Characteristics of the selected participants at baseline
Altogether 629 participants were chosen for analysis of the final database after being screened by inclusion and exclusion criteria (see Figure 1 for flow chart). We show the baseline characteristics of these selected participants in Table 1, based on a dichotomy of whether they were multiply infected with HPV. Overall, the mean age of the 629 selected participants was 44.6 ± 10.6 years old. There were no statistically significant differences detectable for age,weight,NES,SCCA,ALT,AST,UA,CREA,BUN,Glucose,vagina cleanness, vaginal leucocytes,menopausal status,number of childbirth,number of miscarriages,TCT, among different multiplicity of HPV groups,but number of pregnancies(P value is 0.047).
Table 1 Baseline Characteristics of participants
Characteristics*
|
Single HPV infection (n = 375)
|
Multitype HPV infection (n = 153)
|
p value
|
Age, Median (IQR)
|
44.520 ± 10.077
|
43.843 ± 11.514
|
0.502
|
Vagina cleanness , n (%)
|
|
|
0.688
|
I~III grade
|
355 (95.430%)
|
140 (94.595%)
|
|
IV grade
|
17 (4.570%)
|
8 (5.405%)
|
|
Vaginal leukocytes, n (%)
|
|
|
0.192
|
+~+++
|
320 (85.791%)
|
121 (81.208%)
|
|
++++
|
53 (14.209%)
|
28 (18.792%)
|
|
Menopausal status, n (%)
|
|
|
0.565
|
premenopause
|
253 (71.875%)
|
104 (69.333%)
|
|
postmenopause
|
99 (28.125%)
|
46 (30.667%)
|
|
Number of pregnancies, n (%)
|
|
|
0.047
|
=0
|
10 (2.695%)
|
8 (5.229%)
|
|
1~3
|
222 (59.838%)
|
75 (49.020%)
|
|
≥4
|
139 (37.466%)
|
70 (45.752%)
|
|
Number of childbirth, n (%)
|
|
|
0.549
|
0
|
77 (20.924%)
|
26 (16.993%)
|
|
1~2
|
207 (56.250%)
|
88 (57.516%)
|
|
≥3
|
84 (22.826%)
|
39 (25.490%)
|
|
Number of miscarriages, n (%)
|
|
|
0.526
|
3≤
|
346 (94.278%)
|
142 (92.810%)
|
|
≥4
|
21 (5.722%)
|
11 (7.190%)
|
|
Tct, n (%)
|
|
|
0.471
|
ASC-US≤
|
120 (39.735%)
|
51 (43.590%)
|
|
≥LSIL
|
182 (60.265%)
|
66 (56.410%)
|
|
Weight, Median (IQR)
|
55.261 ± 9.250
|
55.082 ± 7.817
|
0.833
|
NSE, Median (IQR)
|
12.565 ± 3.347
|
13.114 ± 3.709
|
0.129
|
SCCA, Median (IQR)
|
0.484 ± 0.944
|
0.419 ± 0.205
|
0.42
|
ALT, Median (IQR)
|
18.463 ± 9.627
|
18.929 ± 11.312
|
0.632
|
AST, Median (IQR)
|
18.425 ± 6.164
|
18.661 ± 6.760
|
0.698
|
GLU, Median (IQR)
|
5.017 ± 0.594
|
4.906 ± 0.678
|
0.068
|
UA, Median (IQR)
|
286.877 ± 67.695
|
299.448 ± 64.100
|
0.05
|
CREA, Median (IQR)
|
62.517 ± 10.582
|
62.337 ± 10.127
|
0.858
|
BUN, Median (IQR)
|
4.463 ± 1.100
|
4.401 ± 1.184
|
0.562
|
*IQR = interquartile range;NSE=neuron-specific enolase;SCCA=neuron-specific enolase;ALT=alanine aminotransferase;AST=aspartate aminotransferase;UA=uric acid;GLU=glucose;CREA=Creatinine;BUN=urea nitrogen.
|
Table2 Number of HPV infection and related factors in single-factor analysis.
number of HPV infection
|
OR_95CI
|
P_value
|
Age
|
0.99 (0.98~1.01)
|
0.527
|
Vagina cleanness
|
1.43 (0.64~3.19)
|
0.386
|
Vaginal leukocytes
|
1.44 (0.89~2.33)
|
0.134
|
Menopausal status
|
1.11 (0.74~1.66)
|
0.607
|
Number of pregnancies
|
0.43 (0.17~1.08)
|
0.074
|
Number of childbirth
|
1.53 (0.87~2.67)
|
0.138
|
Number of miscarriages
|
1.31 (0.63~2.74)
|
0.466
|
Tct
|
0.87 (0.57~1.33)
|
0.529
|
Weight,
|
1 (0.98~1.02)
|
0.909
|
NSE
|
1.04 (0.98~1.1)
|
0.164
|
SCCA
|
0.79 (0.41~1.52)
|
0.476
|
ALT
|
1 (0.98~1.02)
|
0.829
|
AST
|
1 (0.97~1.03)
|
0.945
|
Glu
|
0.79 (0.59~1.07)
|
0.126
|
UA
|
1 (1~1)
|
0.114
|
CREA
|
0.99 (0.97~1.01)
|
0.354
|
BUN
|
0.99 (0.84~1.16)
|
0.904
|
Glutritertile1
|
1
|
|
Glutritertile2
|
0.66 (0.42~1.03)
|
0.067
|
Glutritertile3
|
0.49 (0.31~0.77)
|
0.002
|
Unvariate analysis
We present the outcomes of the univariate analysis in Table 2. Using univariate binary logistic regression, we observed that glucose was not associated with the diversity of HPV. But we found that glucose tretiles of high (0.49,0.31~0.77 vs ref) were adversely associated with multiplicity of HPV.
Results of unadjusted and adjusted binary logistic regression.
In this study, Three models were constructed to analyse the independent effect of glucose on HPV multiple infections (univariate and multi-variate binary logistic regression). Effects sizes (Odds ratio (OR)) and 95% confidence intervals are presented in Table 3. In the non-adjusted model (model 1), the effect size based on the model can be interpreted as the difference between 1 unit of glucose and the risk of multiple effect events. For instance, an unadjusted model with an effect size of 0.75 for the risk of multiple HPV infections meant that a 1-unit difference in glucose was associated with a 25% difference in the risk of multiple effect events (0.75, 95% CI 0.58 to 0.98). In the minimum Adjusted Model (Model 2), a 1 unit change in glucose was associated with a 19% increase in the risk of multiple effect events (0.81,95% CI 0.59 to 1.11) . Within the fully adjusted model (Model 3) ( adjustment for all covariates listed in Table 1), for each additional 1 unit of glucose, risk of event of multiple effection increased by decrease 16% (0.84, 95% CI 0.57~1.22).
Table 3 The results of univariate and multivariate analyses
|
non-adjusted model OR (95%CI)
|
Minimally-adjusted model OR (95%CI)
|
Fully-adjusted model OR (95%CI)
|
Glu
|
0.75 (0.58~0.98)
|
0.81 (0.59~1.11)
|
0.84 (0.57~1.22)
|
Glu(tritertile)
|
|
|
Q1
|
Ref
|
Ref
|
Ref
|
Q2
|
0.73 (0.49~1.07)
|
0.69 (0.46~1.02)
|
0.78 (0.45~1.33)
|
Q3
|
0.45 (0.3~0.68)
|
0.43 (0.28~0.65)
|
0.48 (0.27~0.84)
|
p for trend
|
<0.001
|
<0.001
|
0.01
|
CI: Confidence interval Non-adjusted model: we did not adjust any covariate Minimally-adjusted model: we only adjusted age and weight fully-adjusted model: we adjusted Age,Vagina cleanness,Vaginal leukocytes,Menopausal status,Number of pregnancies,Number of childbirth,Number of miscarriages,Tct,Weight, NSE,SCCA,ALT,AST,UA,CREA,BUN
|
We switched glucose from a continuous variable to a categorical variable ( triple quantile of glucose) for the purposes of sensitivity analysis, and the P for the trend in glucose in the fully adjusted model with a categorical variable was inconsistent with the results when glucose was a continuous variable(Table3). In addition, we found that the trends in effect sizes were non-equally spaced across glucose groups.
The results of nonlinearity of glucose and multiplicity of HPV
In the present study, we investigated the non-linear relationship between glucose and multiple HPV infections (Figure 2).The results of the smoothing curve and the generalised additive model showed a non-linear association with glucose and blood glucose after adjusting for age (years), weight (kg),NES(ug/ml),SCCA(ng/ml), ALT,AST,UA,CREA,BUN, vagina cleanness(I~II was defined as normal, and III~IV as abnormal), vaginal leucocytes, menopausal status( pre-menopause or post-menopause),number of pregnancies(0, 1-3, ≥4) ,number of childbirth(unfed females or one primiparity (1 delivery), multiparous (2 delivery) and grand multiparous (≥ 3 deliveries) and code with 0, 1and 2) , number of miscarriages( 0 denotes no event occurring and 1+ denotes one or more event) ,TCT(NILM or ASC-US code with 1,ASC-H or HSIL orLSIL code with 2 ). Both binary logistic regression and two-segment binary logistic regression were used to fitted the association and the best-fit model based on P was selected for the log-likelihood ratio test.
As the p for the log-likelihood ratio test was smaller than 0.05, we chose dichotomous binary logistic regression to fit the association between glucose and HPV multiplicity as it accurately represented the relationship. Using a two-segment binary logistic regression and recursive algorithm, we computed an inflection point of 5.4. In the left-hand side of the inflection point, the effect size and 95% CI were 0.379 (0.196, 0.732), respectively. On the right-hand side of the inflection point, the effect size and 95% CI are 5.083 (1.592, 16.229), respectively (Table 4).
Table 4 The results of two-piecewise linear model
|
multiple HPV infections (OR, 95%CI)
|
Fitting model by standard linear regression
|
0.841 (0.534, 1.323)
|
|
Fitting model by two-piecewise linear regression
|
|
|
Inflection point of glucose
|
5.4
|
|
<5.4
|
0.379 (0.196, 0.732)
|
|
>5.4
|
5.083 (1.592, 16.229)
|
|
P for log likelihood ratio test
|
0.001
|
|
CI: Confidence interval we adjusted we adjusted Age,Vagina cleanness,Vaginal leukocytes,Menopausal status,Number of pregnancies,Number of childbirth,Number of miscarriages,Tct,Weight, NSE,SCCA,ALT,AST,UA,CREA,BUN
|