Role of bacteria in immunopathology

Chronic rhinosinusitis is considered to represent a spectrum of disorders with varying combinations of immunopathological mechanisms. Although the exact role of infection in the pathogenesis of chronic rhinosinusitis is unknown, bacteria probably contribute to the persistence and severity of chronic rhinosinusitis.^{Reference Stewart and Costerton2}

Historically, chronic rhinosinusitis has been primarily categorised into two groups: chronic rhinosinusitis with nasal polyps and chronic rhinosinusitis without nasal polyps. Both types are characterised by epithelial disruption, ciliary dysfunction, mucus gland hyperplasia, bacterial overgrowth and biofilm formation.^{Reference Kennedy and Borish3}

Bacteria may become pathogenic by acting as antigens and activators of pathogen-associated molecular receptors.^{Reference Kennedy and Borish3} They can also secrete toxins which may have superantigen activity and immune adjuvants that stimulate non-specific adaptive immune responses.^{Reference Kennedy and Borish3} Staphylococcus aureus is especially common in eosinophilic chronic rhinosinusitis, in which it is associated with a superantigen-mediated increase in T-cell type 2 cytokine secretions and immunoglobulin E (IgE) sensitisation.^{Reference Zhang, Van Zele, Perez-Novo, Van Bruaene, Holtappels and DeRuyck4,Reference Foreman, Holtappels, Psaltis, Jervis-Bardy, Field and Wormald5} Observations of improvement in individuals with chronic rhinosinusitis treated with doxycycline and macrolide antibiotics may be linked to the role of these antibiotics in reducing bacterial load and superantigen production, as well as their more direct anti-inflammatory effects.^{Reference Wallwork, Coman, Mackay-Sim, Greiff and Cervin1,Reference Kennedy and Borish3,Reference Van Zele, Gevaert, Holtappels, Beule, Wormald and Mayr6}

Despite the widespread use of antibiotics in the treatment of chronic rhinosinusitis, there is some evidence to indicate that antibiotics may be paradoxically implicated in the causation of this condition. Some evidence suggests that an imbalance of immunogenic bacteria against tolerogenic bacteria may promote the persistence of chronic rhinosinusitis.^{Reference Hoggard, Biswas, Zoing, Mackenzie B, Taylor and Douglas7–Reference Lee, Frank and Ramakrishnan9} In addition, with repeated use of broad-spectrum antibiotics, there is concern that more resistant bacterial strains will emerge. These bacteria reside in biofilms, are extremely difficult to eradicate, and may form a nidus for future exacerbations.

Biofilms can alter antibiotic effectiveness, as the bacteria in their centre are less metabolically active and so less responsive to the cellular mechanisms that are central to the action of many antibiotics.^{Reference Kennedy and Borish3} The gradients in hypoxia and pH created within the biofilms also aid survival of the core bacteria throughout treatment. Further defence mechanisms which promote bacterial persistence in biofilms include quorum sensing, and the ability of an organised multicellular complex that forms the biofilm to evade phagocytosis.^{Reference Kennedy and Borish3}

A dose-dependent effect of topical tobramycin in penetrating and destroying biofilms in a rabbit model has been reported.^{Reference Antunes, Feldman, Cohen and Chiu10} However, effective treatment requires much higher doses of antibiotics than usual, with up to a thousand-fold higher concentration than the predicted mean inhibitory concentration being necessary.^{Reference Antunes, Feldman, Cohen and Chiu10} Although the goal of topical antibiotic therapy is to achieve higher local delivery of drugs while reducing systemic effects, the evidence for the effectiveness of topical antibiotics has not been proven.^{Reference Lim, Citardi and Leong11} It is likely that topical antibiotics in safe prescribed doses are usually ineffective against biofilms.^{Reference Ha, Psaltis, Butcher, Wormald and Tan12}

Efficacy of antibiotics

Antibiotic intervention has not been proven to be generally effective in chronic rhinosinusitis, other than during acute exacerbations.^{Reference Stewart and Costerton2,Reference Kennedy and Borish3} Although bacteria are ubiquitous in the sinuses of patients with chronic rhinosinusitis, definitive evidence for a primary infectious aetiology of this condition remains elusive. A few prospective studies have investigated the efficacy of short-term antibiotic use in chronic rhinosinusitis. The lack of placebo arms in these studies limits the interpretation of any clinical response, and overall they have not detected significant differences between the treatment arms.^{Reference Legent, Bordure, Beauvillain and Berche13–Reference Sydnor, Scheld, Gwaltney, Nielsen, Huck and Therasse15}

The international consensus statement recommends the use of oral macrolides as an option in the treatment of chronic rhinosinusitis on the basis that they have shown benefit in some studies in reducing endoscopy scores and improving symptoms in chronic rhinosinusitis patients.^{Reference Orlandi, Kingdom, Hwang, Smith, Alt and Baroody16} Most studies of macrolide therapy in chronic rhinosinusitis patients without nasal polyps have shown a benefit over placebo, while limited data are available to determine the efficacy for chronic rhinosinusitis with nasal polyps. There is a paucity of evidence supporting the efficacy of non-macrolides in chronic rhinosinusitis.^{Reference Orlandi, Kingdom, Hwang, Smith, Alt and Baroody16}

A meta-analysis of the efficacy of long-term macrolide therapy in chronic rhinosinusitis did not demonstrate evidence for a clinically significant impact across three studies included in the review.^{Reference Wallwork, Coman, Mackay-Sim, Greiff and Cervin1} A single study reported a significant difference in patient-reported outcomes after 12 weeks of treatment, favouring roxithromycin, but only in those patients with a normal serum total IgE level. However, these results are difficult to interpret, as the patient-reported outcome scale used was not validated and could be interpreted as biased in having more points to describe improvement than worsening.^{Reference Suzuki, Shimomura, Ikeda, Furukawa, Oshima and Takasaka17} Macrolides function by downregulating proinflammatory cytokines including interleukin 8, a potent neutrophil chemoattractant.^{Reference Pynnonen, Venkatraman and Davis18} Thus, it is possible that patients in the high serum IgE subgroup (often with associated eosinophilic infiltration) may be less likely to benefit from macrolide therapy compared to those in the normal IgE subgroup.^{Reference Pynnonen, Venkatraman and Davis18}

Analysis of drug molecules in nasal and paranasal tissues and secretions

A number of studies have examined antibiotic concentrations in the sinonasal tissues and/or mucus.^{Reference Ambrose, Anon, Bhavnani, Okusanya, Jones and Paglia19–Reference Tolsdorff38} However, the majority of these studies have been performed in patients with acute sinusitis, acute exacerbations of chronic rhinosinusitis or upper respiratory tract infections. Furthermore, there are limitations in the interpretation and comparison of these studies, including small sample sizes, and heterogeneity of study populations, specific antibiotic and dosage regimens, nasal sampling methods, time of sampling in relation to the time of dose, and drug analysis methods.

The lack of clinical difference seen in chronic rhinosinusitis patients treated with oral antibiotics may be attributed in part to a sub-therapeutic level of drug penetration into sinonasal biofilms, in addition to the presence of growing antibiotic-resistant strains.^{Reference Stewart and Costerton2,Reference Kennedy and Borish3,Reference Antunes, Feldman, Cohen and Chiu10} However, the extent to which drug distribution to the human sinonasal mucosa can contribute to the efficacy of oral antibiotics in patients with chronic rhinosinusitis remains largely undefined.

Predicting the efficacy of an antibiotic is complicated by its varying distribution across the body.^{Reference Langdon, Crook and Dantas39} The total plasma concentration of a drug is not a reliable predictor of its clinical efficacy, as it may not always reflect the drug concentration and activity in the target site. Accordingly, the total plasma concentration of a drug is not an ideal pharmacokinetic parameter on which to base rational dosing. The effects of antibiotics also differ by body site. For example, the pharynx and saliva recover their initial microbial diversity after antibiotic therapy much more quickly than does the gut.^{Reference Jakobsson, Jernberg, Andersson, Sjolund-Karlsson, Jansson and Engstrand40,Reference Zaura, Brandt, Teixeira de Mattos, Buijs, Caspers and Rashid41} Macrolides are highly lipophilic and consequently penetrate well into tissue, especially bronchial secretions, prostatic tissue, middle-ear exudates and bone tissues.^{Reference Periti, Mazzei, Mini and Novelli42} Similarly, second-generation tetracyclines like doxycycline are also lipophilic and so penetrate well into most tissues, including respiratory tract tissue, with the highest concentrations in the liver, kidney and digestive tract. Biliary doxycycline concentrations are found to exceed those of serum by many folds, possibly reflecting higher active transport and secretion of the drug in the biliary tract.^{Reference Agwuh and MacGowan43} Both the macrolide and tetracycline drug groups are found to penetrate poorly into cerebrospinal fluid and saliva.^{Reference Periti, Mazzei, Mini and Novelli42,Reference Agwuh and MacGowan43}

The analysis of drugs or their metabolites in tissue target compartments such as cerebrospinal fluid and mucus can improve our understanding of drug penetration and likely efficacy at the site of infection.^{Reference Mouton, Theuretzbacher, Craig, Tulkens, Derendorf and Cars44} The collection of sinonasal tissues or secretions has been shown to be an appropriate approach in studies measuring drug concentrations in nasal secretions, nasal mucosa, paranasal sinus mucosa, ethmoid bone and septal cartilage after intranasal and oral dosing. These studies have provided useful information in the fields of toxicology and pathophysiology.^{Reference Legent, Bordure, Beauvillain and Berche13,Reference Ambrose, Anon, Owen, VanWart, McPhee and Bhavnani20,Reference Cherrier, Tod, Gros, Petitjean, Brion and Chatelin22–Reference Dinis, Monteiro, Martins, Silva and Gomes25,Reference Eneroth, Lundberg and Wretlind27,Reference Gehanno, Darantière, Dubreuil, Chobaut, Bobin and Pages30–Reference Kuehnel, Schurr, Lotter and Kees32,Reference Lundberg, Gullers and Malmborg34,Reference Pea, Marioni, Pavan, Staffieri, Bottin and Staffieri36–Reference Tolsdorff38,Reference Mouton, Theuretzbacher, Craig, Tulkens, Derendorf and Cars44,45} However, nasal tissue samples and/or mucus are much more difficult to obtain than blood or urine, and are much smaller in volume. Accordingly, special attention is required in the various stages of bioanalysis as compared to that required for conventional samples such as the blood and urine, and validation procedures need to be considered.^{Reference Serralheiro, Alves and Falcao46}

Nasal secretions can be sampled by nose blowing, aspiration, absorption or washing techniques, and each of these has its own advantages, limitations and influences on the results obtained.^{Reference Serralheiro, Alves and Falcao46} Notably, different collection techniques provide heterogeneous matrices and analyte concentrations, which limit comparison between studies.^{Reference Serralheiro, Alves and Falcao46} A sample preparation stage is mandatory for drug elution, as biological samples are not directly compatible with quantification techniques such as high-performance liquid chromatography analysis.^{Reference Serralheiro, Alves and Falcao46} Nasal secretions are protein-rich samples that require an efficient sample clean-up procedure.^{Reference Pea, Marioni, Pavan, Staffieri, Bottin and Staffieri36,Reference Bimazubute, Rozet, Dizier, Gustin, Hubert and Crommen47} Liquid–liquid extraction and solid-phase extraction procedures have been found to improve sensitivity by removing matrix interferences.^{Reference Pea, Marioni, Pavan, Staffieri, Bottin and Staffieri36,Reference Bimazubute, Rozet, Dizier, Gustin, Hubert and Crommen47}

It is essential that all bioanalytical methods for determining analytes in a specific biological matrix undergo validation to assure that they are reliable and reproducible.⁴⁵ As nasal and paranasal specimens tend to be smaller in volume as compared with blood or urine samples, they impose a need for the partial validation of a method based on a previously validated method applied to a different matrix.^{45,Reference Serralheiro, Alves and Falcao46} While selectivity and sensitivity are key parameters, the linearity, precision and accuracy of the measured concentrations, and the stability of the analytes in the biological matrix, should also be assessed.⁴⁵

The determination of drug levels in sinus secretions in the past was largely performed by microbiological assays such as agar and disc diffusion.^{Reference Axelsson and Brorso21,Reference Dewever23,Reference Eneroth, Lundberg and Wretlind27,Reference Fraschini, Scaglione, Pintucci, Maccarinelli, Dugnani and Demartini29,Reference Gnarpe and Lundberg31,Reference Liss and Norman33,Reference Lundberg, Gullers and Malmborg34} However, in light of technological advances and the development of new analytical methodologies and instrumentation, more efficient quantitative analysis can be performed by chromatography.^{Reference Serralheiro, Alves and Falcao46} Novel sample preparation methods have improved the extraction of drug analytes from biological matrices so that less than 10 µl of nasal secretions may be adequate for analysis.

High-performance liquid chromatography is frequently used as a bioanalytical method, as it provides several advantages over others such as gas chromatography.^{Reference Serralheiro, Alves and Falcao46} Nevertheless, few studies have used high-performance liquid chromatography to determine the concentrations of various antibiotics commonly used for chronic rhinosinusitis, particularly those from the macrolide and tetracycline groups, compared with fluoroquinolone drugs.^{Reference Ehnhage, Rautiainen, Fang and Sanchez26,Reference Fang, Palmer, Chiu, Blumer, Crownover and Campbell28,Reference Kuehnel, Schurr, Lotter and Kees32,Reference Margaritis, Ismailos, Naxakis, Mastronikolis and Goumas35}

Older studies have described the pharmacokinetic properties in the sinus secretion of doxycycline and roxithromycin, using microbiological assay techniques, with many reporting the concentration of sinus secretion drug levels at varying time points to be above reference mean inhibitory concentration levels for susceptible strains of pathogens.^{Reference Axelsson and Brorso21,Reference Dewever23,Reference Eneroth, Lundberg and Wretlind27,Reference Fraschini, Scaglione, Pintucci, Maccarinelli, Dugnani and Demartini29,Reference Gnarpe and Lundberg31,Reference Liss and Norman33,Reference Lundberg, Gullers and Malmborg34} Although sinonasal drug levels are suggested to be therapeutic by being above the mean inhibitory concentration level for susceptible strains,^{Reference Axelsson and Brorso21,Reference Dewever23,Reference Eneroth, Lundberg and Wretlind27,Reference Fraschini, Scaglione, Pintucci, Maccarinelli, Dugnani and Demartini29,Reference Gnarpe and Lundberg31,Reference Liss and Norman33,Reference Lundberg, Gullers and Malmborg34} the definitions of therapeutic concentrations are variable between studies.

Microbiome studies have largely not been performed in parallel, thereby hindering investigation of the relationship between drug levels and activity on an intrinsic level, and there is often a lack of clinical correlation. Microbial analysis, when performed, appears to only have occurred in the context of pathogens in acute sinusitis or exacerbations of chronic rhinosinusitis.^{Reference Ambrose, Anon, Bhavnani, Okusanya, Jones and Paglia19,Reference Ambrose, Anon, Owen, VanWart, McPhee and Bhavnani20,Reference Lundberg, Gullers and Malmborg34,Reference Margaritis, Ismailos, Naxakis, Mastronikolis and Goumas35}

Impact of antibiotics on sinonasal microbiome

There is increasing evidence to support the concept that a more diverse microbiome is associated with improved health outcomes and less disease burden.^{Reference Levine and D'Antonio48–Reference Turnbaugh, Ley and Hamady50} Some studies have shown that treatments such as intranasal corticosteroids in the management of acute exacerbations of chronic rhinosinusitis may decrease nasal microbiome diversity.^{Reference Liu, Soldanova, Nordstrom, Dwan, Moss and Contente-Cuomo51,Reference Liu, Kohanski and Mendiola52} A decrease in sinonasal bacterial diversity following a period of oral antibiotics and corticosteroids in patients with acute exacerbations of chronic rhinosinusitis has also been reported.^{Reference Liu, Soldanova, Nordstrom, Dwan, Moss and Contente-Cuomo51} However, as no baseline steady-state sample was collected, the decrease in diversity may also be attributed to resolution of the acute exacerbation.^{Reference Liu, Soldanova, Nordstrom, Dwan, Moss and Contente-Cuomo51}

In a previous study, we observed that the immediate effects of oral doxycycline or prednisone on bacterial communities and cytokines were unpredictable and highly variable between individuals.^{Reference Jain, Hoggard, Zoing, Jiang, Biswas and Taylor53} This is supported by further studies of changes in bacterial communities following systemic medical therapies, which also have not shown significant differences in bacterial diversity or richness, and which have described unpredictable and complex community shifts.^{Reference Cleland, Bassiouni, Vreugde and Wormald54–Reference Hauser, Ir, Kingdom, Robertson, Frank and Ramakrishnan56} Small changes in the relative abundance in a number of dominant taxa have been demonstrated, including trends of increase in staphylococcal species.^{Reference Cleland, Bassiouni, Vreugde and Wormald54–Reference Hauser, Ir, Kingdom, Robertson, Frank and Ramakrishnan56} The effects of medical therapies on the sinonasal microbiome in chronic rhinosinusitis patients have yet to be correlated with clinical responses.

Impact of antibiotics on gut microbiome

Collateral harm from antibiotics is not limited to rising antibiotic resistance. Antibiotic use can account for 19 per cent of emergency department visits for adverse drug reactions; furthermore, the risks of adverse reactions with some antibiotics are reportedly comparable to those of insulin, warfarin and digoxin.^{Reference Shehab, Patel, Srinivasan and Budnitz57} In a prospective study of antibiotic-associated suspected adverse drug reactions among 762 hospitalised patients, 269 patients suffered an adverse drug reaction.^{Reference Kiguba, Karamagi and Bird58} The system organ classes most frequently affected with serious adverse effects were the gastrointestinal (50 per cent, 135 out of 269), neurological (24 per cent, 64 out of 269), body-general (10 per cent, 27 out of 269), and skin or appendages (6 per cent, 17 out of 269), among others.^{Reference Kiguba, Karamagi and Bird58}

It is well known that antibiotics can cause lasting changes to the gut microbiome, including a loss in diversity, the emergence of new genome sequences, growing antibiotic-resistant strains and the upregulation of antibiotic resistance genes.^{Reference Jakobsson, Jernberg, Andersson, Sjolund-Karlsson, Jansson and Engstrand40,Reference Korpela, Salonen, Virta, Kekkonen, Forslund and Bork59,Reference Raymond, Deraspe, Boissinot, Bergeron and Corbeil60} Dysbiosis of the microbiome has been causally implicated in a large number of metabolic, immunological and developmental disorders, and may affect susceptibility to the development of infectious diseases, which can impact a wide variety of systems.^{Reference Angelakis, Million, Kankoe, Lagier, Armougom and Giorgi61–Reference Yoon and Yoon65} Just as dysbiosis is implicated in the development of chronic rhinosinusitis, it is also implicated in inflammatory bowel disease – representing a spectrum of chronic inflammatory intestinal disorders – via a dysregulated immune response to host intestinal microflora.^{Reference Angelakis, Million, Kankoe, Lagier, Armougom and Giorgi61,Reference Yoon and Yoon65,Reference Gevers, Kugathasan, Denson, Vázquez-Baeza, Van Treuren and Ren66}

Complete recovery after short-term antibiotic treatment may still not be achieved for as long as four years following treatment.^{Reference Jakobsson, Jernberg, Andersson, Sjolund-Karlsson, Jansson and Engstrand40} Many of these perturbations in gut microflora have been evaluated in relation to susceptibility to enteritis with Clostridium difficile and Salmonella typhimurium.^{Reference Langdon, Crook and Dantas39,Reference Slimings and Riley67} However, few studies have described the changes in gut microflora associated with acute, short-term adverse gastrointestinal symptoms not necessarily linked to common pathogens. The patterns of disruption to the gut microbiome associated with chronic gastrointestinal diseases such as inflammatory bowel disease and irritable bowel syndrome have been determined. However, the extent to which these can be attributed to previous antibiotic exposure remains to be elucidated.^{Reference Ianiro, Tilg and Gasbarrini63,Reference Gevers, Kugathasan, Denson, Vázquez-Baeza, Van Treuren and Ren66}

Disruption to gut microflora has been frequently studied in relation to the use of broad-spectrum medications, such as clindamycin^{Reference Slimings and Riley67} and beta-lactam antibiotics for infections affecting various other systems,^{Reference Korpela, Salonen, Virta, Kekkonen, Forslund and Bork59,Reference Boursi, Mamtani, Haynes and Yang68,Reference Villarreal, Aberger, Benrud and Gundrum69} or metronidazole and vancomycin used to treat C difficile.^{Reference Turnbaugh, Ley and Hamady50,Reference Mikkelsen, Frost, Bahl, Licht, Jensen and Rosenburg64,Reference Lankelma, Cranendonk, Belzer, de Vos, de Vos and van der Poll70} On the other hand, there is a relative lack of data for tetracyclines^{Reference Angelakis, Million, Kankoe, Lagier, Armougom and Giorgi61} and macrolides.^{Reference Jakobsson, Jernberg, Andersson, Sjolund-Karlsson, Jansson and Engstrand40,Reference Korpela, Salonen, Virta, Kekkonen, Forslund and Bork59,Reference Villarreal, Aberger, Benrud and Gundrum69}

In a large cohort study of Finnish children, macrolides were shown to induce long-term alterations of microbiota; for instance, there were reductions in actinobacteria (mainly bifidobacteria), Firmicutes (mainly lactobacilli) and total bacterial diversity, and increases in the relative abundance of Bacteroidetes and Proteobacteria. These results are supported by a study investigating the long-term impact of a short-term course of clarithromycin and metronidazole on patients with Helicobacter pylori infection.^{Reference Jakobsson, Jernberg, Andersson, Sjolund-Karlsson, Jansson and Engstrand40} Doxycycline has been shown to reduce faecal bacterial concentrations of Bacteroidetes, Firmicutes and lactobacillus in a study with healthy volunteers.^{Reference Liss and Norman33}

Antibiotics from different classes, such as tetracyclines and macrolides, are expected to cause unique patterns of microbiota alteration because of their differing spectra of activity and bacterial targets.^{Reference Ianiro, Tilg and Gasbarrini63} Accordingly, they may play different roles in the development of acute adverse gastrointestinal symptoms or acute gastrointestinal infection. By investigating the short-term changes in the gastrointestinal microbiome in parallel with the nasal microbiome with commonly prescribed medication for chronic rhinosinusitis, we can better understand the effects of current medical management, including their adverse effects. This will allow the future development and delivery of more targeted therapy for all chronic rhinosinusitis patients.