Combination agent antibiogram of common antimicrobial classes viz β-lactam drugs, aminoglycosides, fluoroquinolones and folate antagonist for major gram-negative pathogens—an innovative approach to assist appropriate empirical decisions

AST method

The blood culture isolates of major five gram-negative pathogens were enrolled into the study—three Enterobacterales (ENB) species Escherichia coli, Klebsiella pneumoniae, and Enterobacter cloacae complex; and two non-fermenters (NFGNB) i.e., Acinetobacter baumannii complex, and Pseudomonas aeruginosa. The AST was carried out in by VITEK2 system, using two AST panels—N405 for Enterobacterales, N406 for non-fermenters. 9 The AST method performed was in accordance with the guidelines provided by the manufacturer and the Clinical and Laboratory Standards Institute (CLSI) M100 and M07.10, 11 The laboratory was engaging in an external quality assessment services (EQAS) program and adhering to the weekly AST quality control (QC) methodology, as per CLSI M07. 11, 12

Data collection

The AST data [minimum inhibitory concentration (MIC) value and interpretation] were collected using ‘Clinical Microbiology Reporting software’, developed by JIPMER in collaboration with Ibhar Pvt. Ltd., which was routinely used by our institute for reporting of blood culture. The AST report was validated as per CLSI M39. 2 and other standard guideline. 1, 11 All erroneous AST data were excluded from analysis and suspicious AST results were reconfirmed before inclusion into antibiogram.

Routine antibiogram

First, the routine antibiogram was prepared for major five gram-negative pathogens—E. coli, K. pneumoniae, and E. cloacae complex, A. baumannii complex, and P. aeruginosa. The ‘first-isolate data’ were collected; defined as the first-isolate of a given species per patient per analysis period (e.g., one year in our study). The antimicrobial agents included in the routine antibiogram were—cephems (ceftriaxone [for ENB only], ceftazidime [for NFGNB only] and cefepime), β-lactam combination agents (amoxicillin-clavulanate [for E. coli and K. pneumoniae only], cefoperazone-sulbactam and piperacillin-tazobactam), carbapenems (meropenem, imipenem and ertapenem [for ENB only]), quinolones (ciprofloxacin, levofloxacin [for NFGNB only],) and aminoglycosides (gentamicin [for all except P. aeruginosa] and amikacin), and cotrimoxazole (for all except P. aeruginosa). The isolates of an organism species for which results were not available for any of the antimicrobial agents (for several reasons such as intrinsic resistant, no breakpoint available, VITEK2 got terminated etc.) were excluded from analysis. The routine antibiogram was expressed in terms of susceptibility (S%) percentage. Both susceptible (S) and susceptible dose dependent (SDD) results of the first-isolate were taken for numerator, whereas any all the first-isolates tested and valid AST results are available for particular antibiotic were used as denominator data for calculating S%. The antibiogram developed was validated as per CLSI M39 recommendations and any suspicious or outlier S% were reviewed before their inclusion into antibiogram.

Combination agent antibiogram

The various combination regimens chosen for combination agent antibiogram comprised of one β-lactam agent (cephems or BLBLI or carbapenem), quinolones or aminoglycosides. S% to the combination agents was calculated as % of isolates susceptible to either of the antimicrobial agents. For example, S% to meropenem/amikacin combination included the isolates that were susceptible to meropenem and resistant to amikacin, susceptible to amikacin and resistant to meropenem and susceptible to both meropenem and amikacin. The combination agent antibiogram in the current study was only prepared from common pathogenic GNBs in our settings and also which are common causes nation-wide. These are often MDR in our settings with very less number of single agents having >80% susceptibility and might require combination therapy for empirical choices. Other GNBs are not included in the study as they are encountered rarely as a pathogen in our setting. GPCs were not included in the present study, as almost all of them have good empirical coverage with one single agent and combination empirical therapy is seldom needed. Further stratification of combination agent antibiogram into patientcare locations (e.g., ICUs or IPDs or OPDs), treating specialities (e.g., medicine, surgery etc) and HAIs/CAIs and age wise stratifications etc. were not performed as the minimum number of isolates required to prepare antibiogram at these stratified levels were not met.

Statistical analysis

Chi-square test was used to analyse the differences in S% between the single antimicrobial agents and to combination regimens, and a p value of < 0.05 was considered statistically significant as recommended in the CLSI M39 and other standard antibiogram references 2, 1. All analyses were conducted using IBM SPSS version 23.

Combination agent antibiogram for E. coli

For E. coli, several combination regimens using one β-lactam agent plus either aminoglycoside or cotrimoxazole were found to have statistically significant increase in S% compared to the S% of individual agents (Table 1). The multidrug resistance (MDR) rates in E. coli isolates included in our study was found to be 73.7% which is quite high. However, the extensive drug resistance (XDR) rate was comparatively low, i.e. only 7.6%. This might be the reason why we have observed significant improvement in S% for combination regimens as compared to individual drugs. This provides several options of combination regimen for E. coli as empirical choice. While cotrimoxazole combination regimens were found to be useful when combined with lower line β-lactams such as ceftriaxone, cefepime and amoxicillin-clavulanate; the aminoglycoside combination regimens were found beneficial when combined with higher line BLBLI such as cefoperazone-sulbactam and piperacillin-tazobactam, or carbapenems such as meropenem, imipenem, ertapenem. 7, 14

E. coli is usually the most common gram-negative pathogen isolated from clinical specimens, as in our study. Therefore, when intended to choose a gram-negative empirical regimen, the prescribers often incline towards choosing an agent that gives a broad coverage to E. coli. Moreso, empirical regimen for E. coli is used by the clinicians in two clinical situations—(1) when the identification of the organism is known but AST report is awaited; (2) in clinical situations where E. coli is primarily suspected as the underling pathogen (e.g. community- or healthcare associated urinary tract infection, urosepsis, etc.). Having a knowledge on the estimated S% of various combination regimens will definitely guide the clinicians to choose the correct combination of antimicrobial agents in the regimen. 3, 13 Based on the results of our study, it can be proposed that, the clinicians in our facility, who intend to select a combination regimen with one of the lower line β-lactams such as ceftriaxone or cefepime, or amoxicillin-clavulanate as the first agent, should prefer to add cotrimoxazole as the second agent. But for hemodynamically unstable or severely-ill patients, if clinicians decide to choose a higher line β-lactam agent such as cefoperazone-sulbactam and piperacillin-tazobactam, or carbapenems, then the second agent added should be an aminoglycoside in order to give a maximum increase in expected S% of the combined regimen. 15, 16

Combination agent antibiogram for K. pneumoniae

For K. pneumoniae, in contrast to E. coli, only a limited combination regimens were found to be statistically significant increase in S% compared to the S% individual agents. If cefoperazone-sulbactam or any of the carbapenems as the β-lactam is used, then the second agent selected in the combination regimen should be gentamicin. Whereas combination regimens using piperacillin-tazobactam should use cotrimoxazole as the second agent. In our facility, the XDR rates in the K. pneumoniae isolates included in our study was found to be 39.3%, which is exuberantly high compared to E. coli. This might be the reason why the combination regimens were not found to be significantly effective with respect to the monotherapy, as compared to that in E. coli. 17, 18

Combination agent antibiogram for Enterobacter cloacae complex

For Enterobacter cloacae complex (Table 3), the only combination regimen that was found to have statistically significant increase in S% was ‘imipenem or ertapenem plus gentamicin’. MDR rate in E. cloacae was found to be low (42.9%) in our center as compared to E. coli (73.7%) and K. pneumoniae (71.6%), therefore, coverage with monotherapy is often found to be effective and combination regimen may not be required. 16 As per the statistical table for increase in S% provided in CLSI M39, when the sample size is low, only a higher increase in S% makes it statistically significant, while with a high sample size, even a lower increase in S% makes it statistically significant. As in the present study, the sample size for E. cloacae complex was relatively low (i.e. 85), as compared to E. coli (644) and K. pneumoniae (389), this might be the reason that lower number of combination regimens with statistically significant increase in S% were observed in E. cloacae complex, as compared to E. coli and K. pneumoniae. 2, 17, 18

Combination agent antibiogram for P. aeruginosa and A. baumannii complex

Unfortunately, our study did not find a single combination regimen useful for P. aeruginosa (Table 4), and A. baumannii complex (Table 5), as none of the combination regimens were found to have statistically significant S% (p ≤0.05) compared to the S% of the respective individual agents. 4 This is because of the occurrence of a high rate of drug resistance in P. aeruginosa (26% XDR), and A. baumannii complex (55.9% XDR), in our facility. XDR isolates are resistant to most of the antimicrobial classes and therefore will not be covered even by combination regimens. Therefore, use of combination regimen does not give any added advantage over monotherapy for suspected P. aeruginosa and A. baumannii complex infection, and therefore should not be used empirically. 2, 8, 14

Combination agent antibiogram at GNB-level

When prepared the combination agent antibiogram at GNB-level, based on the joined data of five major gram-negative pathogens, i.e. E. coli, K. pneumoniae, E. cloacae, P. aeruginosa and A. baumannii. It was observed that several combination regimens were found to have statistically significant S% (p= <0.05) compared to the S% of the respective individual agents. The higher statistical significance in GNB analysis is mainly attributed to the higher number of isolates (1640) subjected to analysis. When the isolate number is very high (e.g. >1000), even a smaller increase in S% is found to be statistically significant. 2. However, for a smaller number of isolates, only a higher increase in S % makes it statistically significant. 2 Even if antimicrobial empirical combinations which had statistically significant S% were not found for P. aeruginosa and A. baumannii complex, still the combination agents demonstrated relatively better S% than individual agents and hence would still help as combination empirical choices as compared to individual monotherapy.

The gentamicin combination therapy was found to be statistically significant for majority of β-lactams, except for ceftriaxone and amoxicillin clavulanate. Ceftriaxone and amoxicillin clavulanate are in general not a good empirical agent for suspected gram-negative infection as several major gram-negative pathogens like Ps and Ac are IR to these agents. Gentamicin monotherapy is as such not recommended for bloodstream infections. However, in the current study we found that it gives an added advantage in S% when combined with combination therapy fourth generation cephems, higher BLBI and carbapenems; hence gentamicin combination therapy is a good choice for empirical therapy. 14

Cotrimoxazole combination therapy with β-lactams is rarely a preferred combination regimen as empirical therapy for suspected gram-negative sepsis. However, in the current study, we have found a significant improvement in S% when cotrimoxazole is combined with most of the β-lactams agents. Therefore, we urge to keep cotrimoxazole combination therapy with β-lactams as empirical option for suspected gram-negative infections. Our study found ciprofloxacin combination therapy with β-lactams is a less useful regimen as in most of the instances, the increase in S% in not significant. 1, 8, 15

Combination agent antibiogram can be particularly useful in settings where — (i) high-risk patients like immunocompromised, transplant recipients, etc. are common; (ii) S% of individual agents are less than optimal to be a good empirical choice as monotherapy; (iii) S% of different organisms against various antimicrobial agents grossly vary from each other and therefore a combination therapy is preferred to provide better coverage. 16 However, the S% obtained in combination agent antibiogram is not derived from any in vitro synergy testing, and the potential synergistic or antagonistic interactions between the individual agents are unknown. More so, higher S% to combination agents does not indicate that the two agents given together is necessarily more effective than either of the agent given as monotherapy. 4 It only indicates that the chance of the suspected organism being susceptible is higher when given in combination than as monotherapy, and therefore there is a higher likelihood of therapeutic success. Such kind of combination regimen is only meant for empirical therapy and must be de-escalated (monotherapy) to the most possible narrow-spectrum agent once the culture and AST report are available. 7, 17, 18, 19