Demographics
The patients’ mean age was 48.4 ± 11.0 years. In all of those patients, 6 (3.6%) patients got diabetes mellitus. Superior endplate fracture occurred in 78 (46.8%) patients. There were 6 (3.6%) fractures in T11, 66 (39.3%) in T12, 72 (42.8%) in L1, 24 (14.3%) in L2. The average preoperative BMI was 24.8 ± 2.9. In addition, preoperative TLICS was 6.5 ± 1.0, and LSC score was 5.9 ± 1.1. The mean follow-up time was 27.5 ± 1.7 months.
There were significant improvements in postoperative parameters compared with preoperative parameters, such as VAS, CA, VWA, AVH%, A/P%, PVH% (P < 0.001, respectively), UIVA (P = 0.02) (Fig. 1). However, no significant difference was found between preoperative LIVA and postoperative LIVA (P = 0.420).
Risk factors of postoperative KR before implant removal
The average correction loss before implant removal was 4.3 ± 3.0˚. Kyphosis reoccurred in 78 (46.4%) patients. There were significant between-group differences among the number of females, age, BMI (P < 0.001, respectively), the number of cases with preoperative AVH%, preoperative VAS, preoperative LSC (P = 0.001, respectively), diabetes mellitus (P = 0.009), preoperative PVH% (P = 0.007), and preoperative A/P% (P = 0.002) (Table 1)
Table 1 Comparison of preoperative parameters between KR Group and NKR Group
|
Characteristic
|
KR Group
|
NKR Group
|
t/χ2 value
|
P value
|
Implant removal
|
Before
|
After
|
Before
|
After
|
Before
|
After
|
Before
|
After
|
Female (n)
|
48 (28.6%)
|
72 (42.9%)
|
30 (17.6%)
|
6 (3.6%)
|
13.365
|
39.908
|
< 0.001
|
< 0.001
|
Age (year)
|
54.2 ± 11.1
|
52.1 ± 10.3
|
43.4 ± 8.0
|
40.8 ± 8.1
|
-7.164
|
-7.089
|
< 0.001
|
< 0.001
|
BMI
|
26.6 ± 2.5
|
26.0 ± 2.7
|
23.3 ± 2.3
|
22.3 ± 1.3
|
-8.815
|
-12.166
|
< 0.001
|
< 0.001
|
Diabetes (n)
|
6 (3.6%)
|
6 (3.6%)
|
0 (0%)
|
0 (0%)
|
7.179
|
2.947
|
0.009
|
0.178
|
Upper endplate injury (n)
|
42 (25%)
|
60 (35.7%)
|
36 (21.4%)
|
18 (10.7%)
|
3.221
|
5.487
|
0.88
|
0.021
|
Preoperative CA (°)
|
20.8 ± 10.4
|
19.6 ± 9.1
|
19.1 ± 5.1
|
20.5 ± 4.9
|
-1.346
|
0.828
|
0.181
|
0.409
|
Preoperative RA (°)
|
9.4 ± 5.9
|
8.9 ± 5.6
|
9.3 ± 4.4
|
10.2 ± 3.8
|
-0.155
|
166
|
0.877
|
0.110
|
Preoperative VWA (°)
|
17.9 ± 6.8
|
17.6 ± 6.2
|
18.4 ± 4.4
|
19.3 ± 4.1
|
128.009
|
2.112
|
0.573
|
0.036
|
Preoperative UIVA (°)
|
-3.4 ± 2.4
|
-3.1 ± 3.0
|
-4.0 ± 3.5
|
-5.2 ± 2.6
|
-1.250
|
-4.552
|
0.213
|
< 0.001
|
Preoperative LIVA (°)
|
-4.9 ± 2.3
|
-5.1 ± 2.5
|
-5.5 ± 2.6
|
-5.5 ± 2.4
|
-1.475
|
-0.901
|
0.142
|
0.369
|
Preoperative AVH% (°)
|
55.6 ± 11.8
|
57.0 ± 10.7
|
63.0 ± 10.9
|
65.0 ± 12.5
|
4.222
|
166
|
< 0.001
|
< 0.001
|
Preoperative PVH% (°)
|
84.1 ± 5.8
|
84.6 ± 5.2
|
86.1 ± 2.9
|
86.2 ± 2.7
|
2.745
|
2.607
|
0.007
|
0.010
|
Preoperative A/P% (°)
|
53.3 ± 11.6
|
54.4 ± 10.3
|
58.3 ± 8.9
|
59.4 ± 10.2
|
3.118
|
2.994
|
0.002
|
0.003
|
Preoperative VAS
|
6.5 ± 0.8
|
6.4 ± 0.7
|
6.1 ± 0.7
|
6.0 ± 0.8
|
-3.259
|
-2.688
|
0.001
|
0.008
|
Preoperative TLICS
|
6.5 ± 0.6
|
6.5 ± 1.2
|
6.5 ± 1.3
|
6.4 ± 0.5
|
0.034
|
-0.224
|
0.973
|
0.823
|
Preoperative LSC
|
6.2 ± 1.0
|
6.1 ± 1.2
|
5.7 ± 1.2
|
5.7 ± 1.1
|
-3.355
|
-2.079
|
0.001
|
0.039
|
Notice: CA, Cobb angle; RA, regional angle; VWA, vertebral wedge angle; UIVA, upper intervertebral angle; LIVA, lower intervertebral angle; AVH%, anterior vertebra height ratio; PVH%, posterior vertebra height ratio; A/P%, anteroposterior ratio; VAS, visual analogue scale; TLICS, thoracolumbar injury classification and severity score; LSC, load-sharing classification.
|
Results of multivariable logistic regression analysis showed two significant risk factors of KR before implant removal: age and BMI (P < 0.001, respectively). However, preoperative AVH% (P = 0.008) and preoperative PVH% (P = 0.008) were protective factors for KR before implant removal (Table 2).
The ROC curve and the area under the curve (AUC) were used to analyze the predictability of the risk factors. The results showed that age (threshold value = 49.0, AUC = 0.828), BMI (threshold value = 24.0, AUC = 0.846) were good predictors; however, the predictabilities of preoperative AVH% (threshold value = 49.5, AUC = 0.348) and preoperative PVH% (threshold value = 85.5, AUC = 0.423) were unsatisfactory (Table 3).
Risk factors of postoperative KR after implant removal
The average correction loss after implant removal was 7.5 ± 4.4˚. Kyphosis reoccurred in 114(67.9%) patients (Fig. 2). There were significant differences between KR group and NKR group in the number of females, age, BMI (P < 0.001, respectively), the number of cases with preoperative AVH%, preoperative UIVA, preoperative AVH% (P < 0.001, respectively), diabetes mellitus (P = 0.021), preoperative A/P% (P = 0.002), preoperative VAS (P = 0.008), preoperative PVH% (P = 0.007), preoperative PVH% (P = 0.010), preoperative A/P% (P = 0.003), and preoperative LSC (P = 0.039) (Table 1).
Figure 2 Radiographs of 39 years old male patient was obtained. (a) The preoperative Cobb angle (CA) of fractured segments was 27°. (b) one week after operation, the CA was 8°. (c) CA was 11° before the operation. (d) CA was 21° at last follow-up.
Results of multivariable logistic regression analysis showed two significant risk factors of KR after implant removal: BMI (P < 0.001) and AVH% (P = 0.008). However, preoperative AVH% was a protective factor for KR after implant removal (Table 2).
Table 2 Results of Logistic regression analyzing risk factors of kyphosis recurrence
|
Implant removal
|
Characteristics
|
B value
|
Wals value
|
P value
|
Exp (B) value
|
Nagelkerke R2 value
|
Total percent
|
Before
|
Age
|
0.391
|
16.331
|
< 0.001
|
1.479
|
0.867
|
96.4
|
|
BMI
|
2,522
|
16.388
|
< 0.001
|
12.459
|
|
|
|
Preoperative AVH%
|
-0.2038
|
6.954
|
0.008
|
0.816
|
|
|
|
Preoperative PVH%
|
-1.246
|
11.207
|
0.001
|
0.288
|
|
|
|
Constant
|
35.044
|
3.765
|
0.052
|
1.657x1015
|
|
|
After
|
BMI
|
0.811
|
33.287
|
< 0.001
|
2.250
|
0.611
|
82.1
|
|
Preoperative AVH%
|
-0.086
|
15.687
|
< 0.001
|
0.918
|
|
|
|
Constant
|
-13.380
|
18.419
|
< 0.001
|
< 0.001
|
|
|
Notice: BMI, body mass index; AVH%, anterior vertebra height ratio; PVH%, posterior vertebra height ratio.
|
The ROC cures showed that BMI (threshold value = 25.17, AUC = 0.871) was a good predictor; however, the predictability of preoperative AVH% (threshold value = 61.5, AUC = 0.317) was unsatisfactory (Table 3).
Table 3 Results of ROC curve analyzing risk factors of kyphosis recurrence
|
Implant removal
|
Characteristics
|
Area under the curve
|
Cut-off value
|
Sensitivity
|
1-specificity
|
Youden index
|
Before
|
Age
|
0.828
|
49
|
0.769
|
0.200
|
0.646
|
|
BMI
|
0.846
|
24.0
|
0.846
|
0.267
|
0.579
|
|
Preoperative AVH%
|
0.348
|
49.5
|
0.615
|
0.933
|
-0.318
|
|
Preoperative PVH%
|
0.423
|
85.5
|
0.385
|
0.667
|
-0.282
|
After
|
BMI
|
0.871
|
25.17
|
0.684
|
< 0.001
|
0.684
|
|
Preoperative AVH%
|
0.317
|
61.5
|
0.3685
|
0.778
|
-0.409
|
Notice: BMI, body mass index; AVH%, anterior vertebra height ratio; PVH%, posterior vertebra height ratio.
|