Bacteroides fragilis has been reported for being the cause of a wide range of infections  and is often associated with drug resistance . Most patients receive empiric treatment as anaerobic susceptibility test results are unavailable or delayed . Determining MICs by the recommended agar dilution method gives information regarding resistance trends over time . Due to recent developments in molecular microbiology, there are reported PCR methods to rapidly identify resistance by genetic determinants.
Though, phenotypic testing is more reliable, phenotypic resistance in anaerobes may not always positively correlate with the genotypic findings. This is mainly due to the multiple resistant markers and the varying mechanisms of resistance, including the involvement of IS elements for gene regulation [16, 17, 18, 19]. This study is the first of its kind in India to correlate phenotypic and genotypic resistance mechanism in B. fragilis, in order to define the most plausible way of treating B. fragilis infections.
Antimicrobial resistance in B. fragilis is mediated by various mechanisms including the production of drug modifying enzymes, efflux of drug or inactivation of drug through the expression of resistance genes [20, 21]. This study utilised nim (metronidazole), ermF (clindamycin) and cfiA (meropenem) genes as molecular markers.
Phenotypic results revealed that 41% of B. fragilis were resistant to metronidazole, the commonest drug used for treating anaerobic infections . This is comparable with an Indian study by Sethi et al.  which reported 41% resistance, and Sood et al.(2021) who reported 32.6%. In contrast, 100% susceptibility was reported from Korea by Byun et al and Veloo et al from The Netherlands [24, 25]. The increased prevalence of metronidazole resistance (41%) reported in this study could be region specific, as antimicrobial resistance varies among different regions.
In this study, 50% of the metronidazole resistant isolates carried the nim gene, which agrees with studies by Vishwanath et al  and Gal et al . To date, 11 variants of nim genes (nimA to nimK) have been reported . It has been recorded that both phenotypic metronidazole resistance and nim gene positivity were 50% and 48%, respectively [26, 27]. In contrast, Akhi et al (Iran) showed 0% prevalence of the nim gene in resistant isolates. This might be due to other mechanisms such as overexpression of the multidrug efflux pump, overexpression of recA or deficiency of feoAB . Interestingly, in a study from Gal and Brazier (2004), 14% of susceptible isolates were reported to carry the nim gene. Silent nim genes might be due to the absence of an IS element promoter region . There is strong evidence that these IS elements carry regulatory signals for the expression of nim genes . Studies have shown that silent nim genes can be expressed when isolates carrying them are exposed to metronidazole for longer periods of time .
The present study reports 45% phenotypic resistance for clindamycin. This high percentage resistance to clindamycin may be due to its widespread use in the treatment of anaerobic infections, in intra-abdominal, pelvic, lower respiratory, bone, and skin and soft tissue infections [32, 33]. The study data is comparable to the observation by Vishwanath et al  which reported resistance rate of 38%. Further, 74% phenotypically resistant isolates were positive for ermF. The ermF gene was present only in phenotypically resistant isolates. Similar findings were reported by Kouhsari et al  and Eitel et al  which showed prevalence rates of 76% (206/364) and 74% (23/31) for ermF in resistant isolates. It has been observed that the ermF gene is frequently present on conjugative transposons. In this study, 9% of B. fragilis isolates were not carrying ermF. Resistance in these isolates might be due to the other mechanisms conferred by linA, other erm genes such as ermG/ermS, msrA or efflux pump mechanism .
Meropenem has been reported to have good coverage against anaerobes . In this study, 15% of B. fragilis were resistant to meropenem. Similar results have been reported by Jamal et al  and Wang et al  where they found 17% and 19% resistance respectively whilst slightly reduced rates of resistance were reported Wybo et al  at 10%. Contrary to this, Vishwanath et al  and Sood et al.  reported the nearly complete absence of resistance for meropenem.
Overall the prevalence of the cfiA gene in this study was 37%. All isolates (n = 8) with high MICs (≥ 16 µg/ml) for meropenem were carrying resistance gene. Similar result was reported by Gao et al in China  and Wybo et al . However, 21% of cfiA positive isolates displayed lower levels of meropenem MICs. This might be due to the proven fact that cfiA without an upstream insertional sequence (IS) will display lower MICs as opposed to high MICs for cfiAs with an upstream IS, which provides a strong promoter for cfiA expression [18, 19]. All cfiA negative isolates were phenotypically susceptible. Based on this observation, it could be interpreted that cfiA negative isolates will mostly be phenotypically susceptible.
In most cases, meropenem is used for treating B. fragilis along with metronidazole. Though, both the antimicrobials have good coverage for anaerobes, use of redundant antimicrobials will further add to the existing resistance and complicate antimicrobial stewardship. This has been observed from our study centre, where unnecessary double coverage for anaerobes were offered by second antibiotics (19.8%) .
Beta-lactam, beta-lactamase inhibitor combinations (BL-BLICs) are frequently used for mixed infections (aerobic-anaerobic) as they have a wide range of activity against the majority of anaerobic bacteria . Interestingly, studies from Scandinavian countries and one report from India have presented increased resistance to piperacillin/tazobactam in comparison to meropenem [45, 46, 41]. Contrary to these reports, the present study showed B. fragilis was highly susceptible to piperacillin/tazobactam as only 2 out of 51 isolates had high MICs. Our result was concordant with other studies where resistance reported for piperacillin/tazobactam varied from 0-5% [8, 12, 47, 45], and was less than for meropenem, for example Snydman et al.  studied around 2722 B. fragilis isolates over 8 years.
Observations from this study and the corresponding literature reveal that to decide empirical therapy for B. fragilis, the local site should generate AMR surveillance data. This aggregated data would help to recommend appropriate anaerobic cover for empiric therapy and would avoid the use of redundant antibiotics thereby saving “watch” antibiotics such as meropenem.