A total of 60 patients with rectal cancer treated in Maria Skłodowska-Curie National Research Institute of Oncology in Warsaw between 08.2017 and 12.2020 were prospectively enrolled in the study. Forty-three males and seventeen females were included. The median age was 66.5 years (range, 29-89 years old). Distribution of cancer stages were as followed: stage II-IIIA - 8 (13%), stage IIIB – 41 (68%) and stage IIIC – 10 (17%). Stage of one of the patients remained undefined. There were no stages I or IV. Characteristics of patients is presented in table 1.
Table 1. Characteristics
|
All patients (n=60)
|
Age (years), median (range)
|
66.5 (29-89)
|
Sex, n (%)
|
Male
|
43 (71.7)
|
Female
|
17 (28.3)
|
Tumor, n (%)
|
|
T3
|
55 (91.7)
|
T4
|
5 (8.3)
|
Lymph nodes, n (%)
|
N0
|
8 (13.3)
|
N1
|
35 (58.3)
|
N2
|
16 (26.7)
|
Nx
|
1 (1.7)
|
Grade, n (%)
|
G1
|
2 (3.3)
|
G2
|
42 (70)
|
G3
|
2 (3.3)
|
Gx
|
14 (23.3)
|
Stage, n (%)
|
|
II-IIIA
|
8 (13.3)
|
IIIB
|
41 (68.3)
|
IIIC
|
10 (16.7)
|
Time between measurements
(days), median (range)
|
1st-2nd
|
9 (1-42)
|
2nd-3rd
|
11 (1-34)
|
1st-3rd
|
21 (7-55)
|
The median value of LMR in three consecutive measurements was 2.71 (range 1.11-6.42), 2.93 (0.24-7.32) and 2.8 (0.87-6.98) (p = 0.766). Median value of NLR was 2.71 (1.17-7.58), 2.84 (0.81-8.96) and 2.47 (1.04-10.22) respectively (p = 0.344). Medians of PLR were 150 (67-551), 141 (54-479) and 141 (67-430) (p = 0.627). All median and mean values of lymphocyte, monocyte, neutrophil and platelet count as well as their ratios are shown in table 2. The results of three measurements of LMR, NLR and PLR are illustrated in appendix fig.1.
Table 2. Median values of measurements
X
|
1st measurement
|
2nd measurement
|
3rd measurement
|
All measurements (mean)
|
p value1
|
ALC (10^9/l), median (range)
|
1.72 (0.70-3.79)
|
1.65 (0.69-4.02)
|
1.67 (0.52-3.92)
|
1.67 (0.52-4.02)
|
0.541
|
ALC (10^9/l), mean (SD)
|
1.84 (0.69)
|
1.87 (0.76)
|
1.86 (0.73)
|
1.86 (0.69)
|
|
AMC (10^9/l), median (range)
|
0.61 (0.30-1.30)
|
0.64 (0.27-5.26)
|
0.64 (0.33-1.21)
|
0.63 (0.27-5.26)
|
0.800
|
AMC (10^9/l), mean (SD)
|
0.68 (0.25)
|
0.77 (0.65)
|
0.68 (0.24)
|
0.71 (0.30)
|
|
ANC (10^9/l), median (range)
|
4.89 (2.42-12.36)
|
5.10 (2.11-12.27)
|
4.43 (2.35-13.69)
|
4.84 (2.11-13.69)
|
0.770
|
ANC (10^9/l), mean (SD)
|
5.43 (2.05)
|
5.26 (2.00)
|
5.02 (2.03)
|
5.24 (1.91)
|
|
Platelets (10^9/l), median (range)
|
273.00 (116.00-666.00)
|
254.00 (149.00-607.00)
|
269.00 (152.00-601.00)
|
264.00 (116.00-666.00)
|
0.198
|
Platelets (10^9/l), mean (SD)
|
293.00 (109.60)
|
284.00 (94.12)
|
291.00 (101.82)
|
289.00 (99.16)
|
|
LMR, median (range)
|
2.71 (1.11-6.42)
|
2.95 (0.24-7.32)
|
2.8 (0.87-6.98)
|
2.89 (0.24-7.32)
|
0.766
|
LMR, mean (SD)
|
2.94 (1.22)
|
2.93 (1.25)
|
2.91 (1.18)
|
2.93 (1.11)
|
|
NLR, median (range)
|
2.71 (1.17-7.58)
|
2.84 (0.81-8.96)
|
2.47 (1.04-10.22)
|
2.65 (0.81-10.22)
|
0.344
|
NLR, mean (SD)
|
3.25 (1.47)
|
3.13 (1.56)
|
3.10 (1.80)
|
3.16 (1.47)
|
|
PLR, median (range)
|
150.00 (67.00-551.00)
|
141.00 (54.00-479.00)
|
141.00 (67.00-430.00)
|
142.00 (54.00-551.00)
|
0.627
|
PLR, mean (SD)
|
179.00 (95.69)
|
173.00 (91.15)
|
180.00 (95.01)
|
177.00 (87.90)
|
|
1p value calculated using the Friedman's test
Patients were divided into high and low groups according to baseline value of each SIR marker. The predetermined cut-offs were 2.6 for LMR, 3.0 for NLR and 150 for PLR. The number of patients who belonged to each group in each measurement is presented in table 3.
Table 3. LMR,NLR, PLR-high/low patients in each measurement.
|
Measurement
|
p1
|
1st
|
2nd
|
3rd
|
2nd vs. 1st
|
3rd vs. 1st
|
3rd vs. 2nd
|
n (%)
|
CI95 for %2
|
n (%)
|
CI95 for %2
|
n (%)
|
CI95 for %2
|
LMR
|
LMR-high (>2.6)
|
34 (56.7)
|
43.2-69.4
|
35 (58.3)
|
44.5-70.9
|
33 (55.0)
|
41.6-67.9
|
>0.999
|
0.791
|
0.607
|
LMR-low (≤2.6)
|
26 (43.3)
|
30.6-56.8
|
25 (41.7)
|
29.1-55.1
|
27 (45.0)
|
32.1-58.4
|
"True LMR-high / LMR-low" –
patients who remained in the same group
(LMR-high or LMR-low) after the 2nd / 3rd measurement
|
x
|
x
|
49 (81.7)
|
69.6-90.5
|
41 (68.3)
|
55.0-79.7
|
x
|
x
|
0.804
|
NLR
|
NLR-high (≥3.0)
|
23 (38.3)
|
26.1-51.8
|
27 (45.0)
|
32.1-58.4
|
21 (35.0)
|
23.1-48.4
|
0.455
|
0.754
|
0.146
|
NLR-low (<3.0)
|
37 (61.7)
|
48.2-73.9
|
33 (55.0)
|
41.6-67.9
|
39 (65.0)
|
51.6-76.9
|
"True NLR-high / NLR-low" –
patients who remained in the same group
(NLR-high or NLR-low) after the 2nd / 3rd measurement
|
x
|
x
|
44 (73.3)
|
60.3-83.9
|
41 (68.3)
|
55.0-79.7
|
x
|
x
|
0.146
|
PLR
|
PLR-high (≥150)
|
29 (48.3)
|
35.2-61.6
|
24 (40.0)
|
27.6-53.5
|
26 (43.3)
|
30.6-56.8
|
0.180
|
0.581
|
0.581
|
PLR-low (<150)
|
31 (51.7)
|
38.4-64.8
|
36 (60.0)
|
46.5-72.4
|
34 (56.7)
|
43.2-69.4
|
"True PLR-high / PLR-low" –
patients who remained in the same group
(PLR-high or PLR-low) after the 2nd / 3rd measurement
|
x
|
x
|
47 (78.3)
|
65.8-87.9
|
42 (70.0)
|
56.8-81.2
|
x
|
x
|
>0.999
|
1 Comparison between measurements with McNemar test
2 95% confidence interval (CI) for proportion based on binomial exact calculation
Over half of the patients - 56.7% (95% CI, 43.2-69.4%) were classified as LMR-high; 61.7% (95% CI, 48.2-73.9%) and 51.7% (95% CI, 38.4-64.8%) of patients were assigned to NLR-low and PLR-low group accordingly. After the second measurement 81.7% (95% CI, 69.6-90.5%) of patients belonged to the same group (LMR-high or LMR-low). In terms of NLR and PLR there was 73.3% (95% CI, 60.3-83.9%) and 78.3% (95% CI, 65.8-87.9%) of patients in the same group, respectively. Interestingly, after three measurements the percentage of patients who stayed in the same group (“true LMR, NLR, PLR-high or low”) was almost identical: 68.3% (95% CI, 55.0-79.7%) for LMR and NLR and 70.0% (95% CI, 56.8-81.2%) for PLR. For LMR, NLR and PLR there were no significant changes in percentage of patients classified as low/high between all 3 measurements p > 0.05 in all comparisons).
Mean percentage change between the third and the first measurement of lymphocytes, monocytes, neutrophils and platelets count ranged from -5.59% to 4.76% and the standard error from 2.0 to 3.9 (table 4.; fig. 1).
Table 4. Calculation as % change 3rd measurement vs 1st measurement
% change
|
n
|
Mean
|
Standard deviation
|
Standard error
|
Median
|
Minimum
|
Maximum
|
L
|
60
|
4,76
|
30,23
|
3,9
|
0,75
|
-60,61
|
92,86
|
M
|
60
|
3,88
|
24,39
|
3,1
|
4,78
|
-40,00
|
85,71
|
N
|
60
|
-5,59
|
20,57
|
2,7
|
-8,20
|
-47,70
|
43,66
|
WBC
|
60
|
-2,39
|
17,28
|
2,2
|
-3,86
|
-39,78
|
42,12
|
PLT
|
60
|
1,29
|
15,30
|
2,0
|
-0,70
|
-29,32
|
44,60
|
Cohen’s Kappa statistic for the extent of agreement between the 1st and the 2nd measurement for LMR was κ = 0.59 (95% CI, 0.39-0.79), p < 0.001. For NLR the Kappa was κ = 0.45 (95% CI, 0.22-0.68), p < 0.001 and for PLR κ = 0.53 (95% CI, 0.32-0.75), p < 0.001, meaning in all cases a moderate agreement between both measurements.
If LMR at the first time was out of the range of 2.2-3.0 (+/- 0.4 from the cut-off) the risk of misclassification in the second measurement, defined as an affiliation to a different (high or low) group than initially, dropped to 5.0% (95% CI, 1.0-13.9%). In case of NLR, when outside of the range of 2.5-3.5 (+/- 0.5) in the first test, it was 8.3% (95% CI, 2.8-18.4%) and in case of PLR outside of the range of 125-175 (+/- 25) 10.0% (95% CI, 3.8-20.5%). The distribution of misclassifications based on initial values of SIR markers are shown in fig. 2.
The analysis of correlation between the first and the third measurement of LMR, NLR and PLR was conducted. LMR values were correlated with the coefficient of 0.776 (p<0.00001). NLR and PLR were correlated with the coefficient of 0.696 (p<0.000089) and 0.751 (p<0.00001) (appendix fig. 2.).