Oral flora in acute stroke patients: A prospective exploratory observational study

OBJECTIVE
To describe the bacterial profile of the oral flora during the first 2 weeks following a stroke, examining changes in the condition of the oral cavity and infections.


BACKGROUND
Dysphagia is common after a stroke and can lead to aspiration pneumonia. Oral flora changes associated with stroke have been implicated as a possible source of bacteria that can cause systemic infections.


MATERIALS AND METHODS
Seventy-seven participants were recruited over a period of 9 months. Fifty participants had a complete set of swabs from four different oral sites and a saliva sample taken at three time points over a 14-day period. Molecular identification of bacteria was performed on the pooled DNA extracted.


RESULTS
A total of 103 bacterial phylotypes were identified, 29 of which were not in the Human Oral Microbiome Database (HOMD). Fourteen of the twenty most common bacterial phylotypes found in the oral cavity were Streptococcal species with Streptococcus salivarius being the most common. The condition of the oral cavity worsened during the study period. Fifteen (30%) patients had at least one infection.


CONCLUSIONS
There appears to be huge diversity of bacterial organisms in the oral cavity of stroke patients, and as most phylotypes identified were only found in one or two participants, no particular patterns linked to infection or the condition of the oral cavity could be discerned.

abrasions, caries and systemic infections. [8][9][10][11] The cognitive and physical impairments that often occur after a stroke, combined with the disruption associated with being hospitalised, can make it difficult for stroke patients to attend to their own oral hygiene effectively. [12][13][14] Cognitive impairment can lead to difficulties processing information and neglect, as well as an inability to recognise the need for oral care. Common stroke sequelae such as hemiplegia may weaken limb movement and motor control. Facial muscle weakness, a reduction in oral sensation and difficulties chewing, swallowing and coughing may also increase the risk of aspiration and precipitate pneumonia. 15,16 Whilst not specific to stroke patients, a systematic review of 36 studies concluded that oral colonisation by respiratory pathogens, fostered by poor oral hygiene and periodontal disease, appears to be associated with the development of hospital-acquired pneumonia. 17 In the first few weeks after a stroke, patients are at increased risk of acquiring a respiratory infection and those with higher levels of dependency have a greater risk of infection. 18,19 About 10% of patients develop pneumonia following a stroke, and this is significantly associated with increased risk of death (odds ratio 3.62 95% CI 2.80-4.68). 20 Compared with healthy people, medically compromised patients such as those who have recently had a stroke tend to harbour a greater number of aerobic and facultative anaerobic gram-negative bacilli in their oral cavity, which appears to increase their risk of systemic infections. 21 In one study, aerobic gram-negative bacilli were found in 34% of acute stroke patients but were not found at all in healthy volunteers. 22 The same study found that presence of aerobic gram-negative bacilli was associated with an unsafe swallow and increased risk of mortality. 22 Sedgley found that potentially pathogenic Enterobacteriaceae such as Escherichia coli are frequently carried in the oral cavity and hospitalisation appears to create conditions that favour oropharyngeal colonisation. 23 Little is known about the condition of the oral cavity, the oral flora and associated risk of infections in stroke patients. Advances in genesequencing technology have improved our ability to explore and understand the oral flora and have revealed a high level of diversity with about 10,000 microbial phylotypes identified in the human oral microflora in 2008. 24 This is considerably more than the 700 oral microbial phylotypes previously identified in 2000. 25 The primary purpose of this exploratory observational study was to apply molecular barcoding technology to describe the bacterial profile of the oral cavity during the first 2 weeks following a stroke. A secondary goal was to explore changes in the condition of the oral cavity and associated clinical factors such as infection.

| MATERIALS AND METHODS
The study was undertaken in an acute stroke unit in a hospital in the north-west of England. Consecutively admitted patients to the stroke unit were recruited. Inclusion criteria were that patients had to be recruited within 48 hours of admission; were 18 years or older; and with a clinical diagnosis of acute stroke. Patients on an end-of-life pathway were excluded. Patients were screened for eligibility by the nursing staff and provided with an information sheet, and those wishing to participate gave written consent. A written consultee declaration was obtained for patients who lacked capacity, and witness consent was obtained for patients who had communication difficulties. All recruited patients were allocated a study number.  29 Sepsis was identified through the presence of a serious systemic infection. 30 Age data were not normally distributed, so they were reported using medians and interquartile ranges. The Barthel Index was reported according to categories: 0-10, dependent; 11-17, moderate dependency; and 18-20, independent. Descriptive statistics were used to depict the distribution of the different bacterial species across time points.
A saliva sample and swabs from four different sites within the oral cavity were taken: the buccal mucosa, tongue, gingiva and hard palate. All samples were taken with plain plastic Sterilin swabs (Fisher Scientific UK Limited, F155CA). Each swab was labelled with the study number, swab site, time point and date. All swabs were transported in a cool box at 4°C to a biomedical research facility. On arrival, all specimens were checked for proof of receipt, logged electronically and then immediately stored at −80°C prior to subsequent analysis.
Laboratory-based molecular identification of bacteria was performed following genomic DNA isolation from all swabs and saliva samples in 20 mg mL −1 lysozyme in buffer (20 mmol L −1 Tris-HCl pH 8.0, 2 mmol L −1 EDTA, 1.2% Triton X-100). A Qiagen DNA easy blood & tissue kit 69504 was used (as per the manufacturer's instructions) to isolate and purify DNA. The purified DNA pellets were resuspended in 50 μL of AE buffer (from kit) and quantified using a Nanodrop spectrophotometer. Genomic DNA was pooled and then subjected to amplification as previously described by Paster et al. 31 using the universal 16S rRNA bacterial gene primers D88F and E94R. The amplification product was cleaned and cloned using the TA TOPO cloning kit (Invitrogen).
Colonies were chosen and screened for having taken up the DNA of interest. Plasmid DNA was isolated from the colonies. Assembly of the data was carried out using Geneious Pro v6.1.5 (www.geneious.com).
The sequences were submitted to the European Nucleotide Archive (ebi.ac.uk), and only those with more than 200 bases and 98%-100% alignment were considered sufficiently similar to confidently identify the phylotype. 32 A phylogenetic tree was constructed using the Molecular Evolutionary Genetics Analysis 33 software and the maximum likelihood method based on the Tamura-Nei model. 34 The phylogenetic tree with the highest log likelihood was obtained by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the maximum composite likelihood approach, and then selecting the topology with superior log likelihood value. The analysis involved 95 nucleotide sequences. Codon positions included were 1st+2nd+3rd+non-coding. All positions with less than 95% site coverage were removed.
Associations between the phylotypes and various patient characteristics were explored using chi-squared test and t test as appropriate. All analyses were performed using Statistical Package for the Social Sciences software (SPSS) version 22.

| RESULTS
A total of 937 patients were admitted between July 2012 and April 2013, and of these, 390 had an initial clinical diagnosis of acute stroke.
Seventy-seven (20%) participants were recruited into the study, and a complete data set was obtained for 50 participants (Figure 1). Readmission -previously recruited = 1

Participants with partial data (n = 27)
Participants were discharged before all swabs were collected = 25 Died = 2

Participants recruited (n = 77)
On average, participants who provided complete data and were included in the full study sample were older and significantly more likely to be dependent and have swallowing difficulties than those who provided partial data (Table 1).
A total of 103 different bacterial phylotypes were identified with 98%-100% sequence similarity cut-off for defining a phylotype. The phylogenetic tree with the highest log likelihood (−51127.6140) is shown in Figure 2.
The Human Oral Microbiome Database (HOMD) was used as the benchmark for expected phylotypes present in the human oral cavity. 35 Twenty-nine bacterial phylotypes that were not mentioned in the HOMD were recovered. The number of participants found with these   Table 2.
The proportion of participants with a gram-negative phylotype found in the oral cavity was 29.4%, 37.3% and 33.3% at time points 1, 2 and 3, respectively, and 60.8% of participants were harbouring at least one gram-negative phylotype at one or other time point.
Aerobic gram-negative bacteria were found in three participants at time point 1, an additional two participants at time point 2 and two more participants at time point 3. In total, seven participants (14%) were harbouring at least one aerobic gram-negative phylotype at one or other time point. Figure 3 shows that 14 of the 20 most common bacterial phylotypes found in the oral cavity were Streptococcal species. Table 3 indicates that Streptococcus salivarius was the most commonly dis-   Table 4 shows the type of infection that was present in the participants with community-and hospital-acquired infections, and the bacterial phylotypes identified in their oral cavities at each of the three time points.

| DISCUSSION
The bacterial profile of the oral cavity during the first 2 weeks following a stroke was successfully described using TA TOPO cloning. There were 103 different bacterial phylotypes identified, including 29 not previously found by HOMD. 35 The phylotypes found in the most common twenty, other than Treponema pedis and Streptococcus suis, are considered normal commensals of the oral cavity. 8,[36][37][38] The seven phylotypes found most frequently in the oral cavity were all Streptococcal species, with S. salivarius in 36% of the full study population discovered most often. S. salivarius can become pathogenic, but this is very rare and some strains are promoted as a probiotic, as they produce bacteriocins that inhibit other bacteria and The tenth most commonly found phylotype in this study, and one not in HOMD, was Streptococcus suis. This bacterium is part of the normal oral flora in pigs and a recognised pig pathogen but is known to cause infections in people working with pigs. 43 Transmission in pigs is mainly through the respiratory route, and since it was first described in Denmark in 1968, 44 over 1600 human cases of S. suis infection have been reported. 43 Most sporadic cases of human infections appear to be due to occupational contact with pigs or pork products, but the number of human infections is increasing and two epidemics were reported in China in 1998 and in 2005. 45 S. suis is the most common cause of human meningitis in Vietnam and has been linked to the development of pneumonia in many parts of the world. 46 Treponema pedis was not in HOMD, but was found eight times in six participants, making it the thirteenth most common phylotype found in this study. T. pedis is an anaerobic gram-negative bacterium associated with digital dermatitis in pigs, but it is very close phylogenetically to T. denticola, part of the known human oral flora associated with periodontitis. 47 To the best of our knowledge, T. pedis has not previously been found in the human oral flora.
Streptococcus pseudopneumoniae was also not in HOMD but has previously been isolated in people with gingivitis 48   Mean THROAT total score

Time point
Upper 95% CI Mean Lower 95% CI different participants. S. pseudopneumoniae is phylogenetically related to S. pneumoniae and S. mitis and appears to be a respiratory tract coloniser, with the potential to become pathogenic. 49 It is associated with chronic obstructive pulmonary disease and respiratory tract infections, particularly aspiration pneumonia. [49][50][51] No associations between any particular phylotype and infection with higher levels of dependency had significantly worse THROAT scores, suggesting that by 2 weeks after admission, oral care did not meet their needs. The worsening THROAT score could also be associated with dehydration that is common after a stroke. 52 There did not appear to be any association between the condition of the oral cavity as indicated by the THROAT score and systemic infection. The mean overall THROAT score across all areas for participants with an infection (7.7) was higher than for those without an infection (6.1), but the difference was not statistically significant.
Pressure on staffing levels and an inability to recruit at weekends limited the researchers' ability to recruit consecutively admitted patients, and full data could only be obtained from participants who stayed on the unit for at least 2 weeks. A consequence of this was that participants included in this study were older, were more dependent and had more problems than the wider cohort of stroke patients originally recruited.
Another limitation was that this study adopted the partial community DNA analysis employing cloning of PCR product of interest, All participants who were diagnosed with an infection were given antibiotics or antifungal medication, so it was impossible to control for this as the potential impact of these two factors on the condition of the oral cavity and the oral flora could not be separated.
Although swabs were taken from different regions of the oral cavity, the contents were combined and the oral flora in each specific area was not identified.
These results concur with a study conducted in Hong Kong which found that periodontal health as measured by plaque and took samples from nine different sites and found the greatest diversity of phylotypes in samples taken from the tonsils, tooth surface and the subgingival area, which were facets of the oral cavity not sampled in this study. We used TA TOPO cloning, which appears to be relatively restrictive in revealing the diversity of bacteria and a 98% sequence similarity as the cut-off for defining our phylotypes. It is possible that Aas 8 used a different technique and a lower threshold as it was not mentioned in their paper.
The aerobic gram-negative bacterial carriage rates identified in this study were lower than those found through more traditional culture techniques described in previous studies. 22,60 In these earlier studies, E. coli was one of the most common aerobic gram-negative bacteria found in the oral cavity after a stroke. 22 Guidelines for the oral health care of stroke survivors recommend the use of mechanical methods of oral care for stroke patients. [62][63][64] Antimicrobial paste appears to reduce the risk of aspiration pneumonia following a stroke, but does not have any impact on mortality. 60 Training health workers on how to use the THROAT effectively may help identify stroke patients with oral health problems who need greater care and attention with their oral hygiene.
There is a need for improved research about how best to maintain oral health in patients who have had a stroke including the benefits of maintaining a good diversity of bacteria in the oral cavity.
It is not known whether the discovery of 29 bacterial phylotypes that were not in the HOMD is due to the different techniques employed to identify phylotypes or the incomplete nature of the HOMD, or whether the oral flora is altered either before or immediately following a stroke and this has not been captured in the HOMD. However, this study provides a baseline that will be useful for future research in this area.
Differences in the phylotypes found between this and other studies could be due to variations in cloning, number of colonies analysed and amplification and sequencing primers used. Should any further work be undertaken in this area, it would be worth amending protocols so that they are better able to identify a greater variety of phylotypes.
This description of the bacteria found in the oral cavity adds to our knowledge and understanding of the oral flora and how it changes in the 2 weeks after a stroke. It provides data to support the development of larger observational or interventional studies that could explore the impact of mechanical or other interventions on the oral flora and associated risk of pneumonia after a stroke.
Avenues for future studies might also include the impact of probiotic oral bacteria on the diversity of phylotypes, risk of pneumonia and oral health in people who have had a stroke. 65,66

| CONCLUSIONS
There appears to be huge diversity of bacterial organisms in the oral cavity of stroke patients, and as most phylotypes identified were only found in one or two participants, no particular patterns linked to infection or the condition of the oral cavity could be discerned.
Risk of infection did not appear to be related to the presence or absence of any particular bacterial phylotype in the oral cavity.