Abstract
Background
Differentiation of a relapse from a new infection is challenging in patients with a recurrent bone or joint infection (BJI). We compared clinical, microbiological, and genomic definitions of relapse among patients with methicillin-resistant Staphylococcus aureus (MRSA) BJI.
Methods
All MRSA isolates obtained from BJIs between July 2018 and December 2022 from patients with at least 2 episodes of BJI from 2 U.S. hospitals underwent whole-genome sequencing. Distinct intrasubject lineages (ISLs) were defined as MRSA genomes from the same individual differing by <100 single nucleotide polymorphisms. Clinical, microbiological, and population genomic criteria were each separately compared with a gold standard for relapse versus new infection of genomically defined ISLs. The level of agreement was calculated with Cohen's kappa. A multivariable analysis was performed to define factors associated with the occurrence of recurrent episodes with different ISLs.
Results
We included 264 isolates from 80 subjects with a range of 2–5 episodes spanning 16–1403 days. In total, 29 subjects (36%) had >1 ISL. Multilocus sequence type (ST) 8 was the most common (n = 147, 55.7%). In multivariable analysis, female sex and antibiotic susceptibility differences in MRSA strains were associated with >1 ISL. Compared with the genomic definition of relapse (same ISL), the level of agreement was poor (Cohen's kappa = −0.16) for the clinical definition, fair for the microbiological definition (Cohen's kappa = 0.29), and substantial for the MLST definition (Cohen's kappa = 0.63).
Conclusions
Clinical and microbiological criteria were not accurate in distinguishing a relapse of BJI from a new infection.
Keywords: Staphylococcus aureus, bone and joint infection, osteomyelitis, MRSA, intrasubject lineage
Differentiation of a relapse from a new infection is challenging in patients with a recurrent bone or joint infection. Compared to a genetic definition, clinical and microbiological criteria did not accurately distinguish a relapse from a new infection.
Staphylococcus aureus is the most common pathogen responsible for bone and joint infections (BJI) [1], causing an estimated 10–55% of cases, depending on the infection type (eg, foreign body, prosthetic-joint associated, hematogenous) and anatomic site [2]. Staphylococcus aureus possesses numerous virulence factors that enable it to invade host tissue, evade host immunity, and form biofilms, which often lead to recalcitrant infections [3–5]. The outcome of S. aureus BJI is dependent on host factors, medical and surgical management, and bacterial phenotypes. Biofilms are particularly difficult to eradicate, and relapsed infections are common [6]. Strains that demonstrate antibiotic resistance, such as methicillin-resistant S. aureus (MRSA), can increase the likelihood of poor outcome, perhaps due to the more complicated choice of appropriate of antibiotic therapy. The genetic evolution of BJI isolates during relapse infections is poorly understood [7]. Furthermore, it is challenging to clinically differentiate a relapse from a new infection in a patient with distinct episodes of S. aureus BJI.
Here, we aimed to characterize the genomic epidemiology of MRSA strains among patients with at least 2 BJI episodes. First, we applied phylogenetic analysis of MRSA isolates to identify distinct within-host lineages, termed intrasubject lineages (ISLs). Second, we assessed whether isolates from separate episodes of BJI in the same patient belonged to the same ISL, and finally, whether clinical, microbiological and MLST criteria could distinguish a relapse (same ISL) from a new infection (new ISL). Relapses are costly and often imply failed therapy, and distinguishing these cases from a new BJI may inform optimal approaches to therapy in the future.
METHODS
Population Studied
All MRSA isolates obtained by the Hospital of the University of Pennsylvania (HUP) Clinical Microbiology Laboratory (CML) from July 2018 to December 2022 from patients at 2 U.S. academic hospitals (Penn Presbyterian Medical Center and HUP) were prospectively collected and stored at −80°C. Each isolate was reviewed in the electronic medical record (EMR). Clinical and microbiological criteria for inclusion in the present study are shown in Supplementary Data.
Clinical, Microbiological and MLST Definitions for Relapse and New Infections
Clinical definitions for a BJI episode, index infection episode, new infection and relapse were formulated by 2 infectious diseases physicians with expertise in the management of BJIs (KB and MZD). They were developed to align closely with clinical practice (Table 1).
Table 1.
Clinical Definitions Used in the Study
| Term | Definition |
|---|---|
| Bone and joint infection (BJI) episode | A discrete medical encounter that requires work-up for suspected infection. Any episode may include >1 methicillin resistant S. aureus (MRSA) isolate. |
| Index infection episode | The first BJI that the patient had during the study period. This includes at least all MRSA cultures obtained within 14 d of the infection. |
| Subsequent infection episode | Any MRSA BJI episode following the index infection episode. Must be classified as either a new infection episode or a relapse episode. |
| New infection episode | A new MRSA culture from (or, if only positive blood culture, related to) a BJI we presumed to be a new infection (ie, bacteria newly infecting a bone or joint)
|
| Relapse episode | A new MRSA culture from a BJI thought to be a recrudescent growth of MRSA from the previous infection episode in the same anatomic site OR Any new episode that occurs after 14 d of antibiotics for the index episode and that does not meet the criteria for a new infection episode |
Abbreviations: MRSA, methicillin-resistant Staphylococcus aureus; S. aureus, Staphylococcus aureus.
The microbiological-based definitions were the following: (1) Relapse if the MRSA BJI episode occurred with a MRSA strain having the same antibiotic susceptibility pattern as the previous infection episode and (2) New infection if the BJI episode occurred with a MRSA strain differing in susceptibility to at least 1 antibiotic compared with the prior infection episode.
The MLST-based definitions were the following: (1) Relapse if the BJI episode occurred with a MRSA strain belonging to the same ST as the previous infection episode and (2) New infection if a BJI episode occurred with a MRSA strain belonging to a different ST than the strain causing the prior infection episode.
Data Collection
Data were collected retrospectively from the EMR. Bone or joint infections were categorized as community-associated (CA), healthcare-associated community-onset (HACO), or healthcare-associated (HA) infection as previously defined [8].
Microbiology Testing
All S. aureus isolates were identified from clinical cultures by the HUP CML. Details of microbiological analysis methods are given in the Supplementary Data.
Molecular Typing
For each strain, genomic DNA extraction was performed using a commercial kit (Qiagen). Library preparation and whole genome sequencing (WGS) were performed by the Penn/Children's Hospital of Philadelphia (CHOP) Microbiome Center using Illumina MiSeq or Hiseq platforms to produce paired-end 150 bp reads. The complete method of genetic analysis is described in Supplementary Data. ISLs were defined as a genome or group of genomes that differed by no more than 100 single nucleotide polymorphisms (SNPs), based on findings from our recent study of S. aureus intrahost diversity [3].
Statistical Methods
Data were reported as medians (interquartile range [IQR]) for continuous variables and counts (percentages) for categorical variables. Comparisons of continuous data were performed using Student's t-test or the Mann–Whitney U test, and categorical data were compared using Pearson's chi-squared test or Fischer's exact test, as appropriate. Factors independently associated with the occurrence of at least 2 ISLs were determined from a multivariable logistic regression model built using a backward elimination process. All independent variables with a P value < .10 in univariable analysis were included in the multivariable model. Cohen's kappa, sensitivity, and specificity were calculated comparing the different definitions of relapse (ie, clinical or microbiological or MLST) with the genomic definition (ie, same ISL) as the gold standard. The level of agreement for Cohen's kappa was categorized as poor (<0.20), fair (0.21–0.40), moderate (0.41–0.60), substantial (0.61–0.80), or almost perfect (>0.80) [9, 10]. Statistical analyses were performed using SPSS 28.0 (IBM, Armonk, NY, USA).
The study was approved by the University of Pennsylvania Institutional Review Board.
RESULTS
Subject Characteristics
From July 2018 to December 2022, 4088 MRSA isolates from 2415 patients were collected from the HUP CML, including 729 isolates from BJI. We included 264 isolates from 80 patients with a range of 2–5 episodes spanning 16–1403 days (median, 181 days, IQR [37–313]; Figure 1). The most common type of BJI included was diabetic foot infection (n = 36, 45%; Table 2).
Figure 1.
Flow chart for cohort derivation. Abbreviations: BJI, bone or joint and infection; ENT, ear, nose, throat; ISL, intrasubject lineage; MRSA, methicillin-resistant Staphylococcus aureus. *Only 1 isolate per day from the same anatomic site (blood, bone, or deep soft tissue) was included. One episode = a discrete medical encounter that required work-up for suspected infection. Any episode may include >1 MRSA isolate (see text).
Table 2.
Characteristics of Subjects With Methicillin-Resistant S. Aureus (MRSA) Bone and Joint Infections (BJIs), Comparing Those With 1 and More Than 1 Intrasubject Lineage (ISL)
| Characteristic | Total, N = 80 (%) | Single ISL N = 51 (%) | More Than 1 ISL N = 29 (%) | Univariate | Multivariable | |
|---|---|---|---|---|---|---|
| P value | OR (95% CI) | P value | ||||
| Age, y, med (Q25–75) | 56.5 [49–65] | 57 [49–66] | 55 [45–63] | .51 | … | |
| Female sex | 26 (32) | 9 (18) | 17 (59) | <.001 | 12.5 (3.1–50.7) | <.001 |
| Race Black | 27 (34) | 18 (35) | 9 (31) | .7 | … | |
| Race White | 44 (55) | 27 (53) | 17 (59) | .62 | … | |
| Race other | 8 (10) | 6 (12) | 2 (6.9) | .7 | … | |
| Race unknown | 1 (1.2) | 0 (0) | 1 (3.4) | .36 | … | |
| Charlson comorbidity index, med (Q25–75) | 4 [2–7] | 4 (2–7) | 4 (2–5) | .32 | … | |
| History of diabetes | 49 (61) | 30 (59) | 19 (66) | .55 | … | |
| History of neuropathy | 37 (46) | 22 (43) | 15 (52) | .46 | … | |
| IVDU | 11 (14) | 6 (12) | 5 (17) | .52 | … | |
| History of surgery in the last 12 m | 49 (61) | 29 (57) | 20 (69) | .29 | … | |
| HACO or HA | 63 (79) | 39 (76) | 24 (83) | .51 | … | |
| Number of isolates, med (Q25–75) | 3 (2–4) | 3 (2–4) | 3 (2–4) | .32 | … | |
| Source of culture sample | ||||||
| Blood | 32 (40) | 19 (37) | 13 (45) | .51 | … | |
| Bone | 50 (62) | 33 (65) | 17 (59) | .59 | … | |
| fluid | 8 (10) | 4 (7.8) | 4 (14) | .45 | … | |
| Soft tissue | 62 (78) | 37 (73) | 25 (86) | .16 | … | |
| Different culture sources | 63 (79) | 37 (73) | 26 (90) | .072 | NT | |
| Type of BJI | ||||||
| Native arthritis | 6 (7.5) | 3 (5.9) | 3 (10) | .66 | … | |
| Diabetic foot infection | 36 (45) | 21 (41) | 15 (52) | .36 | … | |
| Orthopedic device, total | 22 (28) | 15 (29) | 7 (24) | .61 | … | |
| Prosthetic joint infection | 10 (12) | 6 (12) | 4 (14) | 1 | … | |
| Foreign device vertebral osteomyelitis | 4 (5) | 2 (3.9) | 2 (6.9) | .62 | … | |
| Other foreign device osteomyelitis | 9 (11) | 8 (16) | 1 (3.4) | .14 | … | |
| Osteomyelitis | 14 (18) | 10 (20) | 4 (14) | .51 | … | |
| Vertebral osteomyelitis | 4 (5) | 2 (3.9) | 2 (6.9) | .62 | … | |
| Decubitus ulcer | 10 (12) | 7 (14) | 3 (10) | .74 | … | |
| More than 1 type of BJI | 10 (12.5) | 6 (12) | 4 (14) | 1 | … | |
| More than 1 anatomic site of BJI | 21 (26) | 11 (22) | 10 (34) | .21 | … | |
| Different susceptibility testing between 2 isolates | 38 (48) | 17 (33.3) | 21 (72.4) | <.001 | 9.4 (2.4–36.2) | .001 |
| Polymicrobial infection | 50 (63) | 27 (53) | 23 (79) | .019 | 2.9 (.9–9.8) | .08 |
| Chronic wounda | 45 (56) | 25 (49) | 20 (69) | .084 | NT | |
| Number of episodes, med (Q25–75) | 2 (2–3) | 2 (2–3) | 2 (2–3) | .14 | … | |
| Time between index and last isolate, d, med (Q25–75) | 192 (77–359) | 147 (71–362) | 287 (82–357) | .26 | … | |
| Time between 2 episodes, more than 6 m | 35 (44) | 20 (39) | 15 (52) | .28 | … | |
| Clonal complex (CC) of index isolates | ||||||
| CC8 | 45 (56) | 31 (61) | 14 (48) | .28 | … | |
| CC5 | 31 (39) | 18 (35) | 13 (45) | .4 | … | |
| Other CCb | 4 (5) | 2 (4) | 2 (7) | .47 | … | |
Abbreviations: BJI, bone and joint infection; CC, clonal complex; CI, confidence interval; HA, healthcare-associated; HACO, healthcare-associated community-onset; ISL, intrasubject lineage; IVDU, intravenous drug use; NT, not included in the final model; OR, odds ratio; S. aureus, Staphylococcus aureus.
aPresence of wound >3 m before culture of a MRSA isolate from a BJI.
bIncludes isolates with no identified CC (n = 3), CC30 (n = 1).
Population Structure of MRSA in Patients With More Than 1 Episode
ST8 was the most common ST (n = 147, 55.7%) followed by ST5 and ST105 (Figure 2A). Four novel STs were identified and assigned: ST9083, ST9084, ST9085 and ST9087. The ST9083 strain belonged to CC5, while the ST9084, ST9085 and ST9087 strains belonged to CC8. The population structure illustrating links among strains isolated from the same participant is shown in Figure 2B, which we used to visualize the phylogenetic relationship among co-carried strains.
Figure 2.
A, MRSA isolate lineages causing bone and joint infections (BJIs) in 80 subjects. Number of isolates belonging to each clonal complex (CC), and to each multilocus sequence type (MLST) within each CC is shown. B, Circos plot of maximum likelihood (ML) phylogeny and the relationship between co-carried MRSA strains. A ML phylogeny was generated from a core-SNP alignment of all 264 isolates. Each tip in the phylogeny corresponds to an isolate colored according to the CC (legend shown on the top right). Inside the tree, links are drawn between isolates that belong to the same patient who shared multiple intrasubject lineages (ISLs); each patient is represented by a unique color. Abbreviations: MRSA, methicillin-resistant Staphylococcus aureus; SNP, single nucleotide polymorphism; ST, sequence type.
Identifications of ISLs
WGS of 264 isolates resolved 115 ISLs, with 29 subjects (36%) having >1 ISL (ie, differing by >100 SNPs) (Figure 1). Among subjects with >1 ISL, 17 had strains from different CCs, and 21 had different STs. The distribution of SNP differences in each ISL is shown in Figure 3.
Figure 3.
Distribution of pairwise single nucleotide polymorphism (SNP) distances in each intrasubject lineage (ISL) of methicillin-resistant S. aureus from bone and joint infections. Abbreviation: S. aureus, Staphylococcus aureus.
Different ISLs in Subjects Across Episodes
The median time between 2 BJI episodes was similar between subjects with only 1 ISL and those with different ISLs (103 [52.5–214] vs 151 [75–325] days, P = .13). Neither the presence of a chronic wound for >3 months before culture of an isolate, nor the presence of any wound during BJI (ie, an open wound infection), nor the persistence of a wound in the interval between culture of 2 isolates were different comparing subjects with a single ISL and those with more than 1 ISL (49% vs 69%, P = .084; 75% vs 83%, P = .4% and 65% vs 72%, P = .48, respectively). In multivariable analysis, female sex and differences in antibiotic susceptibility among strains were associated with >1 ISL across episodes (Table 2). When comparing patients by sex, no significant differences were found in tested clinical and demographic variables (Supplementary Table 1). In a sensitivity analysis limited to subjects with a single site of infection across different episodes, the result of the multivariable analysis was similar (Supplementary Table 2). The time elapsed between the first isolate and last episode according to the type of BJI is shown in Figure 4.
Figure 4.
Time elapsed between the first methicillin-resistant S. aureus (MRSA) isolate and last recorded episode of MRSA bone and joint infection (BJI) according to the type of BJI (“x” represents the mean), showing the range for all subjects with that infection type. Abbreviations: BJI, bone and joint infection; DFI, diabetic foot infection; ISL, intrasubject lineage; PJI, prosthetic joint infection; S. aureus, Staphylococcus aureus.
Comparison of Clinical, Microbiological, and MLST Definitions and the ISL (Genomic) Definition
Using a clinical definition, 115 isolates were assigned to index infection episodes (including 80 to first, or index isolates), 93 to relapse, and 56 to new infections. The median time between index isolate culture and the first isolate of relapse was 75.5 (IQR, 41.5–143) days, and the median time between index isolate and new infection was 298 (IQR, 198–458) days. Compared to the genomic definition of relapse, the level of agreement of the clinical definition was poor (Cohen's kappa = −0.16), fair for the microbiological definition (Cohen's kappa = 0.29), and substantial for the MLST definition (Cohen's kappa = 0.63; Table 3 and Supplementary Tables 3–5).
Table 3.
Diagnostic Accuracy and Agreement of Different Definitions for Bone and Joint Infection (BJI) Relapse Compared With the Gold Standard Genome Definition (Intrasubject Lineages [ISLs] by Whole Genome Sequencing)
| Parameters | Sensitivity % (95% CI) | Specificity % (95% CI) | NPV % (95% CI) | PPV % (95% CI) | Cohen's Kappa (CI) | P value |
|---|---|---|---|---|---|---|
| Clinical definition | 62 (54–70) | 36 (28–44) | 23 (14–31) | 75 (67–83) | −0.16 (.08) | .83 |
| Microbiological definition | 80 (72–88) | 47 (39–55) | 43 (35–51) | 83 (78–89) | 0.29 (.09) | <.01 |
| MLST definition | 92 (87–96) | 58 (50–66) | 70 (63–77) | 87 (82–92) | 0.63 (.08) | <.01 |
Abbreviations: CI, confidence interval; MLST, multilocus sequence type; NPV, negative predictive value; PPV, positive predictive value.
Antibiotic Susceptibility
Vancomycin, daptomycin, and ceftaroline resistance were detected in 0, 6, and 1 isolates, respectively (Table 4). Of the 38 subjects with isolates having different antibiotic susceptibility patterns, 20 (52%) had only 1 antibiotic susceptibility difference. Clindamycin had the greatest number of discordant isolates, in 19 subjects (50%), 7 of whom had only 1 antibiotic susceptibility difference. All isolates with daptomycin resistance were reported in a subsequent episode of BJI, and all patients were exposed to daptomycin before or at the same time as culture of the isolates. Of the 23 isolates with differences in susceptibility testing but the same ISL as the prior isolate, 15 (65%) had a susceptibility difference for only 1 antibiotic. Antibiotics with changes in susceptibility were clindamycin (n = 9), erythromycin (n = 5), fluoroquinolones (n = 4), tetracycline (n = 3), daptomycin (n = 3), and trimethoprim/sulfamethoxazole (TMP/SMX) (n = 1).
Table 4.
Phenotypic Antimicrobial Susceptibilities of 264 Methicillin-Resistant S. Aureus (MRSA) Isolates From 80 Subjects With Bone and Joint Infections, Indicating Number Tested for Each Drug
| Antibiotic | Number of Isolates Tested, n (%) | Susceptible, n (%) |
|---|---|---|
| Oxacillin | 264 (100) | 0 (0) |
| Gentamicin | 262 (99) | 260 (99) |
| Ciprofloxacin | 262 (99) | 40 (15) |
| Levofloxacin | 246 (93) | 43 (18) |
| Moxifloxacin | 262 (99) | 50 (19) |
| Erythromycin | 264 (100) | 30 (11) |
| Clindamycin | 264 (100) | 140 (53) |
| Quinupristin/dalfopristin | 263 (99) | 262 (99) |
| Linezolid | 47 (18) | 47 (100) |
| Vancomycin | 264 (100) | 264 (100) |
| Tetracycline | 264 (100) | 206 (78) |
| Tigecycline | 243 (92) | 243 (100) |
| Rifampicin | 263 (99) | 258 (98) |
| Trimethoprim/sulfamethoxazole | 264 (100) | 234 (89) |
| Daptomycin | 71 (27) | 65 (92) |
| Ceftaroline | 9 (3) | 8 (99) |
Abbreviation: S. aureus, Staphylococcus aureus.
When comparing phenotypic antibiotic data with genetic determinants, we observed an overall concordance exceeding 91.5% for 11 out of 16 antibiotics. Ciprofloxacin, levofloxacin, moxifloxacin, erythromycin, and ceftaroline showed lower concordance, ranging from 11.1% (ceftaroline) to 52.7% (erythromycin). Overall, the average concordance between phenotypic resistance and genotype-based predictions was 74.5%.
DISCUSSION
For the first time, in a cohort of subjects experiencing serial MRSA BJI episodes, we characterized the genetic diversity of MRSA. More than one-third of patients had distinct MRSA ISLs across episodes. The likelihood of observing a new ISL in different BJI episodes was 12 times greater when study subjects were female and nine times greater when different antibiotic susceptibility patterns was reported. Persistent ISLs were reported across a median period of 21 weeks. The negative predictive value for a relapse (ie, a single ISL in 2 episodes) was poor using either clinical or microbiological criteria. Our findings underscore the limited reliability of conventional clinical definitions in accurately identifying relapse versus new infection in MRSA BJIs. Misclassification can lead to inappropriate clinical management, such as unnecessary prolonged antibiotics or failure to identify reinfection events. Genomic methods—particularly SNP-based ISL classification—offer improved precision and reproducibility. These methods should be considered in future BJI studies and may help standardize case definitions [11].
Only 1 longitudinal study, by Lavigne et al, has evaluated the capacity of S. aureus to persist in diabetic foot infections (DFI) [12]. Among 48 patients with persistent DFI at the same infection site without healing at 4 weeks, only 12 (25%) had the same S. aureus clone, as determined by microarray-based characterization, during a period ≥ 4 weeks. Also, only 1 patient had a persistence of the same clone for an extended period of 30 weeks. In our study, the increased resolution provided by WGS identified that 64% of patients with BJI had the same S. aureus ISL across episodes, with a median follow-up of 147 days (21 weeks). More specifically, 21/36 (58%) patients with DFI had only 1 ISL with a median follow-up of 147 days (89–271). The differences from our study results may be explained by the fact that we included only patients with osteomyelitis-associated DFI, whereas Lavigne et al included all patients with Grades 3 and 4 of the Diabetic Foot Ulcer Classification System (PEDIS) of the International Working Group of the Diabetic Foot (IWGDF) [12]. We can hypothesize than when S. aureus infected bone, rather than only the soft tissue, the second episode is more often a relapse than a new infection with a different strain. Moreover, Lavigne et al included patients with both MSSA and MRSA DFI, while we included only patients with MRSA. MRSA and MSSA have different virulence profiles, distinct epidemiology, and varied outcomes. Therefore, the capacity of MRSA and MSSA to colonize, infect, and persist in bone may differ [13]. Overall, while the clinical consequences of a relapse and a new infection are similar, the management may be different. In the case of relapse, 2 hypotheses must be considered. The first is the absence or inadequacy of surgical therapy (persistence of a sequestrum, and/or a foreign device, for example). The second is the ineffectiveness of antibiotic therapy, which may be poor (eg, from impaired diffusion, non-adherence, or emergent resistance), necessitating a change of antibiotic. In the event of a new infection, the persistence of the portal of entry or the persistence of another risk factor for BJI should be considered.
Choosing clinical criteria to define relapse infection episodes and new infections is challenging. Among patients with recurrent S. aureus bloodstream infection, for example, Choi et al defined a relapse in patients with indistinguishable PFGE patterns in 2 episodes and time elapsed between 2 episodes <150 days [14]. In case of BJIs, especially in BJI with a foreign body, it may be more difficult to define a time interval distinguishing a relapse from a new infection. Indeed, in BJI studies, treatment failure is often defined as a function of time after completion of antibiotic therapy—from 2 months in patients with DFI or native arthritis, 1 year for osteomyelitis, to 2 years in patients with PJI [15–18]. In patients with persistent clinical signs of infection (chronic wound, chronic pain), the definitions of relapse or new infection are even more complex and depend on the type of BJI. For example, in patients with PJI, persistence of pain is difficult to interpret, as a patient who has undergone multiple surgeries may have painful sequelae independent of infection. The clinical definitions that we chose were, we believe, simple, reproducible, and reflect assumptions used in typical clinical practice.
A similar antibiotic susceptibility profile was more accurate than the clinical definition in identifying a genomically defined relapse in our study. However, it was far from perfect. The use of MLST profiles had a higher concordance with the genomic ISL criterion to define relapse than clinical and microbiological definitions. However, since bacterial typing is now frequently performed in silico using WGS data, MLST may be used to create a first order approximation of ISLs. This may be particularly useful in settings with limited bioinformatics capabilities. It could therefore be considered in cases where high resolution phylogenetic analysis is not accessible; however, as WGS data generation and analysis becomes more readily available in the clinical setting, these approaches may become routine tools for clinical management.
The association between female sex and the onset of new infection may be a spurious result. Indeed, no clear hypothesis can explain these results, and the study was initially conducted on an exploratory basis. However, this is the first large-scale study to report relapses and new infections in BJI using a genetic definition, so we cannot compare our data with other studies. Further studies are needed to confirm or refute these data.
The greatest level of resolution for MRSA intrahost population structure was achieved using WGS data. The value used for SNP cut-offs to define a S. aureus ISL is inconsistent in the literature and is still a subject of debate. Cut-offs used generally range from 15 to 100 SNPs. A liberal 100-SNP cut-off seems reasonable in our study of BJIs given the distribution of SNPs in our cohort, and it has been used in other studies [14, 19–21]; however, this definition may still underestimate the extent of genetic and phenotypic adaptation occurring over time within a host. For example, it is possible that a genomically defined relapse episode by our criteria could actually be a new infection caused by a strain that causes persistent asymptomatic colonization between infection episodes at another body site (eg, the nares); we were unable to assess this in the current study. Future studies could evaluate this by sequencing several isolates from multiple body sites involved in carriage and disease. These results could then resolve whether a smaller SNP cutoff may more accurately reflect a truly “persistent” population at a site of infection. Last, our study did not include a detailed investigation of intrahost MRSA evolution. In prior studies of chronic MRSA carriage in cystic fibrosis and recurrent SSTIs, distinct mutations and gene content variation were associated with treatment failure and immune evasion [11, 22]. Future work will evaluate the functional impact of these changes using targeted analyses of non-synonymous substitutions and mobile genetic element acquisition. Our findings here provide a foundation for these future studies aimed at investigating the mechanisms of within-host persistence and adaptation.
We observed moderate concordance between phenotypic and genotypic resistance (74.5%), particularly for erythromycin and fluoroquinolones, where discordance was common [23]. These discrepancies may reflect non-expressed resistance genes, regulatory mutations, or unrecognized mechanisms not captured in current genotypic prediction models.
This study represents an important step toward integrating pathogen genomics into the clinical management of BJIs. Our elucidation of the complexity of BJIs highlights the need for more in depth and focused cohort studies. Further, we show that high-resolution pathogen genomic analysis is required to distinguish relapse from new infections. As we continue to accumulate data on the genomic and phenotypic features of persistent versus resolving strains, the potential emerges for predictive tools that can inform risk stratification and therapeutic choices. In particular, phenotypic assays—such as biofilm formation, osteoblast invasion, oxidative stress resistance, and growth kinetics—will complement genomic data, providing a more complete understanding of MRSA persistence and recurrence in BJIs [24, 25]. Each relapse of BJI importantly indicates a failure of medical and/or surgical therapy, resulting in great suffering and financial cost to each patient, indicating that alternative therapies may have been more effective. Ultimately, models that incorporate host comorbidities, pathogen traits, and clinical presentation may guide personalized treatment strategies and reduce recurrence rates.
Supplementary Material
Notes
Author Contributions. K. B.: conceptualization, methodology, formal analysis, investigation, writing—original draft, visualization. E. C.: conceptualization, methodology, formal analysis, writing—review and editing, visualization. N. F. J.: investigation, writing—review and editing. M. H.: investigation, writing—review and editing. T. A.: conceptualization, methodology, writing—review and editing. M. Z. D.: conceptualization, methodology, resources, writing—review and editing, supervision.
Acknowledgments. None.
Data Availability. Data available on request from the authors.
Financial support. None.
Contributor Information
Kevin Bouiller, Department of Infectious and Tropical Diseases, Université Marie et Louis Pasteur, CHU Besançon, UMR-CNRS 6249 Chrono-Environnement, Besançon, France.
Eleonora Cella, Burnett School of Biomedical Sciences, University of Central Florida, Orlando, Florida, USA.
Natasia F Jacko, Division of Infectious Diseases, Department of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.
Maeve E Hiehle, Division of Infectious Diseases, Department of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.
Taj Azarian, Burnett School of Biomedical Sciences, University of Central Florida, Orlando, Florida, USA.
Michael Z David, Division of Infectious Diseases, Department of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.
Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
References
- 1. Urish KL, Cassat JE. Staphylococcus aureus osteomyelitis: bone, bugs, and surgery. Infect Immun 2020; 88:e00932-19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Masters EA, Ricciardi BF, Bentley KLM, Moriarty TF, Schwarz EM, Muthukrishnan G. Skeletal infections: microbial pathogenesis, immunity and clinical management. Nat Rev Microbiol 2022; 20:385–400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Pimentel de Araujo F, Monaco M, Del Grosso M, Pirolo M, Visca P, Pantosti A. Staphylococcus aureus clones causing osteomyelitis: a literature review (2000–2020). J Glob Antimicrob Resist 2021; 26:29–36. [DOI] [PubMed] [Google Scholar]
- 4. Pimentel de Araujo F, Pirolo M, Monaco M, et al. Virulence determinants in Staphylococcus aureus clones causing osteomyelitis in Italy. Front Microbiol 2022; 13:846167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Smeltzer MS, Gillaspy AF. Molecular pathogenesis of staphylcoccal osteomyelitis. Poult Sci 2000; 79:1042–9. [DOI] [PubMed] [Google Scholar]
- 6. Stoodley P, Ehrlich GD, Sedghizadeh PP, et al. Orthopaedic biofilm infections. Curr Orthop Pract 2011; 22:558–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Bouiller K, David MZ. Staphylococcus aureus genomic analysis and outcomes in patients with bone and joint infections: a systematic review. Int J Mol Sci 2023; 24:3234. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Morrison MA, Hageman JC, Klevens RM. Case definition for community-associated methicillin-resistant Staphylococcus aureus. J Hosp Infect 2006; 62:241. [DOI] [PubMed] [Google Scholar]
- 9. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas 1960; 20:37–46. [Google Scholar]
- 10. Hazra A, Gogtay N. Biostatistics series module 7: the statistics of diagnostic tests. Indian J Dermatol 2017; 62:18–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Webb J, Cella E, Sutcliffe CG, et al. Multi-strain carriage and intrahost diversity of Staphylococcus aureus among indigenous adults in the USA. Microb Genom 2025; 11:001367. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Lavigne J-P, Hosny M, Dunyach-Remy C, et al. Long-term intrahost evolution of Staphylococcus aureus among diabetic patients with foot infections. Front Microbiol 2021; 12:741406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. David MZ, Daum RS. Community-associated methicillin-resistant Staphylococcus aureus: epidemiology and clinical consequences of an emerging epidemic. Clin Microbiol Rev 2010; 23:616–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Choi S-H, Dagher M, Ruffin F, et al. Risk factors for recurrent Staphylococcus aureus bacteremia. Clin Infect Dis 2021; 72:1891–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Bernard L, Arvieux C, Brunschweiler B, et al. Antibiotic therapy for 6 or 12 weeks for prosthetic joint infection. N Engl J Med 2021; 384:1991–2001. [DOI] [PubMed] [Google Scholar]
- 16. Li H-K, Rombach I, Zambellas R, et al. Oral versus intravenous antibiotics for bone and joint infection. N Engl J Med 2019; 380:425–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Lebowitz D, Gariani K, Kressmann B, et al. Are antibiotic-resistant pathogens more common in subsequent episodes of diabetic foot infection? Int J Infect Dis 2017; 59:61–4. [DOI] [PubMed] [Google Scholar]
- 18. Gjika E, Beaulieu J-Y, Vakalopoulos K, et al. Two weeks versus four weeks of antibiotic therapy after surgical drainage for native joint bacterial arthritis: a prospective, randomised, non-inferiority trial. Ann Rheum Dis 2019; 78:1114–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Coll F, Raven KE, Knight GM, et al. Definition of a genetic relatedness cutoff to exclude recent transmission of meticillin-resistant Staphylococcus aureus: a genomic epidemiology analysis. Lancet Microbe 2020; 1:e328–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Talbot BM, Jacko NF, Petit RA, et al. Unsuspected clonal spread of methicillin-resistant Staphylococcus aureus causing bloodstream infections in hospitalized adults detected using whole genome sequencing. Clin Infect Dis 2022; 75:2104–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Coll F, Toleman MS, Harrison EM, et al. Genomic evaluation of phenotypic antibiotic susceptibility patterns as a surrogate for MRSA relatedness and putative transmission during outbreak investigations. Infect Prev Pract 2024; 7:100435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Azarian T, Ridgway JP, Yin Z, David MZ. Long-term intrahost evolution of methicillin resistant Staphylococcus aureus among cystic fibrosis patients with respiratory carriage. Front Genet 2019; 10:546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Dermota U, Šturm AC, Triglav T, Smrdel KS, Velimirović I. Whole genome sequencing and molecular epidemiology of methicillin-resistant Staphylococcus aureus isolated from patients with bacteraemia in Slovenia. Eur J Clin Microbiol Infect Dis 2024; 43:969–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Wen Q, Gu F, Sui Z, Su Z, Yu T. The process of osteoblastic infection by Staphylococcus aureus. Int J Med Sci 2020; 17:1327–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Podkowik M, Perault AI, Putzel G, et al. Quorum-sensing agr system of Staphylococcus aureus primes gene expression for protection from lethal oxidative stress. Elife 2024; 12:RP89098. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.