Thirty patients with MM and 19 healthy controls were enrolled in this study, with significant accumulations of TMPs (p<0.0001), and CD138+MPs (p<0.0001) in patients compared to controls, while no significant difference in PMPs (p=0.07) among both groups, table (1).
Table 1: Differences of microparticles between patients and controls
|
patients
|
controls
|
p-value
|
TMPs
|
68.95±1.4
|
54.86±0.5
|
0.0001
|
PMPs
|
63.18±2.1
|
56.79±2.8
|
0.07
|
CD138+MPs
|
11.75±0.4
|
4.75±0.2
|
0.0001
|
TMPs; total microparticles, PMPs; platelet microparticles, data expressed as mean ±SE, independent sample t-test, and Mann Whitney U test for significance, p<0.05.
Furthermore, patients with MM showed statistically significant higher percentages of CD8+T cells (p=0.007), PD1+CD8+T cells (p<0.0001), and PD1+CD4+T cells (p<0.0001) compared to their controls, while healthy controls had a significantly higher CD4+T cells (p<0.0001) than patients as shown in table 2.
Table 2: differential accumulation of T lymphocytes and PD 1 expressing T lymphocytes among study groups.
|
Patients
|
Controls
|
p-value
|
CD8+T cells
|
21.2±0.66
|
18.1±0.9
|
0.007
|
CD4+T cells
|
43.1±1.3
|
49.8±0.5
|
0.0001
|
PD1+CD8+T cells
|
14.26±0.5
|
7.5±0.3
|
0.0001
|
PD1+CD4+T cells
|
12.7±0.4
|
8.3±0.6
|
0.0001
|
Data expressed as mean ±SE, independent sample t-test and Mann Whitney U test for significance, p<0.05
Among 30 patients with active multiple myeloma, 7, 17, and 6 patients were in stages I, II, & III respectively, with 11 patients achieved complete response to Bortezomib based combinations while 19 patients didn't achieve CR as illustrated in table 3
Table 3: Characteristics of 30 MM patients
Factor
|
Mean ±SE
|
Hb (g/dL)
|
9.25 ±0.3
|
Platelets
|
130.8±9.2
|
LDH (U/L)
|
932.7± 65.6
|
serum calcium (mg/dL)
|
11.7± 0.3
|
Total protein (g/L)
|
69.9±1.0
|
Serum albumin (g/L)
|
30.5±1.6
|
M protein (g/L)
|
43.5±2.3
|
Urea (mg/dL)
|
45.1±2.6
|
Creatinine (mg/dL)
|
1.9±0.2
|
Bone marrow plasma
|
41.8±3.0
|
Lytic bony lesions
|
≤1=8 patients
>1=22 patients
|
β2M (mg/mL)
|
4.4±0.4
|
ISS staging
stage I
stage II
stage III
|
7 (23.3%)
17 (56.7%)
6 (20%)
|
Response
CR
non CR
|
11/30
19/30
|
LDH; lactate dehydrogenase, Hb; hemoglobin, MM; multiple myeloma, β2M; beta2 microglobulin, ISS; international staging system, CR; complete response.
Table 4 described different correlations between MPs, immune cells, and prognostic factors, with positive correlations were reported between PMP and PD1+CD4+T cells, CD138+MPs and CD8+T cells, PD1+CD8+T cells and PD1+CD4+T cells, PD1+CD8+T cells were positively correlated with LDH, M-protein, bone marrow plasma percentage, and β2 microglobulin, and β2 microglobulin was also correlated with PD1+CD4+T cells.
Table 4: Correlations between MPs, PD1, and other prognostic factors of MM
|
TMPs
|
PMP
|
CD138+MP
|
PD1+CD8+
|
PD1+CD4+
|
CD8+
|
CD4+
|
TMPs
|
r
|
NA
|
.252
|
.149
|
-.120-
|
.149
|
+.547
|
-.466
|
p
|
|
.179
|
.432
|
.526
|
.432
|
.002
|
.009
|
PMP
|
r
|
.252
|
NA
|
.295
|
.248
|
+.496
|
.254
|
.123
|
p
|
.179
|
|
.113
|
.187
|
.005
|
.175
|
.516
|
CD138+MP
|
r
|
.149
|
.295
|
NA
|
-.032-
|
.229
|
+.514
|
.113
|
p
|
.432
|
.113
|
|
.868
|
.224
|
.004
|
.551
|
PD1+CD8+
|
r
|
-.120-
|
.248
|
-.032-
|
NA
|
+.457
|
.112
|
.332
|
p
|
.526
|
.187
|
.868
|
|
.011
|
.557
|
.073
|
PD1+CD4+
|
r
|
.149
|
+.496
|
.229
|
+.457
|
NA
|
.306
|
-.026
|
p
|
.432
|
.005
|
.224
|
.011
|
|
.100
|
.893
|
CD8
|
r
|
+.547
|
.254
|
+.514
|
.112
|
.306
|
NA
|
-.215
|
p
|
.002
|
.175
|
.004
|
.557
|
.100
|
|
.254
|
CD4
|
r
|
-.466
|
.123
|
.113
|
.332
|
-.026-
|
-.215
|
NA
|
p
|
.009
|
.516
|
.551
|
.073
|
.893
|
.254
|
|
LDH
|
r
|
-.161-
|
.099
|
.120
|
+.609
|
.153
|
.227
|
.088
|
p
|
.395
|
.602
|
.527
|
.000
|
.420
|
.227
|
.645
|
serum calcium
|
r
|
.165
|
.282
|
.027
|
.319
|
.183
|
.150
|
.049
|
p
|
.384
|
.131
|
.887
|
.085
|
.333
|
.430
|
.795
|
serum albumin
|
r
|
-.198-
|
-.252
|
-.204
|
-.296
|
-.190
|
-.191
|
.214
|
p
|
.294
|
.180
|
.279
|
.112
|
.316
|
.311
|
.255
|
M protein
|
r
|
-.130-
|
.188
|
.048
|
+.703
|
.160
|
.306
|
.234
|
p
|
.494
|
.320
|
.801
|
.000
|
.399
|
.100
|
.214
|
creatinine
|
r
|
-.068
|
.026
|
.227
|
.273
|
.200
|
.257
|
-.154
|
p
|
.721
|
.890
|
.228
|
.144
|
.288
|
.170
|
.416
|
bone marrow plasma
|
r
|
-.150-
|
.158
|
.314
|
+.559
|
.353
|
.097
|
.114
|
p
|
.429
|
.404
|
.091
|
.001
|
.056
|
.609
|
.548
|
β2M
|
r
|
-.146-
|
.304
|
.165
|
+.839
|
+.450
|
.081
|
.349
|
p
|
.441
|
.102
|
.384
|
.000
|
.013
|
.672
|
.059
|
total protein
|
r
|
.353
|
-.111
|
-.157
|
-.057
|
.242
|
.255
|
-.429
|
p
|
.056
|
.560
|
.406
|
.764
|
.198
|
.175
|
.018
|
r; Pearson correlation, p; p-value, MP; microparticles, β2M; beta2 microglobulin.
As shown in table 5, and figures (4, 5, & 6), patients with CR had significantly lower PMP, CD138+MP, PD1+CD8+T cells, PD1+CD4+T cells, and CD8+T cells compared with patients without CR, while no significant difference in TMP and CD4+T cells between response groups.
Table 5: impact of MPs and PD1 expressing lymphocytes on the response of 30 patients with MM
|
response
|
Mean ±SE
|
p-value
|
TMPs
|
CR
|
67.6±2.5
|
0.61
|
non CR
|
69.7±1.6
|
|
PMP
|
CR
|
54.4±3.1
|
0.002
|
non CR
|
68.3±2.1
|
|
CD138.MP
|
CR
|
10.3±0.5
|
0.001
|
non CR
|
12.7±0.4
|
|
PD1+CD8+T
|
CR
|
12.7±0.5
|
0.01
|
non CR
|
15.2±0.6
|
|
PD1+CD4+T
|
CR
|
10.9±0.4
|
0.0001
|
non CR
|
13.7±0.5
|
|
CD8+T
|
CR
|
19.0±1.1
|
0.01
|
non CR
|
22.4±0.7
|
|
CD4+T
|
CR
|
42.1±2.2
|
0.52
|
non CR
|
43.7±1.6
|
|
TMP; total microparticles, PMP; platelet microparticles, CR; complete response, PD1; programmed death-1, Mann Whitney test was applied for TMP, PMP, PD1+CD4+T, and CD4 and independent sample t-test was applied for the remaining variables.
Receiver Operating Characteristic (ROC) curve analysis has shown that among different MPs and immune cells, PMP, CD138+MP, PD1+CD8+T cells, PD1+CD4+T cells, and CD8+T have shown good accuracy in predicting the response to Bortezomib based combination in patients with MM as illustrated in table 6, and figure 7, the largest AUC was observed with PD1+CD4+T cells for predicting non CR among active myeloma patients (AUC=0.94, p=0.0001) with the highest sensitivity (92%), specificity (80%), and performance (72%).
Table 6: performance of different MPs and immune cells in MM
Variable
|
AUC
|
Cut off
|
Sensitivity
|
Specificity
|
Youden index
|
p-value
|
PMP
|
0.84
|
≥61
|
83
|
74
|
57
|
0.002
|
CD138 +MP
|
0.83
|
≥10.6
|
91
|
74
|
65
|
0.003
|
PD1+CD8+T cells
|
0.77
|
≥13.5
|
71
|
75
|
46
|
0.017
|
PD1+CD4+T cells
|
0.94
|
≥11.3
|
92
|
80
|
72
|
0.0001
|
CD8+T cells
|
0.79
|
≥20.1
|
83
|
70
|
53
|
0.008
|
AUC; area under the curve, PMP; platelet Microparticles, MP; Microparticles, PD1; programmed death-1
Differences of MPs and immune cells across ISS stages
One way Manova with Tukey test was applied after performing Levene's test of homogeneity to assess equality of variance for CD138+MP, CD8, PD1+CD8+T cells (p=0.3, 0.4, 0.2), and Box's test to assess equality of dependent covariance across all stages (p=0.316), generally there was a significant effect of all three variables together on ISS stage (Wilks' Lambda p-value=0.009); no significant differences were detected for CD138+MP (p=0.6) and CD8 (p=0.5), there was significant differences in PD1+CD8+T cells across different stages (p=0.001), table (7), figure (8). Independent sample Kruskal Wallis test was applied for other Microparticles and immune cells; no significant differences of the distributions of TMP, PMP, and CD4+T cells across different stages of MM patients (p=0.8, 0.6, 0.07 respectively), however significant difference in the distribution of PD1+CD4+T across ISS stages was detected (p=0.041) as shown in figure (9).
Table (7): multivariate analysis of normally distributed variables on ISS stage
Factor
|
Dependent
|
F-statistics
|
df
|
R2
|
p-value
|
ISS stage
|
CD138MP
PD1+CD8+T
CD8+T cells
|
0.428
9.467
0.524
|
2
2
2
|
0.34
0.412
0.037
|
0.6
0.001
0.5
|
df; degrees of freedom, ISS; international staging system, PD1; programmed death-1, one way Manova for significance, p<0.05
Logistic regression
Due to autocorrelations between MPs and immune cells, logistic regression with forward LR method was carried out, the overall prediction rate of the model for response to treatment was 90% as shown in table (8a), for each point of increase in PD1+CD4+T cells there was a decrease in the odds of CR by 15.5%, p=0.025, also for each point of increase in CD138+MP there was a decrease in the odds of CR by 47.2%, p=0.059, table (8b), furthermore other variables were removed from the model because of autocorrelation and no added value in the prediction rate of model.
Table (8a): model summary of logistic regression done for 30 patients with MM
Independent
variables
|
Dependent
variables
|
R2
|
Chi2
|
p-value
|
df
|
% of correct
classification
|
TMP
PMP
CD138MP
PD1CD8
PD1CD4
CD4
CD8
|
Response
|
Step1
0.653
|
19.487
|
<0.0001
|
1
|
Step 1=
83.3%
|
Step 2
0.77
|
5.509
|
=0.019
|
1
|
Step 2=
90%
|
Response is a binary variable with two categories; 0; non CR, 1; CR, df; degrees of freedom, p<0.05
Table (8b): Variables in the Equation of logistic regression (Ln (dependent= CR)=30.621-0.751*1-1.864*2)
|
|
B
|
S.E.
|
Wald
|
df
|
p-value
|
Exp (B)
|
95% C.I. for Exp(B)
|
Lower
|
Upper
|
Step 1a
|
PD1+CD4+T
|
-1.790-
|
.654
|
7.500
|
1
|
.006
|
.167
|
.046
|
.601
|
Constant
|
21.361
|
7.997
|
7.135
|
1
|
.008
|
1891728042.290
|
|
|
Step 2b
|
CD138+.MP
|
-.751-
|
.398
|
3.553
|
1
|
.059
|
.472
|
.216
|
1.030
|
PD1+CD4+T
|
-1.864-
|
.834
|
5.001
|
1
|
.025
|
.155
|
.030
|
.794
|
Constant
|
30.621
|
12.083
|
6.423
|
1
|
.011
|
19892982541237.910
|
|
|
a. Variable(s) entered on step 1: PD1+CD4+T cells, b. Variable(s) entered on step 2: CD138+MP, Exp(B); odds ratio, C.I.; confidence interval, df; degrees of freedom, p<0.05
|