Characteristics of the study population
The average ages of each group (HPV, LSIL, HSIL, Cancer, Normal) were 47.78±9.63, 46.00±10.19, 43.70±10.74, 56.11±9.02 and 43.00±8.69 years, respectively. The proportions of 16/18 HPV subtypes in HPV positive groups were as follow: HPV (13/78), LSIL (10/51), HSIL (6/23), Cancer (4/9) (Table S1). The demographic characteristics of each group are presented in the supplementary data (Table S2).
Sequencing results
After filtering low-quality reads, 6,585,141 assembled clean reads were obtained from 229 samples, with a mean read length of 444.90±5.89bp. For normalization, the reads in each sample were randomly subsampled to the lowest number of 20,098 in sample 318_LCQ (HPV group). After removing singletons (the OTUs contained less than 2 reads), 1,878 OTUs were identified, ranging from 10 OTUs in sample 4391 (Normal group) to 782 OTUs in sample 152_GPZ (HPV group) (Table S3). The average number of OTUs in HPV-negative group (Normal group: 46) was lower than that in HPV-positive groups (HSIL group: 116; LSIL group: 145; HPV group: 157). We found more OTUs in the samples of Cancer group (Mean=256.70 ±174.78), ranging from 51 to 599 (Table S3).
Vaginal microbiota richness and diversity
At the OTU level, microbial richness and diversity were estimated using Chao and Shannon indices, respectively, as shown in Fig. 1. These two indices revealed that HPV infection increased vaginal bacterial richness and diversity. The means of Chao and Shannon indices were much lower in groups HPV-negative (Normal group: 84.02±73.88 and 0.94±0.95, respectively) than in groups HPV-positive (HPV group: 272.26±191.62 and 1.49±1.01, p<0.001; LSIL group: 231.84±195.39, p<0.001 and 1.30±0.93, p=0.04; HSIL group: 185.74±162.62, p<0.001 and 1.51±0.98, p=0.02; Cancer group: 367.76±208.63 and 2.47±0.98 , p<0.001, respectively) . The means of Chao and Shannon indices were the highest in group Cancer and were significantly higher than in groups HSIL (Chao: 367.76±208.63 vs. 185.74±162.62, p=0.01, Shannon: 2.47±0.98 vs. 1.51±0.98, p=0.02), LSIL (Shannon: 2.47±0.98 vs. 1.30±0.93, p=0.0006) and HPV (Shannon: 2.47±0.98 vs. 1.49±1.01, p=0.005). However, there were no differences found among groups HPV, LSIL and HSIL.
Vaginal bacterial structure and beta-diversity in different groups
In PCoA, the first two principal components explained 21.18% and 8.83%, respectively, of the variance along the first and second axes, with the Cancer, HSIL, LSIL and HPV samples visually separated from the Normal sample (Fig. 2). Comparison between two groups based on the ANOSIM test revealed that the bacterial structure of groups Cancer, LSIL and HPV were significantly different from that of group Normal (Table S4). Meanwhile, the bacterial structure of groups HSIL and LSIL were also different from that of group HPV (Table S4).
Taxonomy of the vaginal microbiota in different groups
Overall, 37 bacterial phyla were recovered across all samples (Fig. S2), and All of the samples were dominated by Firmicutes, Actinobacteria, Bacteroidetes, Fusobacteria and Proteobacteria (Fig. S2). Firmicutes was the most abundant phylum, accounting for 73.15%, 68.91%, 69.70%, 70.43% and 47.47% of the Normal, HPV, LSIL, HSIL and Cancer groups, respectively (Fig. 3A). HPV infection tended to lower the proportion of Firmicutes (Cancer < Normal, p<0.05; HPV < Normal, p<0.01). Furthermore, the proportion of Firmicutes was lower in the group Cancer than in the groups HPV, LSIL and HSIL (p<0.05) (Fig. 3A). However, there were no differences among other HPV positive groups. Accordingly, the proportions of Bacteroidetes, Fusobacteria and Proteobacteria were significantly increased in the HPV positive groups as compared to the Normal group (Bacacteroidetes: Cancer 15.72% > HSIL 13.1% > HPV 8.28% > Normal 6.94%, Fig. 3B; Fusobacteria: Cancer 15.84% > LSIL 3.04% > HSIL 2.62% > HPV 2.64% > Normal 2.54%, Fig. 3D; Proteobacteria: Cancer 10.39% > HPV 5.36% > LSIL 3.59% > Normal 1.18%, Fig. 3C). In the same way, the relative percentages of Bacteroidetes, Fusobacteria and Proteobacteria were much higher in the Cancer group than in the groups HPV, LSIL and HSIL (p<0.05) (Fig. 3B, 3C, 3D). There were no differences among the different groups with regard to the proportion of Actinobacteria.
At the genus level, a total of 749 taxa were found across all samples (Fig. S3), with Lactobacillus being the most dominant genus overall. The proportion of Lactobacillus was significantly reduced in the HPV positive groups (Cancer: 19.72%, p<0.01; HSIL: 42.99%, p<0.05; LSIL: 53.71%, p<0.05; HPV: 49.20%, p<0.001, respectively) compared to the Normal group (64.93%) (Fig. 4A). However, there were no significant differences among the groups HPV, LSIL, HSIL and Cancer with regard to the relative abundance of Lactobacillus, except that the group Cancer was lower than the group LSIL (p<0.05). HPV infection also reduced the abundance of Gardnerella (HPV: 7.35% < Normal: 10.66%, p<0.05, Fig. 4B). In addition, HPV infection had also a negative effect on the abundance of Atopobium, which sharply decreased from 3.12% in the Normal group to 2.07% in the HPV group (p<0.001), 2.72% in the LSIL group (p<0.01), 2.49% in the HSIL group (p<0.05) (Fig. 4E). Accordingly, HPV infection increased the relative abundance of Prevotella (Normal: 5.91% < LSIL: 5.99% < HPV: 7.18% < HSIL: 11.46%, P<0.05, Fig. 4C), Bacillus (Normal: 0.34% < LSIL: 3.30% < HPV: 5.77% < HSIL: 5.87% < Cancer: 9.18%, p<0.001, Fig. 4D), Sneathia (Normal: 2.39% < HPV: 2.61% < LSIL: 3.02% < Cancer: 10.03%, p<0.05, Fig. 4F), Megasphaera (Normal: 1.01% < LSIL: 2.13% < Cancer: 2.30% < HSIL: 2.57% < HPV: 9.18%, p<0.01, Fig. 4G), Streptococcus (Normal: 1.03% < Cancer: 1.29% < LSIL: 1.35% < HPV: 1.93%, p<0.01, Fig. 4H) and Anaerococcus (Normal: 1.22% < Cancer: 3.83%, p<0.05, Fig. 4I). The precancerous lesions increased the proportion of Bacillus, with a greater bacterial abundance in LSIL and HSIL groups than in HPV group (p<0.05) (Fig. 4D). A higher proportion of Anaerococcus was also found in the group Cancer compared to the precancerous groups (HPV, LSIL and HSIL, p<0.05) (Fig. 4I). Two taxa, Bacillus and Anaerococcus, were probably related with the progression of CINs severity.
Identification of vaginal microbiological markers in different groups
LEfSe modeling was employed to identify microbiological markers related to HPV infection and CINs severity (Fig. 5). The threshold for the logarithmic LDA model score for discriminative features in this study was 4.0. The most abundant genus in the Normal group was Lactobacillus. In addition, other two taxa were also more abundant in the Normal group, Bacilli at the class level and Atopobium vaginae. HPV infection without CIN or cancerous lesions (HPV group) was strongly associated with Megasphaera. The most abundant bacterium in the LSIL group was Prevotella amnii. However, Prevotella timonensis, Shuttleworthia and Streptococcaceae at the family level were three taxa related to HSIL. Furthermore, more taxa were associated with the Cancer group including Bacillus, Sneathia, Acidovorax, Oceanobacillus profundus, Fusobacterium, Veillonellaceae at the family level, Anaerococcus and Porphyromonas uenonis.
Characteristics of vaginal community state types (CSTs) for different groups
The vaginal bacterial CST analysis visualized by hierarchical clustering revealed that all samples clustered into five major groups: CST I, CST II, CSTII, CST IV and CST V (Fig. 6). The most commonly observed community was CST IV (91/229, 39.74%), followed by CST III (85/229, 37.12%), CST I (44/229, 19.21%), CST II (5/229, 2.18%) and CST V (4/229, 1.75%) as shown in Table 1. The proportions of CSTs in different groups were shown in Table 1. Samples in the Normal group were mostly assigned to CST III (32/68, 47.96%). HPV infection converted the vaginal bacterial community structure from CST III to CST IV, as all of the HPV positive groups were dominated by CST IV (HPV: 32/78, 41.03%; LSIL: 18/51, 35.29%; HSIL: 13/23, 56.52%; Cancer: 8/9, 88.89%, respectively) and presented less CST III (HPV: 28/78, 35.90%; LSIL: 17/51, 33.33%; HSIL: 7/23, 30.43%; Cancer: 1/9, 11.11%, respectively). Furthermore, the proportions of CST IV were gradually augmented with the progression of the severity of CINs (Cancer: 88.89% > HSIL: 56.52% > LSIL: 35.29%).
We identified the most abundant species in each sample in Figure 6, and the distributions of the 29 most predominant species in five groups were documented in Table 2. 78 samples (34.1%) were predominated by Lactobacillus iners, followed by Lactobacillus crispatus (48 samples, 21%), Gardnerella vaginalis (26 samples, 11.4%), Bacillus unclassified (16 samples, 7%), Sneathia amnii (8 samples, 3.5%) and Prevotella amnii (7 samples, 3.1%). HPV infection was related with the decreased abundance of Lactobacillus iners, Lactobacillus crispatus, Gardnerella vaginalis, Lactobacillus gasseri, Anaerococcus spp., Atopobium unclassified and Porphyromonas somerae as compared to the Normal group (Table 2). In addition, HPV infection was associated with the increased abundance of Bacillus unclassified, Escherichia Shigella, Megasphaera unclassified, Streptococcus unclassified, Lactobacillus jensenii, Sneathia sanguinegens, Bifidobacterium unclassified, Candidatus Mycoplasma, Comamonadaceae unclassified, Veillonella montpellierensis, Faecalibacterium unclassified, Finegoldia unclassified, Fusobacterium mortiferum, Porphyromonas uenonis and Ralstonia pickettii (Table 2). Furthermore, the abundance of Gradnerella vaginalis was gradually reduced with the progression of CINs severity (Cancer 0% < HSIL 4.3% < LSIL 11.8% < HPV 12.8% < Normal 13.2% as shown in Table 2).
Table 1: The distribution of community state types (CSTs) in different groups
|
Normal
|
HPV
|
LSIL
|
HSIL
|
Cancer
|
Total
|
|
N (%)
|
N (%)
|
N (%)
|
N (%)
|
N (%)
|
N (%)
|
CST I
|
14 (20.59)
|
14 (17.95)
|
14 (27.45)
|
2 (8.70)
|
0 (0)
|
44 (19.21)
|
CST II
|
2 (2.94)
|
2 (2.56)
|
1 (1.96)
|
0 (0)
|
0 (0)
|
5 (2.18)
|
CST III
|
32 (47.06)
|
28 (35.90)
|
17 (33.33)
|
7 (30.43)
|
1 (11.11)
|
85 (37.12)
|
CST IV
|
20 (29.41)
|
32 (41.03)
|
18 (35.29)
|
13 (56.52)
|
8 (88.89)
|
91 (39.74)
|
CST V
|
0 (0)
|
2 (0)
|
1 (1.96)
|
1 (4.35)
|
0 (0)
|
4 (1.75)
|
Table 2: The distribution of the 29 most predominant species in different groups
|
Normal
|
HPV
|
LSIL
|
HSIL
|
Cancer
|
Total
|
|
N (%)
|
N (%)
|
N (%)
|
N (%)
|
N (%)
|
N (%)
|
Lactobacillus crispatus
|
16 (23.5)
|
15 (19.2)
|
14 (27.5)
|
3 (13.0)
|
0 (0)
|
48 (21.0)
|
Lactobacillus iners
|
30 (44.1)
|
21 (27.0)
|
20 (39.2)
|
6 (26.1)
|
1 (11.1)
|
78 (34.1)
|
Lactobacillus jensenii
|
0 (0)
|
1 (1.3)
|
0 (0)
|
1 (4.3)
|
0 (0)
|
2 (0.9)
|
Lactobacillus gasseri
|
1 (1.5)
|
1 (1.3)
|
0 (0)
|
0 (0)
|
0 (0)
|
2 (0.9)
|
Anaerococcus spp.
|
1 (1.5)
|
0 (0)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (0.4)
|
Atopobium unclassified
|
1 (1.5)
|
0 (0)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (0.4)
|
Bacillus unclassified
|
0 (0)
|
7 (9.0)
|
5 (9.8)
|
3 (13.0)
|
1 (11.1)
|
16 (7.0)
|
Bifidobacterium breve
|
1 (1.5)
|
0 (0)
|
1 (2.0)
|
0 (0)
|
0 (0)
|
2 (0.9)
|
Bifidobacterium unclassified
|
0 (0)
|
1 (1.3)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (0.4)
|
Candidatus Mycoplasma
|
0 (0)
|
1 (1.3)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (0.4)
|
Comamonadaceae unclassified
|
0 (0)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (11.1)
|
1 (0.4)
|
Veillonella montpellierensis
|
0 (0)
|
1 (1.3)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (0.4)
|
Escherichia Shigella
|
0 (0)
|
3 (3.8)
|
1 (2.0)
|
0 (0)
|
0 (0)
|
4 (1.7)
|
Faecalibacterium unclassified
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (4.3)
|
0 (0)
|
1 (0.4)
|
Finegoldia unclassified
|
0 (0)
|
1 (1.3)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (0.4)
|
Fusobacterium mortiferum
|
0 (0)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (11.1)
|
1 (0.4)
|
Gardnerella vaginalis
|
9 (13.2)
|
10 (12.8)
|
6 (11.8)
|
1 (4.3)
|
0 (0)
|
26 (11.4)
|
Megasphaera unclassified
|
0 (0)
|
1 (1.3)
|
0 (0)
|
1 (4.3)
|
0 (0)
|
2 (0.9)
|
Porphyromonas somerae
|
1 (1.5)
|
0 (0)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (0.4)
|
Porphyromonas uenonis
|
0 (0)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (11.1)
|
1 (0.4)
|
Prevotella amnii
|
2 (3.0)
|
3 (3.8)
|
0 (0)
|
1 (4.3)
|
1 (11.1)
|
7 (3.1)
|
Prevotella bivia
|
1 (1.5)
|
0 (0)
|
0 (0)
|
1 (4.3)
|
1 (11.1)
|
3 (1.3)
|
Prevotella timonensis
|
1 (1.5)
|
2 (2.6)
|
0 (0)
|
1 (4.3)
|
0 (0)
|
4 (1.7)
|
Prevotella unclassified
|
1 (1.5)
|
1 (1.3)
|
0 (0)
|
1 (4.3)
|
0 (0)
|
3 (1.3)
|
Ralstonia pickettii
|
0 (0)
|
1 (1.3)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (0.4)
|
Shuttleworthia uncultured
|
1 (1.5)
|
2 (2.6)
|
1 (2.0)
|
2 (8.7)
|
0 (0)
|
6 (2.6)
|
Sneathia amnii
|
2 (2.9)
|
3 (3.8)
|
2 (3.9)
|
0 (0)
|
1 (11.1)
|
8 (3.5)
|
Sneathia sanguinegens
|
0 (0)
|
0 (0)
|
0 (0)
|
0 (0)
|
1 (11.1)
|
1 (0.4)
|
Streptococcus unclassified
|
0 (0)
|
3 (3.8)
|
1 (2.0)
|
1 (4.3)
|
0 (0)
|
5 (2.1)
|