41
16
|
28
15
|
0.466
|
35
15
|
34
16
|
0.829
|
Age(year)
<60
≥60
|
48
33
|
13
6
|
0.461
|
33
24
|
28
15
|
0.464
|
30
20
|
31
19
|
0.838
|
BMI(kg/m2)
<18.5
≥18.5
|
8
73
|
1
18
|
0.852
|
4
53
|
5
38
|
0.657
|
3
47
|
6
44
|
0.485
|
Site of primary tumor
Right colon
Left colon/Rectum
|
9
72
|
2
17
|
1.000
|
7
50
|
4
39
|
0.882
|
6
44
|
5
45
|
0.749
|
Resected primary tumor
Yes
No
|
65
16
|
14
5
|
0.750
|
46
11
|
33
10
|
0.630
|
43
7
|
36
14
|
0.086
|
Liver metastases
Yes
No
|
59
22
|
17
2
|
0.219
|
45
12
|
31
12
|
0.427
|
36
14
|
40
10
|
0.349
|
CEA(ng/mL)
<5
≥5
|
15
66
|
5
14
|
0.656
|
12
45
|
8
35
|
0.762
|
9
41
|
11
39
|
0.617
|
CA199(U/mL)
<37
≥37
|
46
35
|
11
8
|
0.930
|
37
20
|
20
23
|
0.066
|
30
20
|
27
23
|
0.545
|
CT regimen
FOLFOX/XELOX
FOLFIRI
|
49
32
|
9
10
|
0.297
|
37
20
|
21
22
|
0.107
|
33
17
|
25
25
|
0.105
|
First evaluation results
ORR(CR+PR)
SD+PD
|
51
30
|
12
7
|
0.987
|
38
19
|
25
18
|
0.382
|
34
16
|
29
21
|
0.300
|
PFS(months)
<10.4
≥10.4
|
32
49
|
17
2
|
<0.001
|
20
37
|
29
14
|
0.001
|
13
37
|
36
14
|
<0.001
|
Abbreviations: NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; CEA, carcino-embryonic antigen; CA199, carbohydrate antigen 19-9; CT, chemotherapy; ORR, objective response rate; CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease; PFS, progress-free survival.
3.4. Kaplan-Meier Analysis of Blood Parameters.Kaplan-Meier curve demonstrated that the PFS of the low NLR group (NLR<3.9) was significantly longer than that of the high NLR group (NLR≥3.9) (11.0 vs 5.1 months, P<0.001) (Figure 2A); PFS of low PLR group (PLR<152.2) was significantly longer than that of the high PLR group (PLR≥152.2) (11.5 vs 6.9 months, P<0.001) (Figure 2B); PFS of low SII group (SII<464.3) was significantly longer than that of the high NLR group (SII<464.3) (12.6 vs 6.9 months, P<0.001) (Figure 2C).
3.5. Cox Regression Analysis for the Prognostic Factors of PFS. Univariate analysis showed that the resected primary tumor, liver metastases, NLR, PLR and SII were significantly correlated with PFS. Multivariate analysis revealed that the resected primary tumor (HR: 0.551, 95%CI: 0.329-0.924, P=0.024), liver metastases (HR: 2.033, 95%CI: 1.212-3.407, P=0.007), NLR (HR: 2.596, 95%CI: 1.378-4.888, P=0.003), PLR (HR: 2.002, 95%CI: 1.235-3.246, P=0.005) and SII (HR: 2.202, 95%CI: 1.292-3.751, P=0.004) were independent prognostic factors of PFS (Table 3).
Table 3. Cox Univariate and Multivariate Analyses of Clinical Parameters for PFS Prediction in Training Cohort.
Variables
|
Univariate analysis (P)
|
Multivariate analysis
|
Hazard ratio
|
95%CI
|
P value
|
Gender
Male vs Female
|
0.374
|
|
|
|
Age(year)
<60 vs ≥60
|
0.808
|
|
|
|
BMI(kg/m2)
<18.5 vs ≥18.5
|
0.102
|
|
|
|
Site of primary tumor
Right colon vs Left colon/Rectum
|
0.534
|
|
|
|
Resected primary tumor
Yes vs No
|
0.002
|
0.551
|
0.329-0.924
|
0.024
|
Liver metastases
Yes vs No
|
0.014
|
2.033
|
1.212-3.407
|
0.007
|
CEA(ng/mL)
<5 vs ≥5
|
0.871
|
|
|
|
CA199(U/mL)
<37 vs ≥37
|
0.649
|
|
|
|
CT regimen
FOLFOX/XELOX vs FOLFIRI
|
0.572
|
|
|
|
First evaluation results
ORR(CR+PR) vs SD+PD
|
0.141
|
|
|
|
NLR
<3.9 vs ≥3.9
|
<0.001
|
2.596
|
1.378-4.888
|
0.003
|
PLR
<152.2 vs ≥152.2
|
<0.001
|
2.002
|
1.235-3.246
|
0.005
|
SII
<464.3 vs ≥464.3
|
<0.001
|
2.202
|
1.292-3.751
|
0.004
|
Abbreviations: NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; CEA, carcino-embryonic antigen; CA199, carbohydrate antigen 19-9; CT, chemotherapy; ORR, objective response rate; CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease; PFS, progress-free survival.
3.6. Construction and Evaluation of the Nomogram. A prognostic nomogram was developed using the resected primary tumor, liver metastases, NLR, PLR and SII. Each factor was allocated a certain point value. Patients' short PFS rate were evaluated based on the patient's total point value, which was determined using the prognostic nomogram (Figure 3). The ROC curves were plotted, and the results showed that the prognostic model has the best predictive performance (AUC=0.870) (Figure 4). The calibration curves showed that the nomogram had excellent stability and predictive value in both training (C-index=0.827) and validation cohort (C-index=0.870) (Figure 5). The DCA revealed that the prognostic model had the greatest net benefit in both the training cohort and validation cohorts when compared to the "treat all" and "treat none" strategies (Figure 6).