Antipsychotic (AP) safety has been widely investigated. However, mechanisms underlying AP-associated pneumonia are not well-defined.
The aim of this study was to investigate the known mechanisms of AP-associated pneumonia through a systematic literature review, confirm these mechanisms using an independent data source on drug targets and attempt to identify novel AP drug targets potentially linked to pneumonia.
A search was conducted in Medline and Web of Science to identify studies exploring the association between pneumonia and antipsychotic use, from which information on hypothesized mechanism of action was extracted. All studies had to be in English and had to concern AP use as an intervention in persons of any age and for any indication, provided that the outcome was pneumonia. Information on the study design, population, exposure, outcome, risk estimate and mechanism of action was tabulated. Public repositories of pharmacology and drug safety data were used to identify the receptor binding profile and AP safety events. Cytoscape was then used to map biological pathways that could link AP targets and off-targets to pneumonia.
The literature search yielded 200 articles; 41 were included in the review. Thirty studies reported a hypothesized mechanism of action, most commonly activation/inhibition of cholinergic, histaminergic and dopaminergic receptors. In vitro pharmacology data confirmed receptor affinities identified in the literature review. Two targets, thromboxane A2 receptor (TBXA2R) and platelet activating factor receptor (PTAFR) were found to be novel AP target receptors potentially associated with pneumonia. Biological pathways constructed using Cytoscape identified plausible biological links potentially leading to pneumonia downstream of TBXA2R and PTAFR.
Citation: Sultana J, Calabró M, Garcia-Serna R, Ferrajolo C, Crisafulli C, Mestres J, et al. (2017) Biological substantiation of antipsychotic-associated pneumonia: Systematic literature review and computational analyses. PLoS ONE 12(10): e0187034. https://doi.org/10.1371/journal.pone.0187034
Editor: Kenji Hashimoto, Chiba Daigaku, JAPAN
Received: May 31, 2017; Accepted: October 12, 2017; Published: October 27, 2017
Copyright: © 2017 Sultana et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All the pharmacological data used for this study is known, publicly available, experimental data. For CT-link analyses the following free and publicly available database was used: ChEMBL (http://www.ebi.ac.uk/chembl/). For Cytoscape analyses the following databases were used, all of which are freely available online: Gene Ontology Consortium databases (GO databases) http://www.geneontology.org/, Kyoto Encyclopedia of Genes and Genomes, also known as KEGG (http://www.genome.jp/kegg/), Reactome Pathway Database (http://reactome.org/), Pathway Common (http://www.pathwaycommons.org/pc/dbSources.do), WikiPathways (http://www.wikipathways.org/index.php/WikiPathways) and GeneMania databases (http://genemania.org/data/). Drugbank database (https://www.drugbank.ca/), another freely available online data source, was also used to further identify pharmacological data. All the data above was accessed through Cytoscape plug-ins (as indicated on main text). Finally, the Pubmed website (https://www.ncbi.nlm.nih.gov/pubmed/) was used to manually integrate recently published related data.
Funding: I have read the journal's policy and the authors of this manuscript have the following competing interests: RGS and JM are employed by Chemotargets SL. The funder provided support in the form of salaries and the materials to conduct part of the present study for authors RGS and JM, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the ‘Author Contributions’ section. This study was published with financial support from by the Italian Health Ministry in the context of the Project “Assessment of the safety of antipsychotic drugs in elderly with dementia: an international, population-based study using healthcare databases” [grant number GR-2009-1607316], of which GT is the grant holder. The Italian Health Ministry did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: RGS and JM are employees of Chemotargets SL. This does not alter their adherence to PLOS ONE policies on sharing data and materials. The other authors have no competing interests to declare.
Antipsychotics drugs (APs) have been associated with several adverse reactions including stroke, acute kidney disease and pneumonia, especially in aged patients [1–3]. In 2009 the Food and Drug Administration issued a warning on the increased risk of death when APs were used off-label in persons with dementia, pneumonia being among the most commonly reported causes of death. Many observational studies have evaluated the association between APs and the risk of community or hospital-acquired pneumonia in the elderly population [4–14]. However, most studies investigated APs by class rather than by individual drug, with a few exceptions. In this respect, classifying drugs as conventional or atypical is a traditional perspective still reminiscent of the one drug–one target paradigm . It is now widely recognized that most drugs bind with different affinities to multiple proteins, including but not limited to traditional receptor targets , a property often referred to as polypharmacology .This makes each single drug essentially unique pharmacologically.
The biological pathways underlying antipsychotic-induced pneumonia are not currently known, although several plausible hypotheses have been postulated . Understanding which drugs are associated with the highest risk of pneumonia would allow prescribers to select drugs with a lower risk of pneumonia in persons with risk factors predisposing them to infectious respiratory diseases. Such risk factors include being bedridden, having chronic respiratory diseases and/or being prescribed sedating drugs. There is emerging data from observational studies suggesting a different safety profile for individual APs, which may be missed when studying drug effects by class [19, 20]. The potential effects of such differences in adverse drug reactions (ADRs) among single APs could in turn have an impact on AP persistence and healthcare utilization as a result of ADRs, or relapse of symptoms because of differential AP drug discontinuation. While observational studies are a crucial starting point for the identification of increased risk of adverse drug reactions they cannot point to an underlying biological mechanism. Nevertheless, identifying a biologically plausible cause for drug-related risk, i.e. signal substantiation, is essential to confirm the validity of such risk and identify new off-targets that otherwise cannot be predicted on the basis of ‘traditional’ systems pharmacology, i.e. cholinergic, histaminergic, or serotonergic systems etc. Signal substantiation is defined as any activity aiming to provide a biological explanation of why an ADR occurs. The rationale behind this is that an ADR must be biologically plausible to increase the feasibility of a causal relationship between drug and ADR, in particular for type A ADRs, i.e. those which are dose-dependent and are related to the drug’s known pharmacological activity.
The aims of this study were to: 1) conduct a systematic review aiming to identify all hypothesized mechanisms of AP-associated pneumonia in published literature, including studies employing any study designs (including case reports), in which persons of any age used APs for any indication, provided that the outcome was pneumonia; 2) confirm the most relevant drug targets potentially involved in AP-associated pneumonia using computational approaches for drug safety profiling , as well as to identify novel drug targets, with the ultimate goal of extrapolating clinically relevant information.
Methods and materials
Medline and Web of Science were searched for literature from their inception to 27th February 2017 using the following terms: “pneumonia” and “antipsychotic agents”. Articles were excluded if they were not published in English language or if the full text of the article was not accessible even after contacting the author. Eligibility criteria included treatment with APs for any indication in persons of any age and pneumonia as the outcome under investigation. All study designs and article types were eligible for inclusion, including case reports, letters to the editor, observational studies and systematic reviews/meta-analyses. For observational studies reporting a risk of pneumonia with AP use, the study design, population studied, exposure and comparator, outcome investigated, hypothesized mechanism underlying AP-associated pneumonia and risk estimate was reported. Two investigators independently examined the titles and abstracts of selected articles and obtained full texts of potentially relevant papers. Disagreement on inclusion was resolved by discussion. The findings were reporting accordingly the PRIMA checklist (S1 Fig).
Screening of known and predicted antipsychotic binding to receptors and associated safety events
CT-link software  was used to analyze the experimentally known pharmacology and its potential link to known safety outcomes of AP drugs. Known in vitro pharmacology data was extracted from ChEMBL . Statistically significant protein signatures associated to each safety outcome were generated by analyzing the confusion matrix resulting from the number of drugs with affinity above or below a certain affinity threshold for a given target and associated, or otherwise, to a given safety event. Protein signatures were defined for 1,455 safety terms, mapped to Unified Medical Language System (UMLS) codes, and classified in 25 toxicity categories . CT-link was screened for targets and safety outcomes associated with a set of 7 APs commonly investigated in the relation to the risk of pneumonia, namely, amisulpride, clozapine, haloperidol, olanzapine, quetiapine, risperidone, and zotepine [25–29].
Once targets of interest were identified in CT-Link (i.e., novel targets), the pathways that could potentially explain a novel biological link between AP off-targets, i.e., binding targets other than intended ones, and safety terms related to pneumonia were identified and investigated using the following software: Cytoscape , the GeneMania , ClueGO  and CyTarget Linker  plug-ins (S3 Fig). Cytoscape software was used to find group of elements, either genes or proteins, that interact with the user-defined query (which consists of one or more genes/proteins), representing them graphically. Each of the elements (either genes or proteins) found by Cytoscape are referred as “nodes” (graphically represented by circles). The interactions are referred as edges (graphically represented by lines connecting different nodes).
To construct a possible molecular pathway downstream of antipsychotic binding to its off-targets, two approaches were used. In the first approach, a “Physical Interactions” database was used  to identify possible proteins and/or metabolic products that may be responsive to the function of the identified proteins to which AP drugs bind. Cytoscape was used to identify all known nodes which physically interact with AP off-targets, using data inputs from the “Physical Interactions” database of geneMania plug-in. This step was iterated, using the results of the previous step as input, to widen the biological net of physical interactors.
In the second approach, the “Pathway Commons”  database was used in order to identify nodes related to the function of the off-targets identified. For this step, Cytoscape was further screened to identify the pathways that are functionally linked with AP protein targets and pneumonia-related symptoms. The “Pathway Commons” database of the geneMania plug-in was used in this step. All molecular cascades, including those weakly related to our starting query, were identified. Nodes not related to the phenotype under investigation were removed manually.
Selection and clustering of candidate genes from the biological net identified: Mapping of hypothetical molecular cascades leading to AP-derived pneumonia due to novel targets
The relevance of the interactors to pneumonia-related symptoms was checked through the use of Medline and Cytoscape built-in databases. The nodes having a stronger association were highlighted. These are hereafter referred to as associated nodes(AN). Nodes within the biological net, linking the associated nodes to identified AP off-targets were clustered, where a cluster is defined as a subgroup of nodes within the biological net under investigation that share a particular characteristic. In this case, the common characteristic is that they are related to both AP targets and the associated nodes found in the biological net. ClueGo  and CluePedia  plug-ins were then used to identify pathways involved in the function of the nodes included in the clusters.
The results of the article selection process are reported in Fig 1. The search in Medline and Web of Science yielded 200 articles. Of these, only 48 articles were deemed suitable for further consideration as they specifically concerned APs as drugs of interest which may have led to pneumonia. Seven articles were subsequently excluded because they were not in English or had no full text available, even after the authors were contacted (see S1 Table for details on excluded studies with no full text). Forty-one articles were included in the final assessment and of these, the majority were observational studies (N = 16) [1, 4, 6–10, 12, 13, 18, 25–29, 35–46]. Of the 41 articles included, 30 hypothesized a mechanism of action. Ten studies were reviews/meta-analyses [1, 13, 18, 27, 41–46], while other studies included 12 case reports [47–58]and one letter to the editor. Most of the literature concerned patients aged 65 and older [1, 4, 6, 8–10, 12, 18, 35, 38, 41, 45, 47, 49, 51] while others investigated an adult population with bipolar/schizophrenic disorders [26, 36, 37, 40, 48, 50, 55, 56, 59]. No randomized clinical trials (RCTs) were identified, but one meta-analysis of 6 RTCs was found (S2 Fig). Limited information on the effect of dose and duration regarding the risk of pneumonia was available, although some studies suggested an increased risk of pneumonia with higher AP doses [4, 29] and mostly close to the start of treatment , while another study found no clear link between the duration of AP exposure and risk of pneumonia. Some groups of AP-treated persons were found to be at higher risk of pneumonia, including persons with Parkinson’s disease , schizophrenia , and those concomitantly using multiple antipsychotic drugs .
The most widespread theories about the biological mechanisms underlying pneumonia due to antipsychotic use concern the affinity of the antipsychotics to cholinergic [1, 4, 6, 8, 9, 12, 18, 25, 26, 36, 37, 43, 45, 48, 51, 59], histaminergic [4, 6, 9, 25, 26, 43–45, 59] and dopaminergic [1, 6, 9, 18, 25, 36, 41–44, 49, 58]receptors (Table 1). The clinical events triggered by APs binding to their targets and leading to pneumonia are summarized in Table 1.These include dry mouth and impaired peristalsis due to muscarinic M1 receptor blockade, as well as sedation with subsequent impaired swallowing due to histamine H1 receptor blockade. Lesser known mechanisms proposed involved the immune system [6, 10, 26, 36, 41, 43], cytochrome metabolism system , inflammatory pathways [15, 41], an eotaxin (i.e., a chemoattractant for eosinophils) and serotonin eosinophilic-specific chemo-attracting action leading to eosinophilic pneumonia .
Most of the hypotheses concerned antipsychotic drug classes rather than individual drugs, or referred to antipsychotics in general. Several studies did not specify a biological mechanism leading to pneumonia but described the antipsychotic-associated clinical signs or symptoms which may cause pneumonia [10, 15, 35, 42, 44, 50, 53, 54]. The most commonly hypothesized physiological/clinical conditions leading to antipsychotic-associated pneumonia were extrapyramidal symptoms (5 studies reporting extrapyramidal symptoms with conventional APs, 5 studies where AP class was not specified; none for atypical APs), sedation (7 studies reporting sedation with conventional APs, 5 with atypical APs, 6 where class was not specified), xerostomia (6 studies reporting xerostomia after atypical AP use, 2 studies not specifying AP drug class and none for conventional APs) and oropharyngeal dyskinesia/impaired swallowing (10 studies reporting this for atypical APs, 8 without specifying a drug class; none for conventional APs). Less common clinical features included aspiration, sialorrhea, agranulocytosis, and suppression of the cough reflex.
Of all the observational studies/reviews identified, 17 reported a risk estimate of pneumonia associated with antipsychotic use (Table 2). Most risk estimates were available for AP drug classes rather than individual drugs, suggesting that the overall risk of pneumonia associated with atypical APs is similar to conventional APs, when non-use of APs was the reference group. No comparison of pneumonia risk could be made for individual APs due to the variety of reference groups used.
Off-target pharmacology and links to respiratory safety events
The results of the search in CT-link for the known AP receptor-binding profile broadly confirmed the results of the literature review (Fig 2). The comparison of the experimental affinity profiles of the 7 APs across a set of proteins for which in vitro affinity data was available shows that, even though these drugs are all considered to belong to the same pharmacological class, there are significant differences between their pharmacologic profiles. For example, amisulpride and haloperidol show a clear binding preference for the dopamine receptors D2 (DRD2) and D3 (DRD3), while having a much weaker affinity for the histamine H1 receptor (HRH1) than the other APs. In addition, with the exception of quetiapine and amisulpride, all APs bind strongly to the serotonin receptor subtypes 5-HT2A (HTR2A) and 5-HT2C (HTR2C). Also, in general, clozapine and risperidone seem to interact more strongly with the alpha-2A adrenergic receptors than the other APs. Finally, olanzapine and clozapine interact markedly with the muscarinic receptors compared to other APs. Altogether, the unique polypharmacology of every individual antipsychotic will ultimately translate in a unique drug safety profile associated with those protein interactions.
The affinities for the receptors TRXA2R and PTAFR concern predicted, not experimentally known, values; these are presented here for comparison purposes. Abbreviations: HTR- serotonin receptors; ADR: adrenergic receptors; DRDs- dopamine receptors; HRHs: histamine receptors; CHRs- muscarinic receptors; KCNH2: hERG transporter, SLC6A3: dopamine transporter, SLC6A4: serotonin transporter; TBXA2R- thromboxane A2 receptor; PTAFR- platelet activating factor receptor. Color coding reflects the experimentally known pKi values, yellow being inactive (pKi = 4; Ki = 100 nM), red being highly active (pKi = 9; Ki = 1 nM), and grey meaning that no data is available for that interaction.
Most drugs are the result of a long and careful drug design process aiming at maximizing the affinity of a drug for its primary intended target, in the case of antipsychotics, dopaminergic receptor blockade. The primary targets of a drug are therefore those proteins for which the drug has strongest affinity. Accordingly, a total of 41 protein interactions with binding affinity stronger than 10nM (pKi ≥ 8.0) are currently experimentally known for those 7 APs . Of these, 39 interactions (95%) are with aminergic G protein-coupled receptors. They will be most likely the main factor responsible for some of the antipsychotic-associated pneumonias reported. However, in spite of the optimization process, most drugs will still have residual, weaker, affinities for other proteins. In the case of the 7 APs in this study, the number of known protein interactions goes up to 184 when considering weaker binding affinities, such as 1μM (pKi ≥ 6.0). This off-target pharmacology, i.e. drugs targets other than the known and/or expected ones, could also contribute and be partly responsible for some of the respiratory safety events linked to pneumonia. The identified off-targets as potentially novel mediators of AP-associated pneumonia were then screened using a repository of drugs and their associated safety events. The thromboxane A2 receptor (TBXA2R) and the platelet-activating factor receptor (PTAFR) were preliminarily identified as such novel off-targets of interest for which micromolar affinity was observed for all 7 APs except for amisulpride.
A repository of the data from drug labels, safety predictions and ontologies in CT-link was used to screen safety events associated with the set of 7 APs. It was found that a total of 95 respiratory safety terms were associated with these antipsychotics. The most commonly identified terms were alveolitis, lung diseases, dyspnea, and respiration disorders (S2 Table). The antipsychotics most consistently linked to pneumonia-related signals were quetiapine and risperidone, whereas the least associated with pneumonia-related signals were olanzapine and clozapine. Interestingly, the latter were the two antipsychotics having a receptor binding profile markedly different from the other APs in their preferred binding for muscarinic receptors (Fig 2).
Among the off-targets identified, further analysis was carried out for the two potentially novel proteins involved in AP-associated pneumonia, TBXA2R and PTAFR. Drug safety events (including respiratory events) in CT-link are linked to data on the receptors potentially associated with that safety event. Pneumonia in relation to the antipsychotics screened was indeed found to be associated with TBXA2R and PTAFR.
Biological pathway construction
Findings on the potential role of TBXA2R and PTAFR in the increased risk of antipsychotic-associated pneumonia derived from CT-link were further investigated using Cytoscape. Several candidate genes encoding for proteins that physically interact with TBXA2R and PTAFR were identified. The pathways involved in the function of the clustered genes are shown in Figs 3 and 4. The identified AP off-targets, TBXA2R and PTAFR, in black, are linked, either directly or indirectly, to a group of nodes, shown in blue, which likely contribute to the increased risk of developing pneumonia-related symptoms after activation of TBXA2Ror PTAFR. The molecular cascade highlighted in these analyses is involved in a wide range of biological functions. The biological processes (identified from the Gene Ontology Database - http://geneontology.org/) related to the molecular cascade for both TBXA2R and PTAFR are indicated in Fig 5, while S3 and S4 Tables report the details of biological processes and sub-processes indicated in Fig 5. As shown by the supplementary files (S3–S6 Figs), several pathways were influenced by the function of the clustered nodes. While some of these clusters point to biological functions which seem completely unrelated to pneumonia-related symptoms, others are potentially related to pneumonia. Further detail on the web of biological functions related to PTAFR and TBAX2R are shown in the supplementary files (S7–S9 Figs).
TBXA2R is shown in black, associated nodes are represented in blue, primary interactors of TBXA2 are shown in red and secondary interactors are shown in grey. The nodes are linked through physical interactions (data obtained from geneMania database). Abbreviations- ADRB1: Adrenoceptor Beta 1; ARRB2: Arrestin Beta 2; ADRB1: Arrestin Beta 1;AGTR1: Angiotensin II Receptor Type 1; AGTR1: Angiotensin II Receptor Type 1; ADRB2: Adrenoceptor Beta 2; TBXA2R: Thromboxane A2 Receptor; AGTRAP: Angiotensin II Receptor Associated Protein;BDKRB2: Bradykinin Receptor B2;CAV3:Caveolin 3; CDH15: Cadherin 15; CDH2:Cadherin 2;EDNRA: Endothelin Receptor Type A;G3BP2: G3BP Stress Granule Assembly Factor 2; GNA11: G Protein Subunit Alpha 11;GNAI2: G Protein Subunit Alpha I2;GNA12: G Protein Subunit Alpha 12;GNA13: G Protein Subunit Alpha 13;GNAQ: G Protein Subunit Alpha Q;GNAS: GNAS Complex Locus;GPRASP1: G Protein-Coupled Receptor Associated Sorting Protein 1;GRK5: G Protein-Coupled Receptor Kinase 5;HTR2B: 5-Hydroxytryptamine Receptor 2B;HRH2: Histamine Receptor H2; ITGB1BP1: Integrin Subunit Beta 1 Binding Protein 1; KCNMA1: Potassium Calcium-Activated Channel Subfamily M Alpha 1;NME1-MNE2: NME/NM23 nucleoside diphosphate kinase 1;OPRD1:Opioid Receptor Delta 1;OPRK1: Opioid Receptor Kappa 1;OPRM1: Opioid Receptor Mu 1;PRKCA: Protein Kinase C Alpha; RAF1: Raf-1 Proto-Oncogene, Serine/Threonine Kinase;WDR36: WD Repeat Domain 36; YWHAZ: Tyrosine 3-Monooxygenase/Tryptophan 5-Monooxygenase Activation Protein Zeta.
PTAFR is shown in black, associated nodes from Table 1 are represented in blue, primary interactors of TBXA2R are shown in red and secondary interactors are shown in grey.The nodes are linked through physical interactions (red edges) and pathway interactions (blue edges). The data was obtained from geneMania software. Abbreviations- ADORA1: Adenosine A1 Receptor;ADRB1:Adrenoceptor Beta 1; CCK: Cholecystokinin; CDC42:Cell Division Cycle 42; CNR2:Cannabinoid Receptor 2;ENG:Endoglin;GHR:Growth Hormone Receptor; GLP1R: Glucagon Like Peptide 1 Receptor;GNAI3:G Protein Subunit Alpha I3;GNLY:Granulysin; HDAC5: Histone Deacetylase 5; IKBKB: Inhibitor Of Nuclear Factor Kappa B Kinase Subunit Beta; IL13RA1: Interleukin 13 Receptor Subunit Alpha 1; LYN:LYN Proto-Oncogene, Src Family Tyrosine Kinase; OXTR:Oxytocin Receptor;PTGDS: Prostaglandin D2 Synthase; PTAFR: platelet activating factor receptor;PTGER3: Prostaglandin E Receptor 3;RELN:Reelin; RGS16: Regulator Of G-Protein Signaling 16; SLC40A1: Solute Carrier Family 40 Member 1; TACR1: Tachykinin Receptor 1.
Drug safety is an important public health concern, as can be seen by the continuous efforts in pharmacovigilance to identify, quantify and minimize drug-related risks. Over the years, antipsychotic safety has been investigated using both spontaneous reporting system databases as well as electronic medical records or claims databases. Whenever possible, the plausibility of associations between adverse drug events and drug use is strengthened by confirming the biological plausibility of such events, especially in case of “type A” adverse drug reaction which are dose-dependent and are based on pharmacological properties of the drug mechanism of action effects.
Of the 41 articles identified in the systematic literature review, only 30 hypothesized a pharmacological mechanism that could explain the antipsychotic-associated risk of pneumonia. Furthermore, only one of these studies was designed in such a way that could reasonably associate pneumonia to specific antipsychotic sub-groups, and therefore to a group of antipsychotics with closely related physicochemical properties (e.g., butyrophenones and phenothiazines among conventional antipsychotics).
The approach underlying currently hypothesized mechanisms is limited because it is based on the traditional view of ‘systems pharmacology’, thus excluding a priori any“off-target” receptors and focusing on commonly known pharmacological systems, such as cholinergic, histaminergic, serotonergic and dopaminergic classes. As a result, other targets potentially mediating events of interest are not captured. A case in point is the identification of two novel receptor targets which were not identifiable based on the major pharmacologic systems.
The main clinical implications of the study findings primarily illustrate that antipsychotics of both classes have plausible mechanisms leading to pneumonia and/or lead to signs/symptoms that increase the risk of pneumonia. As a result, all APs should be used with caution and sparingly in persons at risk of pneumonia, especially the elderly. Indeed, the finding of similar risk of pneumonia across classes in the literature review conducted is consistent with a recently published systematic review and meta-analysis . Nevertheless, the very diverse target binding among single APs, as highlighted by the analysis of experimentally-known AP target-binding, suggests that grouping antipsychotics into broad groups such as ‘atypical’ and ‘conventional’ may not be helpful or accurate. This method of classifying antipsychotics is based on the point in time when antipsychotics were marketed, with the older APs being termed conventional or first generation and newer ones being referred to as atypical or second generation. In vitro findings presented in this study confirm that such a classification is misleading, particularly concerning atypical APs, because there is a significant pharmacological heterogeneity among individual APs. Equally prone to being misleading are proposed mechanisms of action for pneumonia, or other adverse outcomes, that assumed all AP drugs to have the same or similar pharmacological action. The literature review conducted contains several such examples of mechanisms of action which related to all APs. Researchers should not be content with evaluating the risk of an outcome but should question the reason why such events occur, considering that biological feasibility is one of the key criteria for causality. The present study suggests that off-target pharmacology can play an important role in substantiating known signals. The novel mechanisms concerning TBXA2R and PTAFR as potential mediators of pneumonia symptoms constitute an important finding of this study.
PTAFR has already been linked to an increased risk of pneumococcal disease in particular by facilitating the adhesion of pneumococcal bacteria to endothelial cells in the respiratory tract, according to a recently published narrative review on the link between PTAFR and pneumococcal disease using in vitro and in vivodata. Platelet-activating factor binds to PTAFR, activating the enzyme GTPase, up-regulating phospholipase C, D and A2 pathways and activating protein kinase C as well as tyrosine kinases, leading to an overall increase in inflammatory response. TBXA2R appears to have been less extensively studied than PTAFR in the context of pneumonia, but has nevertheless been implicated in bronchial hyper-responsiveness  as well as bronchial constriction  in animal models, both of which may be risk factors for pneumonia. We hypothesize that the increased risk of pneumonia-like symptoms mediated by PTAFR and TBXA2R may be exerted through the control of capillary permeability at the alveolar level. Impaired control of permeability may increase the quantity of fluids in the alveoli, thus causing edema, a primary risk factor as well as symptom of pneumonia. In addition, pathways involved with the gene/protein clusters identified in relation to pneumonia symptoms also have a role in maintaining the structure of blood vessels. Impairment of this function may negatively impact circulatory system’s physical properties, leading to leaking of fluids in the alveoli. Control of platelet and coagulation homeostasis may also play a role in the increase of fluids within alveoli after TBXA2R and PTAFR activation. Future studies should aim to quantify the affinity of APs and other drugs associated with a risk of pneumonia also to these two receptors in vitro. From a clinical point of view, it is relevant to identify the drug receptor-binding profiles related leading to a specific adverse effect, in this case, pneumonia, as this may help prevent the concomitant use of drugs having a potentially synergistic or additive risk of that adverse effect. Furthermore, future studies should focus more on the single drug safety profile rather than grouping antipsychotic drugs by class, as this may help to identify high risk drugs as well as high risk patient groups and better support individualized medicine.
This paper has several strengths. The literature review allowed us to screen published material for hypothesized mechanisms underlying pneumonia and understand what importance is given to this aspect of drug safety. In addition, the in silicoapproaches to drug safety substantiation used in the present paper build on each other, first using off-target pharmacology to identify relevant novel receptors associated with pneumonia and then using Cytoscape to construct likely biological cascades connecting the novel receptors to pneumonia symptoms. Indeed, similar approaches to better drug safety, such as “medication-wide association studies” and systems pharmacology  have been proposed to augment the capacity of currently used pharmacovigilance systems. Findings from this study confirmed existing receptor targets for AP drugs but also provided two novel off-targets that may lead to AP-associated pneumonia, namely PTAFR and TBXA2R.
There are however some limitations that should be borne in mind. Some of the hypothesized signal substantiation mechanisms suggested in the published literature were found to be of limited value as they did not mention a specific drug but focused on an entire drug class (conventional or atypical) or even on antipsychotics in general. This constituted a limitation in the starting point of collecting hypotheses, as it is unlikely that such suggested mechanisms underlying AP-associated pneumonia are completely reliable given the physicochemical differences between each drug. With regards to the safety predictions generated by CT-link and the generation of molecular cascades with Cytoscape, there are at least two inherent limitations associated with any approach relying on signals therein. On one hand, the experimentally known in vitro affinity data for drugs may be incomplete , which can have an impact on the list of proteins included in the target signatures linked to safety terms and thus, on the ability to identify protein targets likely to be responsible for the AP-associated pneumonia. Given the data sources used to carry out this study, it was not possible to examine whether there are some mechanisms that are more strongly associated with the occurrence of pneumonia than others. Finally, this work focused on a set of seven antipsychotic drugs for which there is a considerable amount of bibliographical information in the context of the risk of pneumonia [25–29].Therefore, our results cannot be considered exhaustive with regards to the current portfolio of over 50 antipsychotic medications but can certainly be taken as a reference from which substantiation analyses of AP-associated pneumonia could be performed.
The literature review identified several biological mechanisms that could lead to pneumonia with AP use. In vitro pharmacology data confirmed receptor affinities identified in theliterature review. Two targets, thromboxane A2 receptor (TBXA2R) and plateletactivating factor receptor (PTAFR) were found to be novel AP target receptorspotentially associated with pneumonia. Biological pathways constructed usingCytoscape identified plausible biological links potentially leading to pneumoniadownstream of TBXA2R and PTAFR. Innovative approaches for biologicalsubstantiation of drug-adverse event associations may strengthen evidence on drugsafety profiles and help tailoring pharmacological therapies.
S2 Fig. Detail of study extraction during systematic review.
Abbreviations- AP: antipsychotic; RCT: randomized controlled trial.
S3 Fig. Flowchart of the analysis carried out in Cytoscape.
Abbreviations- ANs: associated nodes; AP: antipsychotic. Definitions- Biological net: This refers to a group of genes/proteins which interact with each other. Associated nodes: The term refers to genes/proteins, within the biological net, which are linked to pneumonia related processes, according to literature data (mainly from Medline database).
Panel A: The biological net associated with TBXA2R obtained through Cytoscape’s “Physical Interaction” enrichment. Panel B: The direct interactors of TBXA2R within the biological net. Panel C: the downstream elements of the biological net (in blue) and the connecting nodes (in grey).Abbreviation-TBXA2R: thromboxane A2 receptor.
Panel A: The biological net associated with PTAFR obtained through Cytoscape’s “Physical Interaction” enrichment. Panel B: The direct interactors of PTAFR within the biological net. Panel C: the downstream elements of the biological net (in blue) and the connecting nodes (in grey).Abbreviation- PTAFR: platelet activating factor receptor.
Panel A: The biological net related to TBXA2R obtained through Cytoscape’s “Pathway interaction” enrichment. Panel B: Direct interactors of TBXA2R within the biological net. Panel C: The downstream elements of the biological net (in blue) and the connecting nodes (in grey).Abbreviation-TBXA2R: thromboxane A2 receptor.
Panel A: the biological net related to PTAFR obtained through Cytoscape’s “pathway interaction” enrichment. Panel B: direct interactors of PTAFR within the biological net. Panel B: the downstream elements of the biological net (in blue) and the connecting nodes (in grey).Abbreviation- PTAFR: platelet activating factor receptor.
S8 Fig. Biological processes (colored captions and nodes) and sub-processes (captions underneath each node) associated with TBXA2R-related elements.
Nodes belonging to multiple processes are indicated by multicolored sections.Abbreviation-TBXA2R: thromboxane A2 receptor.
S9 Fig. Biological processes (colored captions and nodes) and sub-processes (captions underneath each node) associated with PTAFR-related elements.
Nodes belonging to multiple processes are indicated by multicolored sections. Abbreviation- PTAFR: platelet activating factor receptor.
S1 Table. List of studies excluded due to lack of available full-text.
The information in table above was identified from the abstract, where this was available.
S2 Table. Main respiratory safety terms, potentially related to pneumonia, which may be associated with antipsychotics as found in CT-link.
S3 Table. The biological processes associated with the genes of biological net obtained for TBXA2R.
Abbreviation-TBXA2R: thromboxane A2 receptor.
- 1. Trifiro G., Sultana J., and Spina E., Are the safety profiles of antipsychotic drugs used in dementia the same? An updated review of observational studies. Drug Saf, 2014. 37(7): p. 501–20. pmid:24859163
- 2. Jiang Y., McCombs J.S., and Park S.H., A Retrospective Cohort Study of Acute Kidney Injury Risk Associated with Antipsychotics. CNS Drugs, 2017. 31(4): p. 319–326. pmid:28290080
- 3. Ryan P.B., et al., Atypical Antipsychotics and the Risks of Acute Kidney Injury and Related Outcomes Among Older Adults: A Replication Analysis and an Evaluation of Adapted Confounding Control Strategies. Drugs Aging, 2017. 34(3): p. 211–219. pmid:28124262
- 4. Trifiro G., et al., Association of community-acquired pneumonia with antipsychotic drug use in elderly patients: a nested case-control study. Ann Intern Med, 2010. 152(7): p. 418–25, W139-40. pmid:20368647
- 5. Huybrechts K.F., et al., Differential risk of death in older residents in nursing homes prescribed specific antipsychotic drugs: population based cohort study. BMJ, 2012. 344: p. e977. pmid:22362541
- 6. Knol W., et al., Antipsychotic drug use and risk of pneumonia in elderly people. J Am Geriatr Soc, 2008. 56(4): p. 661–6. pmid:18266664
- 7. Barnett M.J., et al., Risk of mortality associated with antipsychotic and other neuropsychiatric drugs in pneumonia patients. J Clin Psychopharmacol, 2006. 26(2): p. 182–7. pmid:16633149
- 8. Gau J.T., et al., Pharmacotherapy and the risk for community-acquired pneumonia. BMC Geriatr, 2010. 10: p. 45. pmid:20604960
- 9. Aparasu R.R., Chatterjee S., and Chen H., Risk of pneumonia in elderly nursing home residents using typical versus atypical antipsychotics. Ann Pharmacother, 2013. 47(4): p. 464–74. pmid:23548651
- 10. Star K., et al., Pneumonia following antipsychotic prescriptions in electronic health records: a patient safety concern? Br J Gen Pract, 2010. 60(579): p. e385–94. pmid:20883613
- 11. Hatta K., et al., Antipsychotics for delirium in the general hospital setting in consecutive 2453 inpatients: a prospective observational study. Int J Geriatr Psychiatry, 2014. 29(3): p. 253–62. pmid:23801358
- 12. Huybrechts K.F., et al., Risk of death and hospital admission for major medical events after initiation of psychotropic medications in older adults admitted to nursing homes. CMAJ, 2011. 183(7): p. E411–9. pmid:21444611
- 13. Pratt N., et al., Choice of observational study design impacts on measurement of antipsychotic risks in the elderly: a systematic review. BMC Med Res Methodol, 2012. 12: p. 72. pmid:22682666
- 14. Wang P.S., et al., Ventricular arrhythmias and cerebrovascular events in the elderly using conventional and atypical antipsychotic medications. J Clin Psychopharmacol, 2007. 27(6): p. 707–10. pmid:18004143
- 15. Abdelmawla N. and Ahmed M.I., Clozapine and risk of pneumonia. Br J Psychiatry, 2009. 194(5): p. 468–9. pmid:19407284
- 16. Mestres J., et al., Data completeness—the Achilles heel of drug-target networks. Nat Biotechnol, 2008. 26(9): p. 983–4. pmid:18779805
- 17. Peters J.U., Polypharmacology—foe or friend? J Med Chem, 2013. 56(22): p. 8955–71. pmid:23919353
- 18. Gambassi G., Sultana J., and Trifiro G., Antipsychotic use in elderly patients and the risk of pneumonia. Expert Opin Drug Saf, 2015. 14(1): p. 1–6. pmid:25431005
- 19. Huybrechts K.F., et al., Comparative safety of antipsychotic medications in nursing home residents. J Am Geriatr Soc, 2012. 60(3): p. 420–9. pmid:22329464
- 20. Rafaniello C., et al., Predictors of mortality in atypical antipsychotic-treated community-dwelling elderly patients with behavioural and psychological symptoms of dementia: a prospective population-based cohort study from Italy. Eur J Clin Pharmacol, 2014. 70(2): p. 187–95. pmid:24145814
- 21. Bauer-Mehren A., et al., Automatic filtering and substantiation of drug safety signals. PLoS Comput Biol, 2012. 8(4): p. e1002457. pmid:22496632
- 22. Garcia-Serna R., et al., Large-Scale Predictive Drug Safety: From Structural Alerts to Biological Mechanisms. Chem Res Toxicol, 2015. 28(10): p. 1875–87. pmid:26360911
- 23. Gaulton A., et al., ChEMBL: a large-scale bioactivity database for drug discovery. Nucleic Acids Res, 2012. 40(Database issue): p. D1100–7. pmid:21948594
- 24. Remez N., et al., The In Vitro Pharmacological Profile of Drugs as a Proxy Indicator of Potential In Vivo Organ Toxicities. Chem Res Toxicol, 2016. 29(4): p. 637–48. pmid:26952164
- 25. Mehta S., et al., Comparative safety of atypical antipsychotics and the risk of pneumonia in the elderly. Pharmacoepidemiol Drug Saf, 2015. 24(12): p. 1271–80. pmid:26445931
- 26. Kuo C.J., et al., Second-generation antipsychotic medications and risk of pneumonia in schizophrenia. Schizophr Bull, 2013. 39(3): p. 648–57. pmid:22282455
- 27. Haupt M., Cruz-Jentoft A., and Jeste D., Mortality in elderly dementia patients treated with risperidone. J Clin Psychopharmacol, 2006. 26(6): p. 566–70. pmid:17110812
- 28. Kim D.H., et al., Adverse Events Associated with Antipsychotic Use in Hospitalized Older Adults After Cardiac Surgery. J Am Geriatr Soc, 2017.
- 29. Tolppanen A.M., et al., Antipsychotic Use and Risk of Hospitalization or Death Due to Pneumonia in Persons With and Those Without Alzheimer Disease. Chest, 2016. 150(6): p. 1233–1241. pmid:27298071
- 30. Shannon P., et al., Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res, 2003. 13(11): p. 2498–504. pmid:14597658
- 31. Warde-Farley D., et al., The GeneMANIA prediction server: biological network integration for gene prioritization and predicting gene function. Nucleic Acids Res, 2010. 38(Web Server issue): p. W214–20. pmid:20576703
- 32. Bindea G., et al., ClueGO: a Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks. Bioinformatics, 2009. 25(8): p. 1091–3. pmid:19237447
- 33. Kutmon M., et al., CyTargetLinker: a cytoscape app to integrate regulatory interactions in network analysis. PLoS One, 2013. 8(12): p. e82160. pmid:24340000
- 34. Bindea G., Galon J., and Mlecnik B., CluePedia Cytoscape plugin: pathway insights using integrated experimental and in silico data. Bioinformatics, 2013. 29(5): p. 661–3. pmid:23325622
- 35. Ishida T., et al., Plasma substance P level in patients with schizophrenia: a cross-sectional study. Psychiatry Clin Neurosci, 2011. 65(5): p. 526–8. pmid:21851462
- 36. Yang S.Y., et al., Antipsychotic drugs, mood stabilizers, and risk of pneumonia in bipolar disorder: a nationwide case-control study. J Clin Psychiatry, 2013. 74(1): p. e79–86. pmid:23419234
- 37. Hung G.C., et al., Antipsychotic reexposure and recurrent pneumonia in schizophrenia: a nested case-control study. J Clin Psychiatry, 2016. 77(1): p. 60–6. pmid:26613551
- 38. Jackson J.W., et al., Mediators of First- Versus Second-generation Antipsychotic-related Mortality in Older Adults. Epidemiology, 2015. 26(5): p. 700–9. pmid:26035686
- 39. Pratt N., et al., Risk of hospitalization for hip fracture and pneumonia associated with antipsychotic prescribing in the elderly: a self-controlled case-series analysis in an Australian health care claims database. Drug Saf, 2011. 34(7): p. 567–75. pmid:21663332
- 40. Nielsen J., Foldager L., and Meyer J.M., Increased use of antibiotics in patients treated with clozapine. Eur Neuropsychopharmacol, 2009. 19(7): p. 483–6. pmid:19356909
- 41. Jackson J.W., et al., Quantifying the role of adverse events in the mortality difference between first and second-generation antipsychotics in older adults: systematic review and meta-synthesis. PLoS One, 2014. 9(8): p. e105376. pmid:25140533
- 42. Steinberg M. and Lyketsos C.G., Atypical antipsychotic use in patients with dementia: managing safety concerns. Am J Psychiatry, 2012. 169(9): p. 900–6. pmid:22952071
- 43. Trifiro G., Antipsychotic drug use and community-acquired pneumonia. Curr Infect Dis Rep, 2011. 13(3): p. 262–8. pmid:21394430
- 44. Nose M., et al., Antipsychotic drug exposure and risk of pneumonia: a systematic review and meta-analysis of observational studies. Pharmacoepidemiol Drug Saf, 2015. 24(8): p. 812–20. pmid:26017021
- 45. Trifiro G., Spina E., and Gambassi G., Use of antipsychotics in elderly patients with dementia: do atypical and conventional agents have a similar safety profile? Pharmacol Res, 2009. 59(1): p. 1–12. pmid:18977443
- 46. van der Maarel-Wierink C.D., et al., Risk factors for aspiration pneumonia in frail older people: a systematic literature review. J Am Med Dir Assoc, 2011. 12(5): p. 344–54. pmid:21450240
- 47. Kim S.J., et al., A case of drug-induced interstitial pneumonia potentially related to quetiapine (seroquel) therapy for behavioral and psychological symptoms. Respir Care, 2014. 59(10): p. e145–8. pmid:24782551
- 48. Schmidinger S. and Hofer A., Pulmonary embolism and aspiration pneumonia after reexposure to clozapine: pulmonary adverse effects of clozapine. J Clin Psychopharmacol, 2014. 34(3): p. 385–7. pmid:24717250
- 49. Rizos E., et al., Risperidone-induced acute eosinophilic pneumonia. In Vivo, 2013. 27(5): p. 651–3. pmid:23988901
- 50. Patel S.S. and Allin M.P., Physical complications in early clozapine treatment: a case report and implications for safe monitoring. Ther Adv Psychopharmacol, 2011. 1(1): p. 25–9. pmid:23983924
- 51. Patel U., et al., Neuroleptic malignant syndrome presenting as pulmonary edema and severe bronchorrhea. J Natl Med Assoc, 2002. 94(4): p. 279–82. pmid:11991344
- 52. Majumder S., et al., A single fatal dose of olanzapine. Neurol India, 2009. 57(4): p. 497.
- 53. Hinkes R., et al., Aspiration pneumonia possibly secondary to clozapine-induced sialorrhea. J Clin Psychopharmacol, 1996. 16(6): p. 462–3. pmid:8959476
- 54. Liam C.K. and Ong S.B., Neuroleptic malignant syndrome with renal and respiratory complications—a case report. Singapore Med J, 1990. 31(2): p. 182–4. pmid:1973548
- 55. Voulgari C., et al., Clozapine-Induced Late Agranulocytosis and Severe Neutropenia Complicated with Streptococcus pneumonia, Venous Thromboembolism, and Allergic Vasculitis in Treatment-Resistant Female Psychosis. Case Rep Med, 2015. 2015: p. 703218. pmid:25755670
- 56. Liang C.S. and Hsieh T.H., Myoclonus as an indicator of infection in patients with schizophrenia treated with clozapine. J Psychiatry Neurosci, 2011. 36(1): p. E1–2. pmid:21167107
- 57. Jabeen S., Polli S.I., and Gerber D.R., Acute respiratory failure with a single dose of quetiapine fumarate. Ann Pharmacother, 2006. 40(3): p. 559–62. pmid:16467251
- 58. Kruk J., Sachdev P., and Singh S., Neuroleptic-induced respiratory dyskinesia. J Neuropsychiatry Clin Neurosci, 1995. 7(2): p. 223–9. pmid:7626967
- 59. Leung J.G., et al., Characterization of Admission Types in Medically Hospitalized Patients Prescribed Clozapine. Psychosomatics, 2016.
- 60. Iovino F., et al., Signalling or binding: the role of the platelet-activating factor receptor in invasive pneumococcal disease. Cell Microbiol, 2013. 15(6): p. 870–81. pmid:23444839
- 61. Arimura A., et al., Involvement of thromboxane A2 in bronchial hyperresponsiveness but not lung inflammation induced by bacterial lipopolysaccharide in guinea pigs. Eur J Pharmacol, 1993. 231(1): p. 13–21. pmid:8444278
- 62. Cyphert J.M., et al., Allergic inflammation induces a persistent mechanistic switch in thromboxane-mediated airway constriction in the mouse. Am J Physiol Lung Cell Mol Physiol, 2012. 302(1): p. L140–51. pmid:21984570
- 63. Ryan P.B., et al., Medication-wide association studies. CPT Pharmacometrics Syst Pharmacol, 2013. 2: p. e76. pmid:24448022
- 64. Lorberbaum T., et al., Systems pharmacology augments drug safety surveillance. Clin Pharmacol Ther, 2015. 97(2): p. 151–8. pmid:25670520