Haemophilus influenzae blood-stream infection and third-generation cephalosporin susceptibility testing: a comparative case study using EUCAST and CLSI guidelines

Introduction. In this comparative case study, we discuss clinically relevant discrepancies of antimicrobial susceptibility testing (AST) interpretation for ceftriaxone against a non-typable, beta-lactamase negative, ampicillin-resistant (BLNAR) Haemophilus influenzae isolated from a blood culture. Case report. A 74-year-old man presented with a 3 day illness characterized by shortness of breath and dry cough, and was noted to be febrile and hypoxic on admission. A blood culture bottle flagged positive with Gram-negative coccobacilli, later identified as Haemophilus influenzae with the patient commenced on ceftriaxone. The isolate was beta-lactamase negative and antibiotic susceptibility testing (AST) using disc diffusion revealed the isolate resistant to ceftriaxone and ampicillin by EUCAST methodology, with the patient subsequently changed to amoxicillin/clavulanate. Further AST using the CLSI methodology in parallel demonstrated discrepant results between the two susceptibility methods. The patient recovered without complications. Conclusion. This discrepancy could lead to inconsistent reporting of susceptibilities between laboratories, and consequently antibiotic prescribing, especially for invasive isolates. As more laboratories adopt EUCAST methodologies for AST interpretation in Australia and globally, it is important for clinicians to consider the clinical implications of these methodological discrepancies.


INTRODUCTION
Haemophilus influenzae (HI) is a small fastidious Gram-negative bacterium, associated with a spectrum of infections, ranging from mild infections, such as otitis media or bronchitis, to life-threatening infections, such invasive disease or epiglottitis.While HI serotype B is now rare in Australia due to mass vaccination, non-B HI invasive disease, namely blood-stream infections and meningitis, continue to cause significant morbidity and risk of mortality [1].Antimicrobial susceptibility testing (AST) of HI isolates from life-threatening infections remains essential to targeting therapy at the individual level as well as empirical therapy prescribing through population-level surveillance [2,3].Third-generation cephalosporins, such as ceftriaxone and cefotaxime, are broad spectrum antibiotics commonly used in Australia, and remain critical components of life-saving empirical therapy for suspected meningitis, as well as a range of severe Gram-negative infections, including blood-stream infections [1].Therefore, discrepancies in susceptibility reporting for third-generation cephalosporins between the two of the most common AST methodologies [4,5] used by laboratories have major implications for population-level empirical treatment of life-threatening infections, as well as individual patient care.We have previously demonstrated how EUCAST Version 12 and CLSI methodologies give clinically relevant, discrepant results for classifying isolates of HI as beta-lactamase negative, ampicillin-resistant (BLNAR) [6]; although BLNAR detection has major implications for aminopenicillin and amoxicillin-clavulanic acid prescribing, it is increasingly recognized to be a signal for risk of resistance to third-generation cephalosporins in Europe and Japan via acquisition of mutations within penicillin-binding-protein-3 (PBP3) [7][8][9][10][11][12][13]. Nevertheless, implementing the EUCAST methodology means that a clinician would be more likely to change antibiotics to a more restricted agent (e.g.moxifloxacin), with implications for antimicrobial stewardship [14,15].However, PBP3 target alteration is rarely mediated by single mutations, and consistent correlation between PBP3 genotype and resistance phenotype has proven difficult, which can be explained at least in part by considerable heterogeneity of sequence types (STs) between isolates of non-B serotype HI [8].This genotypic/phenotypic uncertainty, combined with a paucity of published clinical outcome data comparing the EUCAST (version 12) and CLSI (M100-S31, 2021) methodologies, creates a dilemma with respect to treatment decisions and antimicrobial stewardship of restricted antibiotics in the Australian context.In this comparative case study, we demonstrate the critical differences in laboratory methodologies and interpretations may have a significant impact on management of life-threatening infection with BLNAR HI, with wider implications for empirical antibiotic choice.

CASE REPORT
We describe the case of a 74-year-old man who presented to the Emergency Department with a 3 day illness characterized by shortness of breath and dry cough.His past medical history included obstructive sleep apnoea, chronic obstructive airway disease, non-insulin-dependent diabetes, hypertension and hypercholesterolemia. On admission, he was noted to be febrile (38.2 °C), tachypnoeic (28 breaths/min), hypoxic (requiring 4 l of oxygen via nasal prongs to maintain peripheral oxygen saturation >95 %), tachycardic (133 beats/min) with an irregular heart rhythm, and he was also noted to have left-sided chest crepitations.There was no clinical evidence of delirium or meningoencephalitis.Initial blood investigations revealed an elevated peripheral white blood cell count (27.3×10 9 /l) with neutrophilia (25.4×10 9 /l), and an elevated C-reactive protein (163 mg l −1 ).Chest x-ray findings were consistent with multi-lobar consolidation and he returned a negative nasopharyngeal/throat swab for SARS-CoV-2.He was admitted to hospital with a diagnosis of severe communityacquired pneumonia (SMART-COP=5) and commenced intravenous antibiotic treatment with ceftriaxone (2 g daily) and azithromycin (500 mg daily) [1].
An aerobic blood culture bottle collected prior to antibiotics and incubated on the automated BacT/Alert blood culture system (bioMerieux, Australia) flagged positive on day 1 of admission, with Gram negative coccobacilli on Gram stain, and was subcultured to routine media.An isolate grown on chocolate agar after 4 h was identified by MALDI-TOF mass spectrometry (Bruker, Australia) as Haemophilus influenzae (HI, score=2.14).In light of this result, azithromycin was ceased after two doses and ceftriaxone 2 g daily was continued empirically as monotherapy for invasive HI, pending the results of AST.The patient's clinical course improved, with resolution of his fevers and reduction in his inflammatory markers.Subsequently, the HI isolate was reported as beta-lactamase negative, but resistant to ceftriaxone and ampicillin (EUCAST version 12.0, 2022, methodology) [4], so monotherapy with ceftriaxone was changed on day 4 of admission to intravenous amoxicillin/clavulanate (1000/200 mg 8-hourly); on day 6 of admission, the patient was changed to oral amoxicillin/clavulanate (875/125 mg 12-hourly) to complete a further 5 days of therapy.He was discharged home on day 10 of admission without any complications.The patient's isolate was later sent to a reference laboratory for molecular serotyping, where it was reported as non-typable HI.
In light of the unusual and clinically important result for ceftriaxone on an invasive isolate, the AST process for this specimen was reviewed in detail.Beta lactamase testing was negative by Nitrocefin disc test (Thermofisher, Australia).AST testing was performed using disc diffusion testing using commercial solid media and antibiotic impregnated discs (Thermofisher, Australia).MIC values were determined by gradient diffusion on solid media with commercial epsilometer (E)-test strips (BioMerieux, Australia).For AST by the EUCAST (version 12.0, 2022) disc and E-test methodologies, the isolate was inoculated on Mueller-Hinton agar with 5 % defibrinated horse blood and 20 mg l −1 NAD (Thermofisher, Australia), and incubated in 5 % CO 2 at 35-37 °C for 18-20 h.
We retested the isolate by the EUCAST (version 12.0, 2022) method [4] and performed parallel testing using the CLSI (M100-S31, 2021) methodology [5] as a comparator, inoculating the isolate on HTM media and incubated in 5 % CO 2 at 35-37 °C for 16-18 h.Both EUCAST (version 12.0, 2022) and CLSI (M100-S31, 2021) methods required discs containing ceftriaxone 30 ug, while only EUCAST had breakpoints for cefotaxime (5 ug disc) [4].The EUCAST (version 12.0, 2022) flow chart [4] based on the penicillin 1 µg disc screen test for beta-lactam resistance mechanism was followed and found to be resistant and further susceptibility tests performed for the relevant agents and interpreted according to their breakpoints (Table 1).Inhibition disc zone diameters (DD) were measured using callipers.For quality control, Haemophilus influenzae ATCC 49247 was used.Our results, shown in Table 1, demonstrate discrepant third-generation cephalosporin AST interpretations despite identical values obtained for disc zone diameters and MICs by the different methodologies, and indicate that differences in media may not be significant for testing these antibiotics against HI.

DISCUSSION
There has been little specific information published on clinical outcomes of invasive infections caused by HI strains demonstrating discrepant ceftriaxone-or cefotaxime-resistant interpretations between EUCAST (version 12.0, 2022) and CLSI (M100-S31, 2021) methodologies.Furthermore, there is limited published information on outcomes for invasive infections caused by strains that are demonstrated to have MICs of 0.125 and 2 mg l −1 by either methodology.
Our clinical case demonstrates how MIC breakpoints for third-generation cephalosporins differ between the EUCAST (version 12.0, 2022) and CLSI (M100-S31, 2021) methodologies for invasive infections caused by HI.This is a major concern for clinical laboratories, especially for those located in regions where invasive beta-lactamase negative, ampicillin-resistant (BLNAR) isolates of HI are very common.EUCAST (version 12.0, 2022) MIC breakpoints values for cefotaxime and ceftriaxone are 0.125 mg l −1 , while CLSI (M100-S31, 2021) MIC breakpoints levels are 2 ug ml −1 ; Nurnberg and colleagues describe in detail that this difference is due to different methodological approaches in breakpoint determination for HI, including the application of underlying assumptions in PK/PD modelling [16].In addition, Nurnberg and colleagues also pointed out in their study that this reduced susceptibility to third-generation cephalosporins is not as rare in BLNAR isolates of HI as suggested by EUCAST (version 12.0, 2022) recommendations [16].
While whole-genome sequencing (WGS) is not routinely available in clinical laboratories to guide decision making, it has provided some mechanistic insights into the clinical relevance of these discrepant breakpoint interpretations.WGS of isolates submitted for population-level surveillance is increasingly informing a conservative approach to breakpoint selection, as genetic mechanisms of selection are identified that produce higher MIC values in vitro.For example, the high-level resistance from the expression of TEM-1 beta-lactamase genes has informed routine laboratory testing of HI for decades, usually rapidly detectable through Nitrocefin disc test [2].However, WGS has shown that HI may harbour non-synonymous mutations in the ftsI gene, producing amino acid substitutions in penicillin-binding protein 3 (PBP3), which may confer in vitro resistance to cefotaxime [7][8][9]11].Such resistance is more problematic to detect without conservative breakpoints, but even conservative breakpoints can fail to distinguish non-wild-type isolates [7,11].In addition, at least one French study has suggested that the genotype and phenotype of invasive and non-invasive strains carrying ftsI gene mutations could be sufficiently different, such that the detection of these mutations in an invasive isolate may be insufficient to predict clinical outcome [8].
Furthermore published evidence on the correlation between PBP3 mutations, in vitro MICs of 0.25-2 mg l −1 , and clinical outcome remains scarce and contradictory.Skaare et al. reported a Norwegiuan HI from a blood-stream infection in an immunocompromised patient, with an MIC ≤0.12 mg l −1 (susceptible; EUCAST version 12.0, 2022, methodology); the patient responded well to an initial 3 days of cefotaxime before being changed to oral ciprofloxacin, but the isolate was subsequently found to have the ftsI mutations, resulting in PBP3 amino acid substitutions (S385T and N526K) [11].a non-invasive French BLNAR isolate from the respiratory tract of an immunocompromised patient with chronic lung disease; the cefotaxime MIC was 0.25 mg l −1 (resistant; EUCAST version 12.0, 2022, methodology) and ceftriaxone MIC of 0.094 mg l −1 (susceptible; EUCAST version 12.0, 2022, methodology) and both clinical and microbiological failure was reported after 8 days of IV ceftriaxone [7].The isolate was subsequently found to have the ftsI mutations, producing PBP3 amino acid substitutions (S357N, M377I, S385T, L389F, A502T and N526K).Nishimura et al. reported a Japanese non-typeable BLNAR HI causing bacteraemia, secondary to infection of a massive uterine adenomyoma with endometritis; the isolate returned an MIC of 1 mg l −1 to cefotaxime (susceptible; CLSI M100-S31, 2021 methodology), and the patient became afebrile after 7 days of ceftriaxone, despite source control not occurring for a further 10 days (ftsI sequencing was not performed) [13].

CONCLUSION
We have demonstrated that two of the most common AST methodologies used in Australian and other global laboratories, when applied to an invasive non-typeable BLNAR HI isolate, produce discrepant results for the most commonly prescribed antibiotics used empirically for this life-threatening infection.Given these discrepancies, and the lack of available clinical outcome for isolates with discrepant results, we cannot say whether or not the patient had a favourable outcome due to the presence of azithromycin in the empirical therapy and its long half-life after cessation, or whether the ceftriaxone therapy was itself sufficient.However, this case demonstrates that the clinical decision to change to intravenous amoxycillin/clavulanate (1000/200 mg 8-hourly) was clearly based on the AST results achieved with the EUCAST methodology, and it remains unclear if the patient would have relapsed had the conflicting results of the CLSI methodology been used instead.
Therefore, clinicians should be aware of this possibility when encountering life-threatening infections caused by non-typeable, BLNAR HI isolates.Clearly, further research is required to determine the clinical implications of these methodological discrepancies for individual patient care and antimicrobial stewardship, as well as population-level determinations of empirical antibiotic regimens.
The Microbiology Society is a membership charity and not-for-profit publisher.
Your submissions to our titles support the community -ensuring that we continue to provide events, grants and professional development for microbiologists at all career stages.

Find out more and submit your article at microbiologyresearch.org
• Editor:There is an interesting discussion in the response to the reviewers' comments about the EUCAST/CLSI discrepancies based on the results of Nurnberg et al. (2021).I think including this in the report's introduction or discussion section (where considered more appropriate) would help to further clarify the point of the study.
Corresponding Author:We have included this in our Discussion -Lines 94 to 103.
Our clinical case demonstrates how MIC breakpoints for third-generation cephalosporins differ between the EUCAST (version 12.0, 2022) and CLSI (M100-S31, 2021) methodologies for invasive infections caused by HI.This is a major concern for clinical laboratories, especially for those located in regions where invasive beta-lactamase negative, ampicillin resistant (BLNAR)isolates of HI are very common.EUCAST (version 12.0, 2022) MIC breakpoints values for cefotaxime and ceftriaxone are 0.125 mg/L, while CLSI (M100-S31, 2021) MIC breakpoints levels are 2 ug/ml; Nurnberg and colleaguesdescribe in detail that this difference is due to different methodological approaches in breakpoint determination for HI, including the application of underlying assumptions in PK/PD modelling [16].In addition, Nurnberg and colleagues also pointed out in their study that this reduced susceptibility to third generation cephalosporins is not as rare in BLNAR isolates of HI as suggested by EUCAST (version 12.0, 2022) recommendations [16].
E have also included reference 16 to the manuscript.
• Editor:Please include when possible the version and release dates of the EUCAST and CLSI guidelines used as reference in this study, not only in the reference list, but in the main text.
Corresponding Author:As Suggested we have included throughout the text the version and release dates of the EUCAST and CLSI guidelines used as reference in this study, not only in the reference list, but also in the main text.
• Editor:L97, L112: there are references to amino acid substitutions as mutations.It is true that amino acid changes are produced by mutations in the DNA sequence, but both concepts are not interchangeable.As mutations happen at the DNA sequence level, it would be ftsImutations those causing amino acid substitutions in PBP3.Please correct these sentences accordingly.

Corresponding Author:
We agree totally to your comments and have amended the manuscript accordingly now appearing in our Discussion Lines 104 to 116.
While whole genome sequencing (WGS) is not routinely available in clinical laboratories to guide decision making, it has provided some mechanistic insights into the clinical relevance of these discrepant breakpoint interpretations.WGS of isolates submitted for population-level surveillance is increasingly informing a conservative approach to breakpoint selection, as genetic mechanisms of selection are identified that produce higher MIC values in vitro.For example, the high-level resistance from the expression of TEM-1 beta-lactamase genes has informed routine laboratory testing of HI for decades, usually rapidly detectable through Nitrocefin disc test [2].However, WGS has shown that HI may harbour non-synonymous mutations in the ftsIgene, producing amino acid substitutions in penicillin-binding protein 3 (PBP3), which may confer in vitroresistance to cefotaxime [7][8][9]11].Such resistance is more problematic to detect without conservative breakpoints, but even conservative breakpoints can fail to distinguish non-wild-type isolates [7,11].In addition, at least one French study has suggested that the genotype and phenotype of invasive and non-invasive strains carrying ftsIgene mutations could be sufficiently different, such that the detection of these mutations in an invasive isolate may be insufficient to predict clinical outcome [8].

2.
No data was generated during this research or is required for the work to be reproduced, this has bee stated explicitly under a Data Summary heading.Comments: Thank you very much for submitting your revised manuscript to Access Microbiology and for your efforts in applying the reviewers' suggestions and comments to your manuscript.This has undoubtedly improved its quality.However, I would like to make some additional suggestions that should be considered in the interest of the reader's understanding of this case report: • There is an interesting discussion in the response to the reviewers' comments about the EUCAST/CLSI discrepancies based on the results of Nurnberg et al. (2021).I think including this in the report's introduction or discussion section (where considered more appropriate) would help to further clarify the point of the study.• Please include when possible the version and release dates of the EUCAST and CLSI guidelines used as reference in this study, not only in the reference list, but in the main text.• L97, L112: there are references to amino acid substitutions as mutations.It is true that amino acid changes are produced by mutations in the DNA sequence, but both concepts are not interchangeable.As mutations happen at the DNA sequence level, it would be ftsI mutations those causing amino acid substitutions in PBP3.Please correct these sentences accordingly.

Authors comments to the Editor:
Dear Editor, Thank you for reviewing our article.I would like to point out to the reviewers that this is a Clinical Case presentation on how MIC breakpoints for third-generation cephalosporins differ between EUCAST and CLSI for invasive infections caused by H. influenzae(HI).This is a major concern in our region especially for invasive beta-lactamase negative ampicillin resistant (BLNAR)isolates which are very common in our region.EUCAST MIC breakpoints values for cefotaxime and ceftriaxone are 0.125 mg/L, while CLSI MIC breakpoints levels are 2 ug/ml.The difference has been explained in detail by a comprehensive study by Nurnberg and colleagues(2021) 1 that this difference is due to different approaches in antimicrobial susceptibility testing for HI.

Nurnberg and colleagues point out that:
"This considerable divergence is due to different approaches.The breakpoint given by CLSI is based on the PK/PD modelling of cefotaxime using the common dose of 2 g × 3. EUCAST lowered the breakpoint according to the epidemiological cut-off value as isolates with reduced susceptibility are rare or have not been reported and clinical outcome is uncertain." Nurnberg and colleagues also pointed out in their study that this reduced susceptibility is not as rare in BLNAR H. influenzaeisolates as depicted by EUCAST.This Clinical Case presentation illustrates the problems that maybe encountered by certain laboratories undergoing changes from antimicrobial susceptibility testing from CLSI to EUCAST without whole genome sequencing in detecting alterations in PBP3 in beta-lactamase negative ampicillin resistant invasive HI isolates.
The reviewers rightly point out that our methodology did not include the testing of penicillin 1 unit disk, we have done this and included our results in Table 1.As expected, it was found to be resistant.To our defence, Skaare and colleagues 2 in another study found that screening for alteration in PBP3 by disk diffusion (EUCAST method) in beta-lactamase negative isolates using penicillin 1 unit disk was 96.2% sensitive and 94 % specific in their study, while our method in their study using AMP2 disk was 89.4% specific but had a greater specificity of 98%.We have included both disk results in Table 1 of the manuscript.

Note:
For the Editor and Reviewers we have included a Figure of our penicillin1 unit disk and Etest with penicillin showing resistance.We wish not to publish this JPEG figure in the article.
We also have included the EUCAST methodology of testing as depicted their guidelines, again not for publication in the article we wish to publish.used the isolate was found to be BLNAR and therefore I would assume that the isolate could be reported as susceptible not resistant and therefore the two sets of guidelines agree?Agreed in part.We agree that these additional data improve the clarity of our article but do not change the central result/ argument.Therefore, we have now included penicillin 1 unit disk diffusion screening test results (EUCAST methodology, see updated Table 1 in article).
However, this screening testing demonstrated that our isolate was penicillin resistant.According to the published EUCAST Version 12 algorithm for beta lactam resistance screening in H. influenzae, this penicillin resistant, beta-lactamase negative isolate should be reported'according to the clinical breakpoint for the agent in question' .Therefore, we cannot infer cephalosporin susceptibility from the penicillin result.In our clinical context, accurate phenotypic determination of cephalosporin susceptibility is of critical importance to early decision-making in patient care.However, we agree that the penicillin screening result is important to clarify our central argument that the EUCAST and CLSI guidelines give conflicting results for critical antibiotic testing in our isolate.Comments: In this manuscript, Merlino et al. present a case report about the course of treatment of a non-typable H. influenzae infection and discuss the discrepancies emerged during the antibiotic resistance tests performed with this isolate.The case is explained with clarity and the manuscript is easy to follow.However, several concerns have emerged after the review process.Please, address the reviewers' suggestions and comments thoroughly, especially those concerning: • Improvement of the analysis of the literature in the Discussion section.• Clarify the observation of the guidelines provided by the EUCAST and the CLSI, addressing the point made by reviewer 2 that the isolate could be reported as "susceptible" based on the guidelines of both EUCAST and CLSI.• Elaborate and discuss how strategies to address the discrepancies presented, if after the revisions this were still appropriate.Please, provide a revised manuscript containing all suggestions and a point-by-point response to the reviewers' comments within 1 month.

Reviewer 2 recommendation and comments
https://doi.org/10.1099/acmi.0.000578.v1.3 © 2023 Anonymous.This is an open access peer review report distributed under the terms of the Creative Commons Attribution License.

Date report received: 20 March 2023 Recommendation: Major Revision
Comments: The EUCAST guidelines state that "The benzylpenicillin 1 unit disk diffusion screening test shall be used to exclude beta-lactam resistance mechanisms.When the screen is negative (zone diameter ≥12 mm) all cephalosporins for which clinical breakpoints are available, including those with "Note", can be reported susceptible without further testing, except for cefuroxime oral, which if reported, should be reported "susceptible, increased exposure" (I)."Although the benzylpenicillin disks were not used the isolate was found to be BLNAR and therefore I would assume that the isolate could be reported as susceptible not resistant and therefore the two sets of guidelines agree?

VERSION 2 Editor
recommendation and comments https://doi.org/10.1099/acmi.0.000578.v2.1 © 2023 de Dios R.This is an open access peer review report distributed under the terms of the Creative Commons Attribution License.Rubén de Dios; Brunel University London, Life Sciences, UNITED KINGDOM Date report received: 25 August 2023 Recommendation: Minor Amendment

FIGURE 1 :
FIGURE 1: Testing of H. influenzaeisolate showing resistance to penicillin.

Table 1 .
E-test MIC values, disc zone diameters and interpretations of antibiotics for the Haemophilus influenzae strain detected in blood culture using third-generation cephalosporins Thomas et al. reported