Reporting of Azithromycin Activity against Clinical Isolates of Extensively Drug-Resistant Salmonella enterica Serovar Typhi

ABSTRACT.

This study aimed to evaluate the minimum inhibitory concentration (MIC) of azithromycin (AZM) in clinical isolates of extensively drug-resistant (XDR) Salmonella Typhi (i.e., resistant to chloramphenicol, ampicillin, trimethoprim-sulfamethoxazole, fluoroquinolones, and third-generation cephalosporin) using the E-test versus the broth microdilution method (BMD). From January to June 2021, a retrospective cross-sectional study was carried out in Lahore, Pakistan. Antimicrobial susceptibility was performed initially by the Kirby-Bauer disk diffusion method for 150 XDR Salmonella enterica serovar Typhi isolates, and MICs of all the recommended antibiotics were determined by the VITEK 2 (BioMérieux) fully automated system using Clinical Laboratory Standard Institute (CLSI) 2021 guidelines. The E-test method was used to determine AZM MICs. These MICs were compared with the BMD, which is the method recommended by the CLSI but not adopted in routine laboratory reporting. Of 150 isolates, 10 (6.6%) were resistant by disk diffusion. Eight (5.3%) of these had high MICs against AZM by the E-test. Only three isolates (2%) were resistant by E-test, having an MIC of 32 μg/mL. All eight isolates had a high MIC by BMD with different MIC distributions, but only one was resistant, having an MIC of 32 μg/mL by BMD. The sensitivity, specificity, negative predictive value, positive predictive value, and diagnostic accuracy of the E-test method versus BMD were 98.65%,100%, 99.3%, 33.3%, and 98.6%, respectively. Similarly, the concordance rate was 98.6%, negative percent agreement was 100%, and positive percent agreement was 33%. The BMD is the most reliable approach for reporting AZM sensitivity in XDR S. Typhi compared with the E-test and disk diffusion methods. Potentially, AZM resistance in XDR S. Typhi is around the corner. Sensitivity patterns should be reported with MIC values, and if possible, higher values should be screened for the presence of any potential resistance genes. Antibiotic stewardship should be strictly implemented.

INTRODUCTION

Enteric fever is caused by Salmonella enterica serovar Typhi (S. Typhi) and Salmonella Paratyphi A, B, and C. Particularly in low-and middle-income countries (LMICs), typhoid fever is the major cause of morbidity and mortality and is linked to substandard personal hygiene and sanitary practices. Enteric fever is endemic in South Asia, South America, Africa, and the Middle East. It spreads via contaminated food and water.¹

The WHO estimates the global burden of enteric fever to be between 11 and 21 million cases, with 128,000 to 161,000 typhoid-related deaths annually.² Antimicrobial resistance in Salmonella can be linked to the horizontal transfer of antibiotic-resistant genes commonly present on mobile genetic elements across Salmonella strains and other enterobacteria. The widespread clonal spread of antimicrobial drug-resistant species is also responsible for antimicrobial resistance.

In Sindh, Pakistan, a substantial number of ceftriaxone-resistant cases have been reported since November 2016. A similar case was reported in a visitor from Pakistan to Australia.^3,4 These S. Typhi strains are resistant to chloramphenicol, ampicillin, trimethoprim-sulfamethoxazole, fluoroquinolones, and third-generation cephalosporin, being characterized as extensively drug resistant (XDR). The treatment choices for this novel XDR strain are azithromycin (AZM) and carbapenems, making AZM the only oral antibiotic in the choice.³ In specific regions of Pakistan where prevailing conditions exist, the prevalence of XDR typhoid has increased from 7/100,000 to 15/100,000.⁵ Since September 12, 2020, 2,883 cases have been reported in Pakistan, and all of them were extensively resistant isolates.⁶ According to the WHO, almost 10,365 cases of XDR S. Typhi were reported by 2019.⁷

Several cases of multidrug-resistance S. Typhi were reported previously in Southeast Asia, with the highest incidence in Vietnam.⁸ Sporadic case reports have emerged from various parts of Pakistan.⁹ The lack of antimicrobial stewardship with over-the-counter antibiotic availability contributes to increasing antimicrobial resistance in endemic regions.¹⁰ As a result of widespread empiric use of AZM during the coronavirus disease 2019 pandemic, there is a growing risk of AZM resistance.¹¹ This misuse of antibiotics may lead to a complete lack of options for treating XDR typhoid.⁶

Antimicrobial susceptibility testing is necessary to correctly diagnose and manage XDR typhoid infections. According to the Clinical Laboratory Standard Institute’s (CLSI’s) M100 Standards 2021, AZM is still an investigational medicine for Salmonella enterica serovar Typhi, and breakpoints are based on minimum inhibitory concentration (MIC) distribution data and little clinical data.¹² Discordance between disk diffusion and MICs for typhoidal Salmonellae has been noted.¹³ Furthermore, some case reports have suggested treatment failure, necessitating reliable susceptibility testing procedures.^14,15 The purpose of this study was to compare the accuracy of the disk diffusion method, which is commonly used in most laboratories across Pakistan, with that of the E-test and broth microdilution methods (BMDs) to report AZM resistance in clinical isolates of XDR S. Typhi.

MATERIALS AND METHODS

Study design and setting.

A retrospective cross-sectional validation study looking at diagnostic accuracy was conducted in the Department of Microbiology, University of Health Sciences (UHS) in Lahore, Pakistan, from January 2021 to June 2021. The ethical review board at UHS approved the study under reference number UHS/REG-19/ERC/398.

Sample collection, processing, and culture.

In all, 835 blood cultures with suspected S. Typhi were collected from multiple tertiary care hospitals in Lahore. Samples from adults and children/infants consisted of 5–10 mL and 1–5 mL, respectively. The samples were inoculated into BACT/ALERT FN Plus (BioMérieux SA, Marcy-l’Étoile, France; SKU number 410852) blood culture bottles for adults containing 40 mL of supplemented complex media comprising adsorbent polymeric beads. BACT/ALERT PF Plus (SKU number 410853) containing 30 mL of supplemented complex media with adsorbent polymeric beads was used for pediatric patients (BioMérieux). Bottles were labeled with the patient’s age and gender and loaded into the BACT/ALERT 3D instrument (BioMérieux). Samples identified as positive by the BACT/ALERT system were subcultured and analyzed per standard protocols. MacConkey and blood agar plates (Oxoid, Basingstoke, United Kingdom) were used for subculture. Inoculated plates were incubated aerobically for 24 hours inverted at 37 °C. If no growth was observed after 24 hours, the plates were re-incubated for an additional 24 hours.

Bacterial isolate storage and identification.

The isolates were stored at −70 °C in tryptic soy broth containing 16% volume/volume glycerol (Oxoid, Basingstoke, United Kingdom). Before use, the bacterial strains were subcultured and retested for distinguishing characteristics. Working cultures were kept on tryptic soy agar slant for up to 2 weeks at temperatures ranging from 2–8 °C. The laboratory standard operating procedure for identifying the organism included colonial morphology, Gram staining, and serological and biochemical tests using the AP1–20E identification system (BioMérieux).

Biochemical identification using AP1–20E.

The AP1–20E system was used to identify organisms, providing easy and quick identification. Twenty microtubes containing a dehydrated substrate for biochemical tests were inoculated with standardized bacterial suspension prepared according to the manufacturer’s instructions and incubated at 29 °C ± 2 °C for 18–24 hours. The test results were converted to a seven-digit profile using a standard coding system, and identification was made through the analytical profile index, following the manufacturer’s instructions.

Serological identification.

Salmonella O, H, and Vi antigens (Difco; BD, Franklin Lakes, NJ) were used to perform serological identification of S. Typhi according to the manufacturer’s instructions. In short, a drop of distilled water was placed on a clean glass slide, and a single colony of the test organism from the blood agar plate was used to create a homogeneous suspension through thorough mixing. The suspension was then mixed with a sterile stick after adding one drop of S. Typhi agglutination sera. The slide was rotated for 10–15 seconds and examined with a naked eye for agglutination. For comparison, positive and negative controls were used.

Antimicrobial susceptibility testing.

The antimicrobial susceptibility pattern of S. Typhi was determined using the commercially available Kirby-Bauer disk diffusion method using Muller-Hinton agar (Oxalis, United Kingdom), per the CLSI 2021 guidelines. Antimicrobial disks with standard antibiotic content were used (Oxoid Inc., Ogdensburg, NY). The isolates were considered sensitive if the zone diameter was ≥ 13 and resistant if the zone diameter was ≤ 12. For all antimicrobial susceptibility tests, Escherichia coli (American Type Culture Collection [ATCC^®] 25922, Manassas, VA) and Pseudomonas aeruginosa (ATCC 27853) were used as the control strain.

MIC determination and interpretation.

VITEK 2 (BioMérieux) completely automated technology was used to determine the MIC of ampicillin, chloramphenicol, trimethoprim-sulfamethoxazole, ciprofloxacin, and third-generation cephalosporins. Extensively drug-resistant S. Typhi was defined as S. Typhi with high MICs against ampicillin, ceftriaxone, chloramphenicol, ciprofloxacin, and trimethoprim-sulfamethoxazole. The MIC for AZM was initially established using the E-test (Liofilchem, Roseto degli Abruzzi, Italy) method. Bacterial suspension was adjusted according to the 0.5 McFarland’s standard. Isolates were inoculated on the surface of a Mueller Hinton agar plate (Hardy Diagnostics, California) using a sterile cotton swab. E-test strips were stored at −20 °C. Prior to use, the strips were taken out and applied to the surface of the agar using sterile forceps, first putting the lowest concentration at the agar plate and then placing the strips carefully and slowly so they touched the surface of the agar completely. After placement of the E-test strips, the plates were incubated at 37 °C aerobically for 18–24 hours. The MIC results were interpreted according to CLSI 2021 guidelines. For AZM, MIC > 32 µg/mL was taken as resistant. The E-test strips were taken out of storage at −4 °C 5 minutes before inoculation.

The tests were done twice to avoid discrepancy or reader bias, and the MICs for each isolate were reported as the higher of the two results. The quality was assessed each time using the standard quality control strains of E. coli ATCC 25922 and Klebsiella pneumoniae ATCC 700603. Sigma-Aldrich Chemicals Pvt Ltd provided the AZM dihydrate used in the study for BMD. CLSI’s BMD method was used on sterile 96-well polystyrene round-bottom microplates. The BMD technique was performed using 2-fold dilutions ranging from 0.03 μg/mL to 32 μg/mL, prepared in Cation Adjusted Mueller Hinton Broth (BBL; BD). The test was repeated twice to ensure reproducibility. Control wells were kept in each row for growth and media control. The quality of every batch was assessed using the standard strains of E. coli ATCC 25922 and K. pneumoniae ATCC 700603. Distinct turbidity or button development in the growth well was a favorable sign. The lowest concentration at which the isolate was inhibited was reported as the MIC (absence of visible bacterial growth). The data were interpreted using the CLSI-suggested MIC breakpoints for AZM (susceptible 16 μg/mL and resistant 32 μg/mL).¹²

Statistical analysis.

The statistical analysis was done using Statistical Package for Social Sciences version 24.0 (IBM, Armonk, NY). Categorical variables were presented as number (percentage). The performance of the BMD was determined by calculating the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy. The sensitivity, specificity, negative predictive value, positive predictive value, and diagnostic accuracy were calculated using the following formulas, where TP is true positive, TN is true negative, FP is false positive, and FN is false negative: 1) specificity = (TN/TN + FP) × 100; 2) sensitivity = (TP/TP + FN) × 100; 3) positive predictive value = (TP/TP + FP) × 100; 4) negative predicted value = (TN/TN + FN) × 100; and 5) diagnostic accuracy = (TP + TN)/(TP + FP + FN + TN) × 100. The concordance rate, positive percent agreement (PPA), and negative percent agreement (NPA) were calculated between the BMD and E-test by using specific formulas.

RESULTS

Three hundred eighty-nine blood cultures collected from different hospitals grew S. Typhi, and 150/389 (38.5%) were XDR S. Typhi on the basis of antimicrobial susceptibility testing. Antimicrobial susceptibility was done according to CLSI 2021 guidelines (Table 1). Of the 150 isolates, 10 (6.6%) were resistant to AZM by the disk diffusion method (Table 2); in addition, 8 (5.3%) had high MIC values by E-test, but 3 (2%) were resistant, having an MIC of 32 μg/mL (Figure 1). The MIC was further confirmed by BMD, and the 8 isolates (5.3%) had high MIC values but with a different distribution (Table 2). Only 1 (0.6%) was resistant to AZM, having an MIC of 32 μg/mL by the BMD method (Figure 2). The comparison of MIC values by both methods is shown in Figure 3. The MIC₅₀ and MIC₉₀ values for AZM were 4 μg/mL for XDR S. Typhi.

Table 1.

MIC distribution of XDR Salmonella Typhi isolated from blood cultures

Antibacterial agent	Breakpoint	No. of isolates having MICs (μg/mL)												Total	Percent resistant (%)
		≤ 0.06	0.125	0.25	0.5	1	2	4	8	16	32	64	≥ 128
Ampicillin	≥ 32	0	0	0	0	0	0	0	0	0	0	95	55	150	100
Piperacillin/tazobactam*	≥ 128/4	0	0	0	8	12	10	45	33	6	13	20	3	150	2
Ceftriaxone	≥ 4	0	0	0	0	0	0	0	30	42	35	15	28	150	100
Cefixime	≥ 4	0	0	0	0	0	0	0	38	15	48	29	20	150	100
Cefotaxime	≥ 4	0	0	0	0	0	0	0	20	12	48	33	37	150	100
Cefepime	≥ 16	0	0	0	0	0	0	0	0	0	31	73	46	150	100
Imipenem	≥ 4	0	30	28	80	12	0	0	0	0	0	0	0	150	0
Meropenem	≥ 4	5	13	32	55	45	0	0	0	0	0	0	0	150	0

MIC = minimum inhibitory concentration; XDR = extensively drug resistant.

^*The MIC value of piperacillin/tazobactam is expressed as MIC values of piperacillin with an equal (4 μg/mL) concentration of tazobactam.

Table 2.

Comparison of disk diffusion, E-test, and BMD

Isolate	Disk diffusion method		MIC (E-test)		MIC (BMD)		MIC50	MIC90
	Zone (mm)	Interpretation	μg/mL	Interpretation	μg/mL	Interpretation	μg/mL	μg/mL
XDR1	10	R	16	S	8	S	4	4
XDR2	11	R	16	S	8	S	4	4
XDR3	9	R	16	S	16	S	4	4
XDR4	9	R	32	R	32	R	4	4
XDR5	12	R	16	S	16	S	4	4
XDR6	11	R	16	S	16	S	4	4
XDR7	10	R	32	R	16	S	4	4
XDR8	11	R	32	R	16	S	4	4
XDR9	12	R	4	S	4	S	4	4
XDR10	9	R	4	S	4	S	4	4

BMD = broth microdilution method; MIC = minimum inhibitory concentration; XDR = extensively drug resistant.

Figure 1.

Minimum inhibitory concentration (MIC) distribution of azithromycin (AZM) by E-test (95% CI). This figure shows MIC distribution of 150 extensively drug-resistant Salmonella Typhi isolates against AZM by the broth microdilution method. Of all 150 isolates, 92 had an MIC value of 2 μg/mL, 50 had a value of 4 μg/mL, 0 had a value of 8 μg/mL, 5 had a value of 16 μg/mL, and 3 had a value of 32 μg/mL.

Figure 2.

Minimum inhibitory concentration (MIC) distribution of azithromycin (AZM) by broth microdilution method (BMD) (95% CI). This figure shows the MIC distribution of 150 extensively drug-resistant Salmonella Typhi isolates against AZM by the BMD method. Of all 150 isolates, 92 had an MIC value of 2 μg/mL, 50 had a value of 4 μg/mL, 2 had a value of 8 μg/mL, 5 had a value of 16 μg/mL, and only 1 had a value of 32 μg/mL.

Figure 3.

Minimum inhibitory concentration (MIC) distribution of azithromycin (AZM) by broth microdilution method (BMD) and the E-test (95% CI). This figure shows MIC distribution of extensively drug-resistant Salmonella Typhi isolates against AZM by BMD and E-test methods.

The sensitivity, specificity, negative predictive value, positive predictive value, and diagnostic accuracy of the E-test versus the BMD method are shown in (Table 3). Sensitivity in our study represents the ability of a method to detect true resistant isolate to AZM, whereas specificity means true sensitive isolate. The specificity of the E-test was 100%, the sensitivity was 98.6%, the negative predictive value was 99.3%, the positive predictive value was 33.3%, and diagnostic accuracy was 98.6%. The concordance rate between the two methods was 98.6%, whereas the NPA was 100% and the PPA was 33% (Table 4).

Table 3.

Comparison of results of BMD with E-test (n = 150)

		BMD		Performance of E-test
		Resistant	Susceptible	Sensitivity	Specificity	NPV	PPV	DA
E-test	Resistant	1	2	98.65%	100%	99.3%	33.3%	98.66%
	Susceptible	00	147	–	–	–	–	–

BMD = broth microdilution method; DA = diagnostic accuracy; NPV = negative predictive value; PPV = positive predictive value.

Table 4.

Calculation of concordance rate, PPA, and NPA between the BMD and E-test

		BMD		PPA	NPA	Concordance rate
		Resistant	Susceptible
E-test	Resistant	1	2	a/a + b × 100	d/c + d × 100	a + d/a + b + c + d × 100
		a	b	33%	100%	98.6%
	Susceptible	00	147	–
		c	d

a = true positive; b = false positive; BMD = broth microdilution method; c = true negative; d = false negative; NDA = negative percent agreement; PPA = positive percent agreement.

DISCUSSION

Enteric fever is a significant health problem in LMICs. The emergence and the spread of XDR S. Typhi across Pakistan have generated a significant problem in controlling and preventing typhoid fever.¹ These S. Typhi strains are resistant to chloramphenicol, ampicillin, trimethoprim-sulfamethoxazole, fluoroquinolones, and third-generation cephalosporins.³ With limited treatment options, emerging AZM resistance would create further problems. This study reports AZM susceptibility in XDR S. Typhi in Pakistan.

Because of the limited amount of currently available clinical and antimicrobial susceptibility data, AZM is still regarded as an investigational drug. Qureshi et al.¹⁶ reported clinical failure in one isolate given oral AZM and three failures combined with meropenem, though AZM was reported as being sensitive with the disk diffusion method. Iqbal et al.¹⁷ reported 2,104 S. Typhi isolates from Karachi, Pakistan. Only one had a high AZM MIC (12 μg/mL). Klemm et al.³ recruited 84 XDR S. Typhi isolates primarily from an outbreak in Hyderabad, Pakistan, and only one isolate had an MIC of 8 μg/mL.

In a small number of cases, AZM-resistant isolates from Pakistan have been reported. One isolate had MICs of 64 μg/mL according to a recent study from Lahore, Pakistan, by Fida et al.¹⁸

In the current study, when the disk diffusion method was used, 10 isolates (6.6%) were resistant. The E-test revealed that eight of them (5.3%) exhibited high MICs against AZM, which was verified by BMD. Only three isolates (2%) were resistant by the E-test, which had an MIC of 32 μg/mL. By BMD, all eight isolates had high MICs with various MIC distributions, but only one was resistant, with a MIC of 32 μg/mL. There was a discrepancy in reporting resistance by disk diffusion, the E-test, and BMD. Because there is no gold standard available and disk diffusion breakpoints are mentioned in the CLSI, it is the most common method of reporting in clinical microbiology laboratories across Pakistan.

Laboratories in Pakistan use disk diffusion to report AZM sensitivity.¹⁷ In our study, 10 isolates (6.6%) were found to be resistant according to this method. This method may overreport resistance, as there is only a 1-mm difference in cutoff zone diameters as recommended by the CLSI (2021),¹² which consequently could reduce the use of AZM, the only oral option available to treat XDR typhoid in clinical settings. This could lead to overuse of carbapenems in XDR S. Typhi and could put selective pressure on an organism to attain more resistant elements such as carbapenamases. Gradient strip methods, called the E-test, are preferable to the disk diffusion approach, as these methods measure MIC values and are more specific for reporting sensitivity.¹⁹ Shoaib et al.¹⁰ indicated in a recent study from Karachi that the E-test could be a better choice than disk diffusion. In addition, few studies have reported problems with the E-test method.²⁰ This discrepancy could be due to reading errors, lack of expertise, inoculum preparation in the E-test method, or reader bias.²⁰ Likewise, in our study there was inconsistency between the E-test and the BMD. A BMD is probably the most appropriate method to report MIC and, hence, sensitivity as recommended by the CLSI, especially when reporting against XDR S. Typhi, for which AZM is the only oral treatment option available. This study could change the reporting paradigm of AZM from disk diffusion to only BMD to avoid overreporting of resistance and start timely treatment based on clinical data. It could also help to conserve carbapenems for use against metallo-beta-lactamase producers.

A BMD seems to be a more sensitive method for detecting AZM resistance in XDR typhoid cases and should be a routine practice in diagnostic laboratories, especially to cross-check isolates reportedly resistant on disk diffusion and E-test methods. It will also help clinicians to decide treatment based on standard reporting.

CONCLUSION

The outcomes of the current study demonstrated that disk diffusion and the E-test may lead to overreporting of AZM resistance and are a less-reliable approach for reporting AZM sensitivity in S. Typhi compared with BMD. As a result of this overreported resistance, AZM use in XDR S. Typhi may decline, which would lead to the extensive use of carbapenems and the acquisition of resistance against them. Careful reporting of AZM against XDR S. Typhi isolates is the need of the hour. A more accurate method for reporting AZM resistance is BMD. The disk diffusion method, which is routinely used in the majority of laboratories, should be avoided, especially in XDR S. Typhi clinical isolates, to avoid reporting over resistance.

LIMITATIONS

Because of budget constraints, we were unable to do molecular analysis of AZM-resistant isolates, which could have given better insight into the molecular mechanism of resistance against AZM. When three tests were compared and there were no definitive gold standards, use of latent class analysis to evaluate performance between the assays was not included in our analysis.