ABSTRACT
Background Several clinical trials of Staphylococcus aureus (S. aureus)-targeted therapies for atopic dermatitis (AD) have demonstrated conflicting results regarding whether they improve AD severity scores. This study performs a model-based meta-analysis to investigate possible causes of these conflicting results and suggests how to improve the efficacies of S. aureus-targeted therapies.
Methods We developed a mathematical model that describes systems-level AD pathogenesis involving interactions between S. aureus and Coagulase Negative Staphylococcus (CoNS). The model was calibrated to reproduce time course data of S. aureus levels, EASI scores, and EASI-75 in response to dupilumab, S. hominis A9 (ShA9) and flucloxacillin from published clinical trials. We simulated efficacies of hypothetical S. aureus-targeted therapies on virtual patients using the model.
Results Our model simulation reproduced the clinically observed detrimental effects that application of ShA9 and flucloxacillin had on AD severity and showed that these effects disappeared if the bactericidal activity against CoNS was removed. A hypothetical (modelled) eradication of S. aureus by 3.0 log10 CFU/cm2, without killing CoNS, achieved comparable EASI-75 to dupilumab. This efficacy was potentiated if dupilumab was administered in conjunction with S. aureus eradication (EASI-75 at week 16; S. aureus eradication: 66.7%, dupilumab 61.6% and combination: 87.8%). The improved efficacy was also seen for virtual dupilumab poor responders.
Conclusion Our model simulation suggests that killing CoNS worsens AD severity and that S. aureus-specific eradication without killing CoNS could be effective for AD patients, including dupilumab poor responders. This study will contribute to design promising S. aureus-targeted therapy.
1. INTRODUCTION
Atopic dermatitis (AD), also called eczema, is the most common inflammatory skin disease1. The symptoms of AD involve relapsing pruritus and skin pain, impairing patients’ quality of life and work productivity2. The pathogenesis of AD is characterised by skin barrier damage, Th2-dominant inflammation and skin dysbiosis3,4,5. The skin dysbiosis in AD patients is frequently accompanied by Staphylococcus aureus (S. aureus) colonisation and decreased commensal bacteria in the skin6. S. aureus skin colonisation is found in 75%-90% of AD patients without clinical signs of superinfection, whereas it is found in only 0%-25% of healthy subjects7-11.
S. aureus levels on skin lesions correlate with AD severity12,13, and S. aureus has been considered as a promising target for AD treatment as it induces both skin barrier damage and inflammation by producing various virulence factors, such as phenol-soluble modulins (PSMs), staphylococcal enterotoxins and the toxic shock syndrome toxin-114,15.
Some clinical trials of S. aureus-targeted therapies for AD16 have indeed demonstrated a reduction in S. aureus. However, they have shown conflicting efficacies as to whether they improve AD severity scores. For example, in several clinical trials, oral and topical antibiotics whose antibacterial spectrum covers Staphylococci were applied to eradicate S. aureus at least temporarily on AD skin lesions. But these interventions often failed to improve AD severity: A Cochrane review concluded that antibiotics may make no difference or only slight improvement in AD severity17. Oral flucloxacillin, one of the antibiotics, worsened AD severity compared to placebo despite a significant reduction of S. aureus levels on skin lesion18. Currently, the use of antibiotics is recommended for AD only in case of overt infection19.
As another S. aureus-targeted therapy, transplant of S. hominis A9 (ShA9), a commensal strain of coagulase-negative staphylococci (CoNS) isolated from healthy human skin, has been tested20. The clinical study showed that ShA9 transplant decreased the S. aureus levels on skin lesions and improved AD severity scores in the patients (N=21) whose skin was colonised with S. aureus that is sensitive to the bacteriocins secreted by ShA9. However, ShA9 transplant deteriorated AD severity scores in the patients (N=11) whose skin was colonised with S. aureus that is resistant to the bacteriocins secreted by ShA920. ShA9 produces bacteriocins that have bactericidal activity against S. aureus21 and secretes autoinducing peptides (AIPs) that inhibit the accessory gene regulatory (agr) system, which regulates the expression of the virulence factors in S. aureus22.
Some therapeutics that do not target S. aureus directly can also reduce S. aureus levels. Dupilumab, an approved biologic for AD, is a monoclonal antibody that inhibits IL-4 and IL-13 signalling. These Th2 cytokines can facilitate S. aureus colonisation as they damage the skin barrier by inhibiting epidermal differentiation23,24and the skin barrier damage induces an increase in skin pH25 that promotes S. aureus growth26. In addition, inhibition of IL-4 and IL-13 by dupilumab can reduce S. aureus levels since IL-4 and IL-13 inhibit the synthesis of antimicrobial peptides (AMPs) against S. aureus27. Dupilumab has reduced S. aureus levels and improved AD severity scores in a clinical trial12.
Taken together, all of flucloxacillin, ShA9 and dupilumab decreased S. aureus levels but showed conflicting efficacies as to whether these drugs improve AD severity scores. Understanding of underlying mechanism for the conflicting efficacies will help optimising S. aureus-targeted therapies for AD that are consistently effective.
To investigate the causes of the conflicting efficacies of S. aureus-targeted therapies, this study applies a quantitative systems pharmacology (QSP) approach. QSP is a framework to describe systems-level pathogenesis and drug effects by integrating data and knowledge into a mathematical model28. A QSP approach facilitates a model-based meta-analysis that integrates data from different clinical trials, as well as knowledge on pathogenesis and mechanism of action (MoA) of drugs, to inform rational drug development29. A QSP model-based meta-analysis is especially suitable for this study which aims to investigate underlying mechanisms for the conflicting efficacies of S. aureus-targeted therapies observed in different clinical studies.
We have recently applied a QSP model-based meta-analysis of multiple biologics for AD and identified IL-13 and IL-22 as potential drug targets for dupilumab poor responders30. However, the previous QSP model of biologics is not suitable for this study’s aim as it did not describe the mechanism of S. aureus-targeted therapies. This study presents a new QSP model of S. aureus-targeted therapies that describes the interactions between S. aureus and CoNS in AD pathogenesis by referring to clinical efficacy data of the three drugs described above (flucloxacillin, ShA9 and dupilumab) to test the following two hypotheses.
The first hypothesis is that the bactericidal effects of S. aureus-targeted therapies on CoNS impair their efficacies on AD severity. Decrease in CoNS levels causes reduction in their AIP secretion, thereby upregulating agr expression. Upregulated agr expression promotes production of virulence factors in S. aureus that can worsen AD severity. While such a hypothesis has already been implied in several studies20,31,32, there has been no quantitative evaluation on the possible dynamic influences of killing CoNS on clinical efficacies to the best of our knowledge.
The second hypothesis is that S. aureus-targeted therapies are effective for dupilumab poor responders as they have a different MoA from dupilumab. The responder rates for dupilumab were 44%-69%33,34 for Eczema Area and Severity Index (EASI)-75 (75% reduction in the EASI score35,36), leaving a significant proportion of dupilumab poor responders. Therapeutic options for dupilumab poor responders are limited to increasing topical corticosteroids and adding additional systemic immunosuppressive agents. However, the dupilumab poor responders are often resistant to these treatments and require monitoring for adverse effects37, leaving unmet medical needs for dupilumab poor responders. We propose promising S. aureus-targeted therapies for AD patients, especially for dupilumab poor responders, by conducting model simulations on virtual patients.
2. METHODS
Our QSP model explicitly describes causal relationships between drugs, biological factors and an AD severity score using a graphical scheme and ordinary differential equations. The model was developed by 1) selecting drugs and biological factors to be modelled, 2) formulating drug effects and causal relationships between the biological factors and 3) optimising model parameters that define virtual patients. The developed model was used to simulate the clinical efficacies of hypothetical S. aureus-targeted therapies in virtual patients.
2.1. Selecting drugs and biological factors
We considered flucloxacillin, ShA9 and dupilumab because they demonstrated a decrease of S. aureus levels in a placebo-controlled double-blinded clinical study where AD severity scores were reported (Table 1 and Supplementary Information (SI) Section 1).
We selected six biological factors as model variables: colony density levels of S. aureus and CoNS and levels of agr expression, IL-4/IL-13 in the skin and skin barrier integrity and the EASI score. S. aureus and CoNS are the core factors in this study. “CoNS” does not include the ShA9 strain applied in the ShA9 treatment. “Agr expression” corresponds to the main mechanism for S. aureus to expression virulence factors in S. aureus22 that induce skin barrier damage and skin inflammation. The IL-4/IL-13 represents Th2-cytokines that are targeted by dupilumab. “Skin barrier integrity” is a critical factor in AD pathogenesis as in our previous models30,38. The EASI score represents an endpoint for AD severity. Some biological factors such as AMPs were not described as model variables but were considered implicitly as a rationale for the causal relationships (e.g., IL-4 and IL-13 increase S. aureus and CoNS via decreasing AMPs) to make the model simpler yet interpretable.
2.2. Formulating drug effects and causal relationships between biological factors
We developed a mathematical model consisting of six equations corresponding to the six biological factors with 26 parameters to simulate the efficacies of the three drugs (SI Section 3). The effects of flucloxacillin were modelled by increasing the killing rates of both S. aureus and CoNS as its antibacterial spectrum covers all Staphylococcus species. The effects of ShA9 were modelled by increasing the killing rates of S. aureus and CoNS and the inhibitory strength against the agr expression because ShA9 produces bacteriocins against both S. aureus and CoNS21 and AIPs that inhibit the agr expression20. The effects of dupilumab were modelled by decreasing effective concentrations of IL-4/IL-13 in the skin by 99%. The value of 99% was obtained from a calculation using the published data on IC50 and the mean concentration of drugs in the skin39 that was estimated from their concentration in the serum measured in clinical trials (SI section 3.2.3). The causal relationships between biological factors were described according to published experimental evidence based on human data (SI section 3.1). The model was implemented in Python 3.7.6 (Python Software Foundation).
2.3. Modelling virtual patients and optimising model parameters
We assumed that the model parameter values (e.g., the recovery rate of skin barrier via skin turnover, k1) vary between AD patients and that a set of 26 parameter values defines pathophysiological backgrounds of each virtual patient (TABLE S3). Each value of the i-th parameter, ki, is taken from a log-normal distribution40 whose probability function, f(ki), is defined by where μi and σi are the distribution parameters that represent the mean and the standard deviation of In ki, respectively.
We optimised the 52 distribution parameters (μi and σi, i=1, …, 26) that define distributions of the 26 model parameters so that the model reproduces the reference data derived from published clinical studies (SI Section 4). The reference data consist of baseline levels of S. aureus, CoNS, IL-4/IL-13 and the EASI scores (TABLE S2) and time courses of S. aureus levels, the EASI scores and EASI-75 assessed in clinical trials of the selected drugs (FIGURE 1). The S. aureus levels, EASI scores and EASI-75 were normalised to compare the clinical efficacies of different clinical trials (SI Section 2). The agr expression and skin barrier integrity were regarded as latent state variables that have no reference data to be compared with simulated values. Simulated baseline levels were obtained by computing steady-state levels of biological factors (at 1000 weeks without drug treatment). All the simulations were conducted on 1000 virtual patients generated by randomly sampling each parameter value from the distribution in Eq. (1).
2.4. Simulating efficacies of hypothetical S. aureus-targeted therapies
We simulated EASI-75 of hypothetical therapies with different strengths for killing of S. aureus and of CoNS and for inhibiting agr expression to explore optimal S. aureus-targeted therapies. Specifically, we examined the efficacies of hypothetical therapies that achieve a maximal reduction of S. aureus level from placebo (reduction of 3.0 log10 CFU/cm2 achieved in published clinical trials for S. aureus-targeted therapies18, 20, 41, 42, 43, 44, 45), the maximal level of CoNS (no bactericidal effects on CoNS, keeping the baseline level of CoNS), an example level of inhibition of the agr expression (we used 90% as we have no reliable evidence to estimate maximal inhibition rates of agr expression) and their combinations.
We also simulated EASI-75 of hypothetical therapies in virtual dupilumab poor responders, which were defined as the virtual patients who did not achieve the EASI-75 criterion at 16 weeks.
3. RESULTS
3.1. QSP model reproduced clinical efficacies of three drugs
We normalised S. aureus levels, the EASI scores and EASI-75 using the reported results in the clinical trials to compare efficacies of flucloxacillin, ShA9 and dupilumab (FIGURE 1). The efficacies of ShA9 were presented for two groups of patients stratified by the sensitivity of S. aureus to ShA9 bacteriocins, as in the original clinical study20. Hereafter, ShA9 applied to patients colonised with S. aureus that is sensitive to ShA9 bacteriocins is referred as ShA9-sensitive, and those with S. aureus that is resistant to ShA9 bacteriocins is referred as ShA9-resistant.
The normalised efficacies demonstrated that all the drugs decreased S. aureus levels and that ShA9-sensitive and dupilumab improved the EASI scores and EASI-75, whereas ShA9-resistant and flucloxacillin worsened the EASI scores and EASI-75. The results confirmed that the three drugs demonstrated conflicting efficacies regarding whether they improve AD severity scores while they all reduced S. aureus levels.
We revised our previously published QSP model of biolgics30 to include the MoA for the three drugs and the interactions between S. aureus and CoNS (FIGURE 2). The new QSP model of S. aureus-targeted therapies reproduced the baseline levels of the biological factors and the clinical efficacies of the drugs on S. aureus levels, the EASI scores and EASI-75 (FIGURE 3a, b). The root mean square errors of the mean and %CV of S. aureus levels, the EASI scores and EASI-75 between the simulated and reference data were 0.3 log10 CFU/cm2, 43%, 1.5 (out of 72 = the maximal EASI score) and 2.9%, respectively.
3.2. Detrimental effects of flucloxacillin and ShA9 on EASI scores disappeared when their bactericidal activity against CoNS was hypothetically removed
Using the new QSP model, we tested the first hypothesis that the bactericidal effects on CoNS impair the efficacies of S. aureus-targeted therapies on AD severity.
Our model simulation demonstrated that flucloxacillin and ShA9-resistant decreased CoNS while increasing the agr expression (FIGURE 3c), and that flucloxacillin and ShA9 could achieve better EASI scores and EASI-75 than placebo if they had no bactericidal effects on CoNS (FIGURE 4). In addition, a sensitivity analysis of the model parameters for %improved EASI elucidated that lower rates of CoNS killing by flucloxacillin (dfh) and ShA9 treatments (dA9h) result in higher %improved EASI (SI section 5). These results suggested that a decrease in CoNS increases the agr expression, thereby worsening the EASI scores.
While CoNS levels were reduced to similar levels in both the ShA9-sensitive and ShA9-resistant groups, agr expression was reduced only in the ShA9-sensitive group (FIGURE 3c). The agr expression decreased due to the stronger decrease of S. aureus levels by ShA9-sensitive, compared to ShA9-resistant, even though the decrease in CoNS resulted in a slight increase in the agr expression. These results suggest that the efficacies of S. aureus-targeted therapies are determined in some part by the balance of their bactericidal strengths against S. aureus vs. CoNS.
3.3. Hypothetical S. aureus-targeted therapies achieved better EASI-75 than dupilumab
The QSP model described antimicrobial effects of S. aureus-targeted therapies by three parameters: the rate of S. aureus killing, that of CoNS killing and the strength of agr expression inhibition (FIGURE 2). The antimicrobial effects result in a decrease of S. aureus level, that of CoNS level and an inhibition of agr expression level, respectively (FIGURE 5a). To explore which antimicrobial effects are responsible for improvement in AD severity, we conducted model simulations for hypothetical S. aureus-targeted therapies with different values of the three parameters.
Our simulation results demonstrated that lower S. aureus levels, higher CoNS levels and stronger inhibition of agr expression resulted in higher EASI-75 after 16 weeks (FIGURE 5b left). The S. aureus-specific eradication (the maximal reduction of S. aureus level without killing CoNS, yellow arrows in FIGURE 5b) led to comparable EASI-75 to dupilumab (66.7% vs. dupilumab: 61.6%). The EASI-75 of the S. aureus-specific eradication was improved by adding 90% inhibition of the agr expression (70.6%, blue arrows in FIGURE 5b).
Simulations for a combinatorial application of dupilumab and hypothetical S. aureus-targeted therapies elucidated that it can achieve better EASI-75 than an application of either one (FIGURE 5b right). The S. aureus-specific eradication improved EASI-75 (87.8%) when it was combined with dupilumab, which was further improved (91.9%) by adding 90% inhibition of agr expression.
3.4. S. aureus-targeted therapies achieved significant response in virtual dupilumab poor responders
We also simulated EASI-75 of S. aureus-targeted therapies in dupilumab poor responders (FIGURE 5c). Similarly to the results shown above for all virtual patients (FIGURE 5b), lower S. aureus levels, higher CoNS levels and higher inhibition of agr expression showed a better EASI-75 in virtual dupilumab poor responders. The hypothetical S. aureus-targeted therapies achieved a significant EASI-75 in virtual dupilumab poor responders (S. aureus-specific eradication: 43.2% and that with 90% inhibition of agr expression: 61.1%), which were potentiated by simultaneous application of dupilumab (S. aureus-specific eradication: 61.5% and that with 90% inhibition of agr expression: 79.6%).
4. DISCUSSION
4.1. QSP model-based meta-analysis reveals mechanism of conflicting efficacies of S. aureus-targeted therapies
We developed a QSP model that describes the interactions between S. aureus and CoNS in AD pathogenesis (FIGURE 2) by integrating data and knowledge from published experiments using human samples (SI section 3). The model reproduced published data of clinical efficacy for flucloxacillin, ShA9 and dupilumab (FIGURE 1) regarding the EASI scores, EASI-75 and S. aureus levels (FIGURE 3).
The QSP model simulation revealed that S. aureus-targeted therapies can worsen the EASI scores if they kill CoNS. The simulation showed that the application of ShA9 and flucloxacillin had detrimental effects on AD severity, and those effects disappeared if their bactericidal activity against CoNS was hypothetically removed (FIGURE 4). The graphical scheme of the QSP model (FIGURE 2) can explain how a decrease in CoNS impairs the EASI scores. The decreased CoNS levels diminish secreted AIPs, thereby upregulating the agr expression. The upregulated agr expression promotes the production of virulence factors that damage the skin barrier (e.g., by PSMα and enterotoxins) and induce inflammation (e.g., by wall teichoic acid to activate dendritic cells), which can worsen AD severity. These results and interpretation indicate an importance of bactericidal specificity on S. aureus in S. aureus-targeted therapies.
4.2. Model simulation quantifies relationships between profiles of antibacterial effects and responder rates
The QSP model simulation also elucidated relationships between antibacterial effects of S. aureus-targeted therapies (decreases in the S. aureus and CoNS levels and in the inhibition level of agr expression) and their EASI-75 responder rates (FIGURE 5b left). In addition, our simulation suggested that the efficacy of S. aureus-targeted therapies can be potentiated by concomitant use of dupilumab (FIGURE 5b right).
Theoretically, S. aureus-targeted therapies will achieve the best efficacy if they could eradicate S. aureus completely. However, some S. aureus may remain on population average after S. aureus-targeted therapies, presumably due to resistance against antibiotics and bacteriocins46. Hence, it is crucial to inhibit agr expression by keeping the AIPs produced by CoNS, in addition to killing S. aureus, to minimise the agr-dependent virulence effects of S. aureus.
The hypothetical S. aureus-specific eradication (the maximal reduction of S. aureus level without killing CoNS), especially in combination with dupilumab, showed higher responder rates than dupilumab (Simulated EASI-75 at week 16: placebo 26.6%, dupilumab 61.6%, S. aureus-specific eradication 66.7% and combination 87.8%. FIGURE 5b right). Recently, JAK inhibitors have demonstrated promising efficacies in AD patients; abrocitinib showed a comparable response to dupilumab (EASI-75 at week 16; abrocitinib 71.0% vs. dupilumab 65.5%, not significant)47, and upadacitinib showed the highest responder rate among Ph3 trials of JAK inhibitors (EASI-75 at week 16. upadacitinib 77.1% vs. placebo 26.4%)48. Our simulation implies that S. aureus-specific eradication combined with dupilumab may achieve higher responder rates than JAK inhibitors.
4.3. S. aureus-specific eradication is potentially effective for dupilumab poor responders
This study also suggested the effectiveness of S. aureus-targeted therapies for dupilumab poor responders. The simulation for virtual dupilumab poor responders showed that S. aureus-specific eradication achieved 43.2% EASI-75 (FIGURE 5c left), which is much higher than EASI-75 achieved (up to 33.8%) when we simulated inhibition of all the cytokines considered in the previous QSP model of biologics30. These results imply that S. aureus, rather than cytokines, is potentially a promising therapeutic target for dupilumab poor responders.
The model simulation also demonstrated that the efficacy of S. aureus-targeted therapies is potentiated by its concomitant use with dupilumab in dupilumab poor responders (FIGURE 5c right). The results suggest that IL-4/IL-13 signalling contributes to the pathogenesis even for dupilumab poor responders and thus needs to be inhibited. Targeting both S. aureus and IL-4/IL-13 could be a promising therapeutic approach for AD patients.
4.4. Limitation of the QSP model simulation
This study aimed to interpret clinical data of S. aureus-targeted therapies obtained under different study conditions using a model-based meta-analysis. We assumed their efficacies are comparable across clinical trials after normalisation, although the study conditions (e.g., topical and systemic therapies) may influence the reported efficacies. The accuracy of the simulated efficacies of the hypothetical S. aureus-targeted therapies needs to be verified by future clinical trials49.
Our model assumed that CoNS has no detrimental effects on the skin barrier and inflammation. However, recent studies have suggested that S. epidermidis, one of CoNS, also has detrimental effects on the skin barrier13. The detrimental effects of S. epidermidis may explain the worsened EASI scores in ShA9 as it increased the proportion of S. epidermidis among microbiome in the AD skin lesion20. Explicit modelling of different CoNS strains may deliver further insights into the roles of CoNS in AD pathogenesis, although our model assumed the detrimental effects of S. epidermidis are negligible compared to those of S. aureus because S. aureus has a higher correlation with AD severity scores than S. epidermidis6,50.
4.5. Prospect for S. aureus-targeted therapies
The results of this study supported the widely accepted idea for S. aureus being a promising drug target for AD and suggested the importance of considering antibacterial activities against both S. aureus and CoNS when developing S. aureus-targeted therapies. How much S. aureus killing is required to achieve a certain efficacy for a therapy would depend on how strong the therapy kills CoNS and inhibits agr expression.
This study presents an example of how QSP model can contribute to model-informed drug development51 for precision medicine. For example, our simulation results will contribute to the design of S. aureus-targeted therapies because the simulated relationship between EASI-75 responder rates and antibacterial effects (i.e., decreases in the S. aureus and CoNS levels and inhibition of agr expression) can be used as a guide to set a target profile of the antibacterial effects to achieve a desirable efficacy (e.g., better EASI-75 than dupilumab). Our simulation results also encourage combinatorial use of S. aureus-targeted therapies and cytokine-targeted therapies such as biologics and JAK inhibitors for AD.
The code of the QSP model is available at https://github.com/Tanaka-Group/AD_QSP_model.
Data Availability
The code of the QSP model is available at https://github.com/Tanaka-Group/AD_QSP_model.
Acknowledgments
We thank Dr. Elisa Domínguez-Hüttinger for her insightful comments on our manuscript.
Footnotes
Conflict of Interest statement Takuya Miyano reports personal fees from DAIICHI SANKYO CO., LTD., during the conduct of the study, outside the submitted work; Alan D Irvine reports personal fees from Sanofi Regeneron, personal fees from AbbVie, personal fees from Eli Lilly, personal fees from Pfizer, personal fees from UCB Pharma, personal fees from Novartis, personal fees from Dermavant, personal fees from Benevolent AI, personal fees from Menlo Therapeutics, personal fees from Chugai, personal fees from LEO Pharma, personal fees from Arena, during the conduct of the study; Reiko J Tanaka reports grants from British Skin Foundation, during the conduct of the study.
Funding information This work was funded by the British Skin Foundation (005/R/18).