Clinical subgroups by cluster analysis.
We sought to classify clinically relevant subgroups among IIM patients with infection and to identify the clinical manifestations associated with infection. Therefore, a total of 66 patients were included in the cluster analysis. The patients were divided into phenotypes based on an unsupervised cluster analysis.
Data from 66 patients (45 women and 21 men) with 50 variables were described. We further removed the samples with significant missing data, and 46 variables were finally entered into the unsupervised analysis. The patients were classified into three groups by cluster analysis (Fig. 1). The clinical features of each cluster are shown in Table 1. The clinical characteristics of the 3 clusters are summarized in Fig. 2.
Table 1
Clusters by partitioning around medoids cluster analysis
Variable
|
Cluster
|
p Value
|
1 (n = 12)
|
2 (n = 21)
|
3 (n = 33)
|
General information
|
|
|
|
|
Male, n (%)
|
2(17)
|
7(33)
|
12(36)
|
0.4479
|
Age, Mean ± SD, years
|
50.08 ± 11.81
|
45.38 ± 12.32
|
52.12 ± 14.44
|
0.2016
|
Disease course, Mean ± SD, months
|
21.75 ± 24.54
|
13.86 ± 28.78
|
20.85 ± 29.84
|
0.6326
|
BMI, Mean ± SD, kg/m2
|
23.6 ± 3.224
|
20.71 ± 6.063
|
22.69 ± 2.828
|
0.1231
|
Clinical manifestations
|
|
|
|
|
Fever, n (%)
|
1(8)
|
8(38)
|
4(12)
|
0.0356*
|
Cough, n (%)
|
9(75)
|
7(33)
|
10(30)
|
0.0199*
|
Expectoration, n (%)
|
5(42)
|
7(33)
|
6(18)
|
0.2212
|
Dyspnea, n (%)
|
7(58)
|
13(62)
|
16(48)
|
0.6014
|
Arthritis, n (%)
|
6(50)
|
12(57)
|
11(33)
|
0.2047
|
Myalgia, n (%)
|
2(17)
|
11(52)
|
15(45)
|
0.1203
|
Myasthenia, n (%)
|
4(33)
|
20(95)
|
28(85)
|
< 0.0001****
|
Rash, n (%)
|
5(42)
|
21(100)
|
9(27)
|
< 0.0001****
|
Heliotrope rash, n (%)
|
3(25)
|
18(86)
|
4(12)
|
< 0.0001****
|
"V" sign, n (%)
|
3(25)
|
15(71)
|
2(6)
|
< 0.0001****
|
Gottron papule, n (%)
|
4(33)
|
18(86)
|
4(12)
|
< 0.0001****
|
Mechanic’s hands, n (%)
|
5(42)
|
2(10)
|
1(3)
|
0.0019**
|
Periungual erythema, n (%)
|
3(25)
|
1(5)
|
0(0)
|
0.0076*
|
Moist, n (%)
|
1(8)
|
6(29)
|
3(9)
|
0.1154
|
Laboratory features
|
|
|
|
|
HGB, Mean ± SD
|
128.5 ± 11.46
|
120.1 ± 23.02
|
129.3 ± 14.37
|
0.1499
|
PLT, Mean ± SD
|
220.3 ± 83.41
|
173.8 ± 66.96
|
217.8 ± 107.70
|
0.1946
|
WBC, Mean ± SD
|
7.084 ± 2.209
|
5.834 ± 2.070
|
9.051 ± 3.358
|
0.0005**
|
PMN, Mean ± SD
|
5.019 ± 1.874
|
4.346 ± 1.726
|
6.862 ± 3.029
|
0.0016**
|
LYM, Mean ± SD
|
1.403 ± 0.639
|
0.841 ± 0.328
|
1.502 ± 0.638
|
0.0003***
|
MONO, Mean ± SD
|
0.552 ± 0.219
|
0.513 ± 0.276
|
0.561 ± 0.287
|
0.813
|
CD3, Mean ± SD
|
1001 ± 203.2
|
565.7 ± 283.9
|
954.7 ± 436.3
|
0.0026**
|
CD4, Mean ± SD
|
589.4 ± 152.5
|
307.9 ± 176.4
|
534.7 ± 279.7
|
0.0044**
|
CD8, Mean ± SD
|
384 ± 172.5
|
235.3 ± 160.6
|
357.5 ± 189.6
|
0.067
|
ALT, Mean ± SD
|
25.08 ± 13.76
|
46.24 ± 38.63
|
99.39 ± 85.55
|
0.0011**
|
AST, Mean ± SD
|
25.08 ± 11.45
|
59.14 ± 39.11
|
119.6 ± 137.6
|
0.0119*
|
CK, Mean ± SD
|
170.1 ± 266.8
|
269 ± 864.7
|
2558 ± 2501
|
< 0.0001****
|
LDH, Mean ± SD
|
238.7 ± 47.22
|
368.5 ± 144.4
|
601.1 ± 416.4
|
0.0012**
|
HBDH, Mean ± SD
|
184.3 ± 32.47
|
287.6 ± 110.5
|
483.3 ± 366.3
|
0.0021**
|
MYO, Mean ± SD
|
136.6 ± 219.6
|
161.2 ± 361.8
|
831.6 ± 824.3
|
0.0004***
|
CK-MB, Mean ± SD
|
3.975 ± 4.899
|
6.879 ± 16.06
|
93.66 ± 105.3
|
0.0002***
|
IgG, Mean ± SD
|
10.56 ± 3.208
|
14.71 ± 5.54
|
12.9 ± 5.803
|
0.1067
|
IgA, Mean ± SD
|
1999 ± 894.4
|
2537 ± 1099
|
2058 ± 917
|
0.164
|
IgM, Mean ± SD
|
1440 ± 656
|
1641 ± 765
|
1367 ± 815.9
|
0.4484
|
C3, Mean ± SD
|
0.9002 ± 0.094
|
0.8214 ± 0.175
|
0.8837 ± 0.167
|
0.2906
|
C4, Mean ± SD
|
0.2115 ± 0.076
|
0.2448 ± 0.097
|
0.2088 ± 0.057
|
0.2199
|
ESR, Mean ± SD
|
34.10 ± 25.77
|
43.70 ± 23.97
|
45.85 ± 27.00
|
0.4569
|
CRP, Mean ± SD
|
11.72 ± 15.68
|
9.935 ± 11.01
|
14.10 ± 22.44
|
0.7373
|
Autoantibody
|
|
|
|
|
ANA, positive, n (%)
|
10(83)
|
12(57)
|
25(76)
|
0.1999
|
Anti-Ro52 antibody, positive, n (%)
|
12(100)
|
11(52)
|
16(48)
|
0.0060**
|
Anti-MDA5 antibody, positive, n (%)
|
3(25)
|
14(67)
|
1(3)
|
< 0.0001****
|
Anti-Jo1 antibody, positive, n (%)
|
6(50)
|
0(0)
|
5(15)
|
0.001**
|
Prognosis
|
|
|
|
|
Infection, n (%)
|
5(42)
|
16(76)
|
12(36)
|
0.0139*
|
*Indicates statistical difference between three clusters, *P<0.05, **P<0.005, ***P<0.0005, ****P<0.0001.
Abbreviations:
HGB, Hemoglobin; PLT, platelet; WBC, White blood cell; PMN, polymorphonuclear neutrophils; LYM, lymphocytes; MONO, monocytes; IgG, immunoglobulin G; IgM, immunoglobulin M; IgA, immunoglobulin A; C3, Complement 3; C4, Complement 4; LDH, lactate dehydrogenase; HBDH, hydroxybutyrate dehydrogenase; MYO, myoglobin; CK-MB, creatinine kinase MB; ALT, alanine transaminase; AST, aspartate aminotransferase; CK, creatine kinase; ESR, erythrocyte sedimentation rate; CRP, C-reactive protein; ANA, antinuclear antibody; Ro52, anti-cytoplasmic ribonucleoprotein of 52 kDa; MDA5, Melanoma differentiation-associated gene 5; anti-Jo1 antibody, anti–histidyl-ARN-t- synthetase antibody.
The patients in Cluster 1 (n = 12; 18.2%) had moderate infection features (moderate-risk cluster), among whom 5 patients (41.6%) had infections. This cluster’s features were obvious mechanic’s hands (41.6% vs. 9.5% vs. 3.0%), periungual erythema (25.0% vs. 4.8% vs. 0%), and rarely fever (8.3% vs. 22.3% vs. 12.1%) and myasthenia (33.3% vs. 95.2% vs. 84.9%), as compared with the other two clusters. Among the 3 clusters, this cluster’s patients were more likely to have the lowest ALT, AST, CK, LDH, HBDH, MYO, and CK-MB levels and the highest anti-Ro52 antibody and anti-Jo1 antibody positive rates. Meanwhile, cluster 1 was accompanied by the highest numbers of CD3 and CD4.
The patients in Cluster 2 (n = 21; 31.8%) had high infection features (high-risk cluster), among whom 16 patients (76.2%) had infections. Almost all the patients in cluster 2 complained of rashes, and the rashes included heliotrope rashes (25% vs. 85.7% vs. 12.1%), “V” sign (25% vs. 71.4% vs. 6.1%), and Gottron papules (33.3% vs. 85.7% vs. 12.1%). Signs of fever and myasthenia were frequent in this group, and this group had a high percentage of anti-MDA5 antibody positive patients. In addition, most patients had decreased WBC, PMN, LYM, CD3, CD4 and anti-JO1 antibody-positive levels.
The patients in Cluster 1 (n = 33; 50%) had low infection features (low-risk cluster), among whom 12 patients (36.4%) had infections. Compared with clusters 1 and 2, this cluster infrequently had accompanying DM-typical rashes of any kind, including the heliotrope rash, “V” sign, Gottron papule, mechanic’s hands and periungual erythema. Meanwhile, anti-Ro52 antibodies and anti-MDA5 antibodies were also rare in the cluster 3 patients. However, this cluster has a high rate of ANA antibody positivity and high levels of ALT, AST, CK, LDH, HBDH, MYO, CK-MB, WBC, LYM, and PMN.
Infection features of the patients in different clusters.
We next analyzed the prevalence of infection in the 3 clusters (Fig. 2A). We found significant differences in the infection rate among the 3 groups. The patients in cluster 2 had a very high infection rate compared to the patients in the other two clusters (76.2% vs. 41.6% vs. 36.4%, p = 0.0139). The common infections in the three clusters of patients included viral, bacterial, fungal, mycoplasma, tuberculosis, and pneumocystis carinii infections. The patients in clusters 1 and 3 commonly had viral infections, followed by bacterial infections. Bacterial infection was the most common infection type in the cluster 2 patients (Fig. 2B). Meanwhile, fungi and Pneumocystis carinii were the common causes of infections in clusters 2 and 3. The patients in the 3 clusters often exhibited single infections rather than mixed infections (Fig. 2C). Of the 33 patients, 72.7% of the patients showed evidence of pulmonary infections, 9.1% had urinary tract infections, and 21.2% had uncertain site infections. Pulmonary infections were the most common infections in IIM patients (Fig. 2D).