Temporal instability of the post-surgical maxillary sinus microbiota

Recent interest in the paranasal sinus microbiota, sparked by increased availability of culture-independent methods, has prompted reconsideration of the role of bacteria in chronic rhinosinusitis (CRS). Molecular techniques identify significantly more organisms than standard cultures [1–5], shedding light on complex bacterial communities within healthy sinuses that had previously been assumed to be sterile. The normal microbiota of the sinuses is currently not well defined and appears to vary widely among individuals [2, 6–8]. Evidence suggests that diminished bacterial diversity and richness are associated with CRS [2, 5, 6]. It has been proposed that specific organisms may be critical for sinus health and that some bacterial species may be associated with development of disease [2, 9]. At this time, however, it remains to be elucidated whether a stable sinus microbiome exists, and whether disruption of normal microbial homeostasis underlies alterations in host mucosal integrity, mucociliary clearance, or expression of local inflammatory factors. Also unclear is how an individual’s sinonasal microbiome is established and how dynamic it is over time.

The upper airway serves as a filter, constantly entrapping airborne microbes and particulates in the mucus blanket and removing them harmlessly by mucociliary clearance [10, 11]. In settings of mucostasis, overgrowth can occur of bacteria that are either resident or transiently moving through the sinonasal tract. Acute bacterial sinusitis is a common occurrence that is generally self-limited or responsive to medical treatment, suggesting that acute changes in the sinus microbiome associated with mucostasis are tolerated and reversible. While it is possible that pathological alteration of the microbiota precedes inflammation and subsequent mucostasis, evidence for this is lacking in acute sinusitis. In CRS, a more heterogeneous disorder with a multifactorial etiology [12, 13], purulent static secretions are frequently present, and a role for bacteria in promoting chronic inflammation is generally presumed. Antibiotics are the most common medication prescribed for managing symptomatic CRS, but evidence of efficacy is limited, particularly in CRS with nasal polyposis [14].

In cases of medically recalcitrant CRS, endoscopic sinus surgery (ESS) can be very effective in clearing mucopus and enlarging drainage pathways, in order to facilitate mucociliary function and access for topical therapies [15, 16]. The restoration of sinus health after ESS again indicates that changes in the microbiota associated with chronic inflammation are also caused, or at least perpetuated, primarily by mechanical obstruction. In a subset of patients, however, ESS does not reverse the chronic inflammatory process, requiring long-term medical management and sometimes repeated surgery. The role of bacteria in the pathogenesis of chronic sinus mucosal inflammation in this setting remains under investigation [17]. In patients who have undergone adequate ESS, ongoing sinusitis is not due to blockage of sinus ostial outflow. Often, periods of disease control are punctuated by acute inflammatory and/or infectious exacerbations, which can be readily visualized endoscopically and sampled in the office setting. In this way, post-operative sinus cavities present a window through which to investigate the relationship between the microbiome and the mucosal inflammatory state over time.

Previous studies suggest that the sinus microbiome varies among individuals, among sub-sites within an individual sinonasal tract, and in response to medical therapy [18–20] or sinus surgery [21–23]. While the microbiota remain remarkably constant over time in the gut and urogenital tract [24–26], temporal variation has not been evaluated in the post-operative sinuses. In order to evaluate the hypothesis that dysbiosis is a contributing factor in surgically-recalcitrant CRS, it is important to establish what constitutes healthy microbiota and the degree to which a stable set of bacteria, or “core” sinus microbiome, exists for individuals. Only with this understanding can the microbiota characteristics associated with chronic inflammatory sinus disease be interpreted. Therefore, the aim of this study is to determine the long-term stability of the microbiota within surgically-opened maxillary sinuses between two temporally separate points. We hypothesize that stable microbiota features exist within individuals, despite changes in disease state and medical therapy.

This prospective study was approved by the Johns Hopkins Medicine Institutional Review Board. All subjects provided written consent to participate. Between 3/2011 and 4/2015, 35 patients with longstanding patent maxillary antrostomies, status post ESS in the remote past, had swab samples of the maxillary sinus taken at two distant time points as part of a routine follow-up. All collections were taken from the maxillary sinus mucosa using a cotton-tipped wire swab (Polyester Dacron Swab Pack, MPC, Amarillo, CA; Dry Swabs, Copan Diagnostics, Murrieta, CA) under endoscopic visualization by the same clinician (APL). Swabs were discarded and retaken if contaminated by mucosa outside the target region. Immediately following collection, the samples were flash frozen and stored at −80C until processed.

The vast majority of patients with CRS experience clinical resolution after medical and surgical therapy, suggesting that sinus dysbiosis, if present, is either fully reversible or is secondary to impaired ventilation and mucociliary function. In cases of persistent CRS after surgery, the ongoing role of bacteria is unknown. The nose and sinuses are not sterile, and therefore bacteria are necessarily found in both healthy and CRS maxillary sinuses after antrostomy. In this study, we compared the maxillary sinus microbiota in a cohort of patients from whom samples were obtained years apart. Our results confirm that the post-operative maxillary sinus (POMS) microbiota is broadly characterized by a group of highly prevalent taxa which differ widely in proportional abundance. Additionally, there is a diverse sinus microbiota that contains an assortment of proportionally low abundance bacteria that vary between individuals and within individuals over time. The POMS microbiota of an individual person is relatively unstable, with an overlap of approximately 25% of taxa between samples. A slightly higher degree of stability is present in the setting of chronic inflammation where there is a depleted, low diversity microbial community. Additionally, beta-diversity metrics that account for the relative abundance of taxa (Fig. 5), show that sample pairs cluster together more strongly by group than they did within-patient. These findings highlight the complexity of identifying a “normal” or “healthy” microbiome upon which to propose a dysbiotic mechanism responsible for driving CRS. As an individual’s POMS microbiota may be in constant temporal flux, perhaps due to differences in inhaled and skin bacteria transiently entering and being cleared from the sinus, the cause and effect relationship of specific bacteria present in CRS may be difficult to demonstrate. Our findings are in accordance with the results of a recent study, which reported on the middle meatus microbiome of 23 patients undergoing ESS [22]. The authors found a large number of taxa that constitute an extremely small fraction of the total detected microbiota [22].

Previous studies have revealed unique “cores” of bacterial organisms that remains present in samples across various time periods in individual body sites. The gut, which has been studied the most extensively, has a stable core group of bacteria in various proportions. Martinez et al. found 33–40 species-level OTUs that remained consistent over 1 year in three subjects [43]. While this was only 12% of the total OTUs detected, those species accounted for >75% of the total sequences obtained for each subject. Faith et al. studied individual subjects’ fecal microbiota over a 5-year period, showing it to be quite stable with 60% of bacterial strains remaining present [25]. Additionally, the most stable components of the fecal microbiota were also the most abundant. Caporaso et al. [24] extended their study to include the mouth and palms as well as the gut. They too identified a small “core human microbiota” at these body sites present over months, representing about 10% of all OTUs identified at each site [24]. Interestingly, the size of the stable core microbiome was largest for mouth followed by the gut and then the palms respectively. The microbiome of the nasal passages of healthy individuals has been reported to be relatively stable, although a stability index was not calculated [44]. Futhermore, Yan et al. [20] examined the microbiota of the anterior nasal cavity, middle meatus, and sphenoethmoid recess in 12 healthy adults, noting a temporally stable, highly individualized baseline, with some differences in microbiota between the anterior nares and deeper anatomic subsites. Another study that examined middle meatus swabs from 28 healthy subjects found that many individuals harbor respiratory pathogens at low abundance, suggesting that pathogenic bacteria may be transient or permanent members of the healthy microbiome at relatively low amounts [8]. This is consistent with our findings in the POMS.

In the gut, the composition of an individual’s stable microbiota is proposed to be derived from diet or environmental exposures from early in life. In contrast, our findings suggest that the POMS microbiota is much more dynamic. A 25% stability at the genus level may indicate a significant and ongoing influence of transient environmental exposures in the upper airway, which, like the palms, are likely more varied than the gut. At the same time, a number of highly prevalent taxa in the POMS are largely stable in the microbiota but shift in abundance over time. Bacteria belonging to the Staphylococcus, Corynebacterium, Proprionibacterium and Pseudomonas genera are the primary examples. These organisms are often among the most abundant taxa found at any given time point, and, when this is the case, they are present at the other time point in over 80% of subjects.

A limitation of this study is that only two points in time are compared. Obviously, collection of more frequent time points over a greater period would most accurately portray maxillary sinus microbiota temporal variation. We performed a preliminary analysis of additional time points in a smaller number of subjects, which suggested increased microbiota stability over short time intervals, but still with marked fluctuation. As more longitudinal samples are collected and analyzed for individual subjects, characterization of the stably present organisms can be refined. It is likely that medications, dietary influences, and changes in living or work environments lead to shifts in sinus microbiota between sampling points. This is expected and important to recognize in a “real-world” descriptive study such as this. The percentage range of the stable taxa is fairly consistent, despite individual differences in multiple external factors. Indeed, even if we excluded patients who were on antibiotics during sampling or have been on antibiotics in the 2 months preceding sampling, the stability index remains unchanged. The precision to which rare taxa can be identified in the specimens may limit interpretation to some extent. However, based on the depth of coverage, the fact that samples are compared within individuals, and that negative controls without a template were processed for each primer pair during PCR, the possibility of contamination has been minimized as much as possible. Accordingly, we believe that rare taxa were accurately detected and their inclusion is justified in the stability analysis. We observe that very rare taxa present at one time point may be present in high relative abundance at a subsequent time point. This finding highlights the dynamic nature of the sinus microbiota and argues that rare taxa should not be automatically dismissed as irrelevant or contaminants in microbiome studies. Lastly, it should be noted that increased PCR cycle numbers are associated with point mutation artifacts and reduced richness, but do not appear to affect community structure [45].

To the extent that “core” post-antrostomy maxillary sinus microbiota exists, it is likely that these bacteria have the capacity to participate in a “healthy” diverse microbiome as well as to pathologically dominate in a disease setting. The reason for the decreased diversity and evenness associated with CRS is not understood, and a cause-and-effect relationship has not been demonstrated. It is conceivable that changes in the local milieu due to chronic inflammation support the growth of certain resident bacterial species over others, thus driving the community to a less diverse state. Alternatively, development of uneven and less diverse microbiota within the sinus may precede and stimulate a subsequent inflammatory response. Our data suggest that clinically healthy POMS are associated with diverse but relatively temporally unstable microbiota. In cases of longstanding infectious CRS without polyps, a highly depleted microbiota may develop, dominated by few stable taxa. This subset of patients can be frustratingly difficult to rid of inflammation and infection, and this disease phenotype seems most consistent with an etiology linked to dysbiosis. Although long-term control of inflammation is also challenging in CRS with nasal polyps, the association with infection or dysbiosis is not apparent, and the present study shows a similar level of temporal stability to healthy sinuses, with a reduced overall diversity.

Culture independent techniques permit a more comprehensive view of the sinus microbiome than previously available, but interpretation of this information has presented new challenges. Although dysbiosis continues to be hypothesized as contributing to the pathogenesis of CRS, studies to date have not revealed a consistent pattern differentiating CRS patients and healthy controls. In focusing on the post-surgical maxillary sinus, our group has described shifts in the microbiota occurring in response to medical therapy for acute infectious CRS exacerbation, as well as the broad impact of topical therapies [18, 19]. Taken together with the present study of temporal variation, it is clear that the sinus microbiome is complex, variable, and likely influenced constantly by internal and external factors. The identification of temporally stable taxa in uninflamed sinuses may lend further support for their role in maintaining health, but against the larger backdrop, much further investigation is necessary to delineate the relationship between these bacteria and CRS, before consideration can be given to concepts such as probiotic therapy or selection of antibiotics based on culture-independent methods.