Bacterial Membrane Vesicles: The Missing Link Between Bacterial Infection and Alzheimer Disease

Abstract

Periodontitis is a common chronic inflammatory disease, affecting approximately 19% of the global adult population. A relationship between periodontal disease and Alzheimer disease has long been recognized, and recent evidence has been uncovered to link these 2 diseases mechanistically. Periodontitis is caused by dysbiosis in the subgingival plaque microbiome, with a pronounced shift in the oral microbiota from one consisting primarily of Gram-positive aerobic bacteria to one predominated by Gram-negative anaerobes, such as Porphyromonas gingivalis. A common phenomenon shared by all bacteria is the release of membrane vesicles to facilitate biomolecule delivery across long distances. In particular, the vesicles released by P gingivalis and other oral pathogens have been found to transport bacterial components across the blood-brain barrier, initiating the physiologic changes involved in Alzheimer disease. In this review, we summarize recent data that support the relationship between vesicles secreted by periodontal pathogens to Alzheimer disease pathology.

Alzheimer disease (AD) is a progressive neurodegenerative disease that results in a loss of cognitive function that eventually interferes with daily life and is ultimately fatal. While around 5% of AD cases are described as familial and have a known genetic cause, around 95% are termed sporadic with an unclear cause [1]. The greatest risk factor for sporadic AD is age, with patients presenting with clinical symptoms in mid- to late adulthood typically from the sixth decade onward. Other possible causes of sporadic AD include unknown genetic and modifiable risk factors [2]. The prevalence of AD is increasing at an alarming rate as the population of the world ages. The 2019 Global Burden of Disease survey reported that AD was ranked as the 20th cause of global deaths in 1990 and the 7th in 2019. In Australia, since 2016, AD has been the leading cause of death of women and the second leading cause of death overall. There are an estimated 6.7 million individuals with AD in the United States and 50 million with AD globally. These populations are predicted to grow to 12.7 million and 150 million in the United States and globally, respectively, by 2050 [3, 4].

AD has traditionally been defined by underlying pathologic processes documented by postmortem examination or use of targeted in vivo biomarkers. The characteristic hallmarks of pathologic AD development are progressive brain atrophy due to dramatic neuronal and synaptic losses, abnormal deposition of extracellular amyloid beta (Aβ) plaques, and the presence of intraneuronal neurofibrillary tangles composed of hyperphosphorylated tau protein [5]. While the amyloid cascade hypothesis has been the most supported theory, several studies now suggest that neuroinflammation may be the key event fueling neurodegeneration in AD [6]. Microglia and astrocytes are key regulators of neuroinflammation, which can take on neurotoxic or neuroprotective phenotypes, depending on the progression of the disease [6]. Numerous studies have shown microglia and reactive astrocytes becoming activated and being found in high numbers around the Aβ plaques in AD, suggesting a role of these cells in AD pathogenesis [6]. However, whether the neuroinflammation is a cause or consequence of neurodegeneration is still a subject of debate [6].

The Aβ and tau pathologies in the brain accumulate for many years before the clinical onset of AD, suggesting that when symptoms develop, significant and irreversible neuronal loss would have already occurred [7]. Thus, it is critical to develop interventions that target the early stages of the disease to prevent progression to AD.

AN INFECTIOUS ETIOLOGY FOR AD

When Alois Alzheimer described what would later become known as AD, he noted pathologic similarities to neurosyphilis, suggesting a bacterial cause to the disease [8]. In 2010, it was discovered that Aβ42, the most toxic form of Aβ peptides, had antimicrobial activity against Candida albicans and 7 bacterial species in vitro, suggesting that Aβ42 could be acting as an antimicrobial peptide as part of the brain's innate immune system [9]. Despite several reports over the years of the components of viruses, bacteria, and fungi detected in AD brains postmortem [10] and growing evidence of a causal link between microbial infection and neuroinflammation [11], there has been little acceptance from the wider AD research community that microbes could play a role in triggering AD. In fact, in a Journal of Alzheimer's Disease editorial in 2016, Itzhaki and 33 coauthors noted that the response to a role for microbes in AD “recalls the widespread opposition initially to data showing that viruses cause some types of cancer, and that a bacterium causes stomach ulcers” [12].

Evidence that bacterial infection is involved with AD continues to accumulate, with bacteria such as the respiratory pathogen Chlamydia pneumoniae [13] and the agents of periodontitis, Porphyromonas gingivalis [14] and Treponema denticola [15], detected in AD brains postmortem by techniques that specifically identified DNA and proteins from those bacteria. The list of bacteria potentially involved in AD has been expanded recently by using less biased or targeted methods of identification; 16S rRNA gene sequencing has been used on sections of postmortem brains from those with and without AD [16]. In fact, a recent study presents evidence for the temporal-spatial development of a pathogenic microbiome in the brain [17].

Evidence Linking Periodontal Diseases to AD

Periodontal diseases are a group of diseases that can afflict the gingival tissue, periodontal ligament, and alveolar bone of the oral cavity and range in severity from gingivitis, a mild nondestructive inflammation of the gingival tissues, to severe periodontitis. Periodontitis is a bacterial-induced inflammatory disease that is characterized by the progressive, irreversible destruction of the supporting structures of the teeth, including the periodontal ligament and alveolar bone. If left untreated, periodontitis can lead to tooth loss and impaired quality of life, such as difficulty biting and chewing, aesthetic issues, and psychological problems [18]. The high prevalence of periodontitis contributes to the global burden of chronic noncommunicable diseases. Periodontitis in dentate adults was estimated to affect approximately 62% of the population and severe periodontitis approximately 24% in a recent systematic review and meta-analysis covering 17 countries [19]. The prevalence of periodontitis increases with age, with a recent US study indicating that 30% of 20- to 44-year-olds had periodontitis, increasing to 46% in 45- to 64-year-olds and 60% in 65- to 79-year-olds [20]. In the United States, the prevalence of severe periodontitis—characterized by periodontal probing depths >5 mm, bleeding on probing, clinical attachment loss, and number of missing teeth—among 45- to 64-year-olds and 65- to 79-year-olds is 10.4% and 9%, respectively [20].

Periodontitis is caused by microbial dysbiosis in the subgingival plaque biofilms that are accreted to the surface of the tooth root. This dysbiosis is characterized by the emergence and proliferation of a limited number of bacterial pathobionts that already exist in the oral cavity during health [21]. These pathobionts—including P gingivalis, T denticola, Tannerella forsythia, Fusobacterium nucleatum, Filifacter alocis, Prevotella intermedia, and Aggregatibacter actinomycetemcomitans—come to dominate the bacterial community of subgingival plaque at the base of the periodontal pocket [21] (Figure 1A). Pathobionts increase substantially in the subgingival plaque biofilm prior to tissue breakdown and alone can exceed 25% of the subgingival plaque microbiota [24]. With the temporal and spatial shifts in species composition, there is a large increase in the bacterial biomass from health to disease. Bacterial pathogen-associated molecular patterns are conserved molecular motifs absent in the host, including lipopolysaccharides (LPSs), glycoproteins, lipoproteins, and peptidoglycans, that elicit an inflammatory response from the host. The host response is targeted to these tissue-located molecules and fails to resolve the bacterial insult, resulting in host tissue destruction and the release of bioavailable nutrients/micronutrients that further favors the proliferation of the inflammophilic pathobionts accreted to the tooth root surface. This leads to more extensive destruction of host tissue, including ulceration of the gingival epithelium and breakdown of the junctional epithelium enabling increased tissue penetration by bacterial products and bacteria.

Figure 1.

A, Development of bacterial dysbiosis and tissue damage in periodontitis. Reproduced from Liu et al [22] under the terms of Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/). B, Scanning electron microscopy micrograph showing high numbers of BMVs in an in vitro polymicrobial biofilm composed of Porphyromonas gingivalis (Pg), Treponema denticola (Td), and Tannerella forsythia (Tf) cultured in a flow cell system. Reproduced from Zhu et al [23] under the terms of Creative Commons Attribution 4.0 International (CC BY) License (https://creativecommons.org/licenses/by/4.0/). C, Membrane vesicles bleb from the outer membrane of Gram-negative bacteria, such as P gingivalis. P gingivalis BMVs are enriched in numerous pathogen-associated molecular patterns, such as LPS, peptidoglycan, glycoproteins, lipoproteins, and enzymes such as gingipains. Created with BioRender. BMV, bacterial membrane vesicle; LPS, lipopolysaccharide.

Periodontitis can be a significant source of systemic and localized inflammation and a source of persistent bacteremia. The disease is linked to an increased risk of diabetes, cardiovascular diseases, cancers, preterm birth, rheumatoid arthritis, and importantly AD [14]. In fact, “it is recognized that most bacteria associated with AD originate in the oral microbiome and several correlations have been drawn between poor oral hygiene and dementia” [25]. In addition, it has been demonstrated that a greater number of missing teeth, a key characteristic of severe periodontitis, is related to a higher probability of AD occurrence [26].

Dominy et al [14] investigated the relationship between P gingivalis and AD and observed that specific proteases (gingipains RgpA and Kgp) produced by P gingivalis were found more frequently in brain samples from patients with AD than healthy controls. These gingipain levels were correlated with the amount of total tau in the samples. Quantitative polymerase chain reaction analysis detected P gingivalis DNA in the cerebrospinal fluid in a majority of the patients diagnosed with probable AD. Finally, this group used an oral gavage of P gingivalis in a mouse model to show that the bacteria invaded and colonized the brains of these animals. The expression of gingipains appeared to be involved in this colonization, as knockout strains demonstrated decreased colonization [14].

A relationship between AD and T denticola infection has been proposed, originating from the observation that T denticola can be detected in the brains of patients with AD [15]. T denticola increased Aβ40 and Aβ42 accumulation in the hippocampus of C57BL/6 mice by upregulating the β and γ secretases, the host enzymes required for the processing of the Aβ precursor protein APP [27]. Such T denticola–promoted Aβ accumulation resulted in neuronal apoptosis [28], while oral infection with T denticola promoted other AD-like pathologies, including tau hyperphosphorylation and neuroinflammation [29].

The first evidence linking A actinomycetemcomitans components to AD was a demonstration that purified LPS increased cytokine levels in hippocampal and microglial cells. When exposed to A actinomycetemcomitans LPS, microglial cells expressed higher levels of interleukins 1β, 6, and 17, as well as tumor necrosis factor α (TNF-α) and toll-like receptor 2, consistent with an M1 proinflammatory phenotype [30]. The purified LPS also upregulated the same cytokine and toll-like receptor expression levels in a mixed hippocampal culture [30]. Importantly, exposure of LPS to the hippocampal culture caused an increase in Aβ42 levels as well as severe changes in neuron morphology [30]. Several studies have correlated these types of inflammatory factors with AD risk [31]; however, the A actinomycetemcomitans–mediated changes in inflammatory factors have not yet been specifically linked to development of AD.

Central to this work of bacterial identification has been the detection of bacterial cellular components, with only a couple of early reports of bacteria able to be cultured from infected brains [8, 13]. This general inability to culture from brains has been ascribed to the “demonstrated difficulty in culturing bacteria associated with chronic infections and biofilms” [17]. However, an alternative explanation is that nonviable bacterial membrane vesicles (BMVs) infiltrate the brain rather than live bacterial cells.

WHAT ARE BMVs?

BMVs are small spherical structures released by all bacteria that enable interaction and communication with other cells and compounds in their environment. These vesicles are generally 20 to 400 nm in diameter and are composed of various types of lipids derived from cellular membranes with numerous other biomolecules, such as membrane, periplasmic, and cytoplasmic proteins, as well as DNA, RNA, and low molecular mass organic compounds that confer various biological functions [32].

One of the primary functions of BMVs is to facilitate intercellular communication. For example, BMVs can contain quorum-sensing molecules, enabling bacteria to coordinate their growth and behavior with nearby bacteria. They can also transfer functional genes, such as antibiotic resistance genes, and deliver toxins and other virulence factors to host cells during infection [33].

BMVs are involved in nutrient acquisition [33, 34] and can protect bacterial cells from antibacterials by acting as “decoys” for membrane-active molecules such as phages or antimicrobial peptides [35] or by encapsulating antibiotic-degrading enzymes [36]. Finally, BMVs are abundant in bacterial biofilms and appear to play a structural role in the formation of these assemblies [37] (Figure 1).

ORAL BACTERIAL BMVS AND AD

T denticola

Although little is known about the specific function of T denticola BMVs in AD pathogenesis, there is evidence that these BMVs carry cargo that may play a role. Proteomic analyses of purified T denticola BMVs showed that the major virulence factor of T denticola, the surface-located proteinase dentilisin, and its 2 accessory proteins were significantly enriched in BMVs [38]. Dentilisin has been demonstrated to disrupt tight junctions, and BMVs carrying dentilisin were observed to cross epithelial cell monolayers [39]. It is therefore reasonable to predict that dentilisin-bearing BMVs promote blood-brain barrier (BBB) disruption to promote passage across this barrier.

A actinomycetemcomitans

Significant increases in proinflammatory cytokine expression were observed in response to intact A actinomycetemcomitans BMVs. When BMVs produced by strain 33384 were incubated with activated U937 cells (human macrophage-like cells), TNF-α expression was significantly increased via the nuclear factor–κB and toll-like receptor 8 signaling pathways [40]. In contrast, incubation of BV2 cells (mouse microglia cell line) with A actinomycetemcomitans 33384 BMVs also activated the nuclear factor–κB pathway but led to an increased expression of interleukin 6 rather than TNF-α [41]. Both responses represent proinflammatory processes and were shown to be specifically mediated by the presence of small RNAs within the lumen of the BMVs that were delivered to the host cell cytoplasm [40, 41]. In addition, both studies demonstrated that the BMVs, whether delivered via intracardiac injection or intravenous injection in the tail, were able to cross the BBB and localize in the mouse brain [40, 41]. Specifically, after tail vein injection, the BMVs were concentrated in the meningeal blood vessel after 4 hours and had colocalized with meningeal macrophages after 8 hours. After 48 hours, the BMVs had crossed the BBB and colocalized with microglial cells outside the meningeal blood vessel [41]. Together, these studies indicate that A actinomycetemcomitans BMVs can cross the BBB in mice, where their encapsulated RNA cargo elicits an increase of proinflammatory cytokine levels within the brain, which has been linked to AD pathogenesis.

P gingivalis

The outer membrane surface of P gingivalis is decorated with many virulence factors, including 3 gingipains (RgpA, RgpB, and Kgp); these are preferentially packaged onto BMVs [42]. Gong et al [43] detected P gingivalis BMVs in the hippocampus and cortex of mice treated with fluorescently labeled P gingivalis BMVs by oral gavage. P gingivalis BMVs have also been detected in the hippocampus of mice treated with P gingivalis BMVs by gingival exposure [44]. P gingivalis BMVs significantly decreased occludin gene expression in the hippocampus of mice [43]. In addition, in vitro models have shown the disruption of tight junction protein zonula occludens 1 by P gingivalis BMVs as a possible mechanism by which they can increase the permeability of the BBB and enter the brain [45]. Chronic treatment with P gingivalis BMVs by oral gavage or intraperitoneal injection increases tau phosphorylation and activates microglia and astrocytes in mice, suggesting that they are translocated via the bloodstream to the brain [43, 46]. Translocation through the trigeminal nerve has been demonstrated as a potential pathway of P gingivalis BMVs to the brain in mice [44]. In vitro studies have shown that P gingivalis BMVs stimulate the expression of proinflammatory markers interleukins 6, 8, and 1β and TNF-α [43, 44, 46]. Structural changes in human neuroblastoma cells have been reported in vitro, including cell agglomeration and reduced viability upon P gingivalis BMV exposure [47]. Proinflammatory cytokine stimulation is gingipain dependent, as Kgp/Rgp mutants do not upregulate proinflammatory genes as compared with buffer controls, suggesting that gingipains on the surface of P gingivalis BMVs are responsible for cytokine activation [46]. However, P gingivalis BMV activity is likely not due to gingipains alone, as neuroblastoma cells treated with P gingivalis BMVs and gingipain inhibitors did not significantly reduce cellular degradation when compared with cells treated with P gingivalis BMVs alone [47]. Treatment of P gingivalis BMVs on microglia-like cells activates the NLRP3 inflammasome, suppresses the expression of brain-derived neurotrophic factor, and can induce cytotoxicity [43, 44]. P gingivalis BMVs may also affect cognitive behaviors, as mice treated with P gingivalis BMVs via oral gavage displayed lower entries and duration spent in the novel arm of Y-maze tests as compared with controls [43]. P gingivalis BMVs delivered by gingival exposure decreased mouse activity in Y-maze tests and detected decreased brain-derived neurotrophic factor and N-methyl-D-aspartate receptor expression in the hippocampus, possibly contributing to the decline in memory processes observed [44].

P gingivalis produces significant numbers of BMVs, especially when cultured as pathogenic polymicrobial biofilms [23] (Figure 1). In batch culture, the ratio of P gingivalis BMVs to cells was approximately 2000:1 by stationary growth phase [48]. Several studies have shown that the level of P gingivalis in subgingival plaque is associated with disease severity and that a threshold of around a million P gingivalis cells per site is required for disease progression [48]. Given these numbers, a single periodontal pocket could produce in the order of 10⁹P gingivalis BMVs per day. Therefore, 4 pockets over 10 years, which would be classified as mild/moderate periodontitis, could produce 1.46 × 10¹³ BMVs. Even if a tiny proportion of these BMVs penetrate the gingival tissue and enter the bloodstream, they represent a significant and sustained assault on the host.

Taken together, these findings, as summarized in Table 1, suggest a mechanism by which BMVs produced by periodontal pathogens are able to cross the BBB by weakening the barrier. Once in the brain, the various components of the BMVs, including LPS, enzymes, peptidoglycan, and/or RNA, activate an immune response manifested as activation of proinflammatory cytokines, including TNF-α and interleukins 6 and 8. It is conceivable that chronic infections such as periodontitis would result in a continuous supply of BMVs, resulting in a prolonged inflammatory response in the brain and potentially contributing to the cognitive decline associated with AD.

Table 1.

Effects of BMVs From Periodontal Pathogens in Cell Culture and Animal Models

Bacterium: Model	Result	Reference
Treponema denticola
Cell culture
HEp-2 cells: human epithelial	BMVs with dentilisin crossed epithelial cell monolayers.	[39]
Aggregatibacter actinomycetemcomitans
Cell culture
U937 cells: human macrophage–like	BMVs increased expression of TNF-α via the NF-κB and TLR-8 pathways.	[40, 41]
BV2 cells: mouse microglial brain–like	BMVs increased expression of interleukin 6 via the NF-κB and TLR-8 pathways.	[41]
Animal
Mouse: C57BL, male, 6 wk a	BMVs crossed the BBB and localized in the brain after delivery by intracardiac injection.	[40]
Mouse: CX3CR1-GFP b	BMVs crossed the BBB and localized in the brain after delivery by tail intravenous injection.	[41]
Porphyromonas gingivalis
Cell culture
BV2 cells: mouse microglial brain–like	BMVs activated NLRP3 inflammasome, suppressed BDNF expression, and increased TNF-α and interleukin 1β.	[43, 44]
Macrophages: mouse peritoneal cavity
N2a cells: human neuroblastoma	BMV-conditioned media from BV2 cells increased phosphorylated tau in N2a neurons.	[43]
SH-SY5Y cells: human neuroblastoma	BMV treatment caused cytotoxicity and suppressed BDNF.	[44]
HBMECs: human brain microvascular endothelial cells	BMVs disrupted zonula occludens 1 protein and weakened the BBB.	[45]
HMC3 cells: human microglial	BMVs increased expression of interleukins 6, 8, and 1β and TNF-α.	[46]
SH-SY5Y cells: human neuroblastoma	BMVs mediated structural changes in neuroblastoma cells.	[47]
Animal
Mouse: C57BL/6, male, 61 wk	BMVs delivered via oral gavage affected cognition and decreased occludin gene expression in the hippocampus.	[43]
Mouse: C57BL/6, male, 6 wk	BMVs delivered via oral gavage or gingival exposure crossed the BBB and localized in the brain, diminished memory processes, and decreased BDNF and N-methyl-D-aspartate receptor expression in the hippocampus.	[44]
Mouse: C57BL/6, male, 61 wk	BMVs activated NLRP3 inflammasome, suppressed BDNF expression, and induced cytotoxicity.	[43]
Mouse: C57BL/6, male, 6 wk		[44, 45]
Mouse: C57BL/6, male, 61 wk	BMVs delivered by oral gavage or intraperitoneal injection increased levels of phosphorylated tau protein and activated microglia and astrocyte cells.	[43]
Mouse: BALB/cAJc1, female, 40 wk		[43, 46]

Abbreviations: BBB, blood-brain barrier; BDNF, brain-derived neurotrophic factor; BMV, bacterial membrane vesicle; NF-κB, nuclear factor–κB; TLR-8, toll-like receptor 8; TNF-α, tumor necrosis factor α.

^aWeek denotes age of mice at time of experiment.

^bGender and age not provided.

A FOCAL INFECTION AS A SOURCE OF BMVS

Humans live with trillions of bacteria, all capable of producing BMVs, which may disseminate through the body via the bloodstream. In fact, a recent report showed that BMVs were detected in therapeutic-grade blood products from several healthy donors [49]. This adds credence to the concept of focal infection, where a focal or localized infection can lead to a chronic or acute disease at another site in the body. As BMVs from pathogenic or pathobiont bacteria carry virulence factors and disseminate from the primary site of disease, they could cause disease elsewhere in the body. Bacteria that cause chronic primary infections, which develop slowly and last months or even years, even with treatment, would cause long-term exposure to pathogenic BMVs, which could potentially damage the BBB and cause the brain pathology associated with AD. This has been demonstrated in paradigm-shifting mouse models of AD where bacterial cells have been removed from the infection model altogether, with only purified BMVs delivered intravenously or via oral gavage, resulting in AD-like brain pathology [40, 43].

CONCLUSIONS AND FUTURE DIRECTIONS

The inability to culture viable bacteria from postmortem brains, combined with the proven ability of purified BMVs to cause AD-like pathology in mouse brains, suggests that BMVs released from a focal infection elsewhere in the body might disseminate to the brain and cause AD-like pathology. Thus, BMVs could provide the missing link between active bacterial infection and AD.

However, evidence is just emerging that purified BMVs from periodontal pathogens can cause AD-like pathology [43]; the next step would be to examine the blood of periodontal patients for the presence of BMVs. Methods are being developed to isolate BMVs from the complex blood/plasma sample that contains eukaryotic extracellular vesicles as well as BMVs, which requires the need to remove extracellular DNA prior to lysis of the BMVs for genomic DNA extraction. These samples are of low biomass, which causes additional issues. Most current BMV isolation techniques are based on the growth of large bacterial cultures; new methods are required to enable analysis of the BMVs found in patient blood, where the BMV concentration is orders of magnitude smaller. This will require the development of a robust, specific, and sensitive assay to detect BMVs in the blood and CNS.

Another important experiment would be to investigate the BMV concentrations before and after treatment for periodontitis. While it is likely that treatment for periodontitis would decrease BMV concentration by reducing bacterial load, such a relationship has not yet been studied.

In addition, while many of the studies highlighted in this article demonstrate that periodontal BMVs are able to cross the BBB and activate the expression of proinflammatory cytokines in the brain, there has not yet been much work to link those specific changes to the cognitive decline associated with AD. This type of correlation would strengthen the strong association between BMVs and AD that has already been drawn.

Long-term exposure to pathogenic BMVs may perpetually infiltrate the brain and quite possibly overwhelm the brain's defenses because the brain cannot remove the source of the BMVs, as the site of infection is in a different part of the body. Given that moderate to advanced periodontitis is mainly a disease of those >50 years of age and that the clinical onset of sporadic AD occurs in those >65 years of age, there is an undeniable temporal relationship between the diseases, where the brains of people with periodontitis would be chronically exposed over years to pathogenic BMVs from oral pathobionts such as P gingivalis, which have been shown in mice to cause AD-like pathology. This aligns with the observation that brain pathologies accumulate for many years prior to the clinical onset of AD. The importance of maintaining oral health cannot be overstated in the prevention of onset of AD.

Notes

Acknowledgments. The authors gratefully acknowledge generous funding from the IDSA Foundation.

Author contributions . Concept and planning by C. A. B. and A. C. B. All authors contributed to writing and editing the manuscript.

Supplement sponsorship. This article appears as part of the supplement “Advances in Identifying Microbial Pathogenesis in Alzheimer's Disease,” sponsored by the Infectious Diseases Society of America.

Financial support. This work was supported by the IDSA Foundation; the National Foundation for Medical Research and Innovation (to C. A. B. and S. G. D.); the National Institutes of Health (R21DE032153 and R15GM152918 to A. C. B.); and the Pennsylvania Department of Health (4100095607 to A. C. B.).

Potential conflicts of interest. All authors: No reported conflicts.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.