Anaerobes are well known to be an important part of the normal human intestinal, vaginal, oral, and skin microbiota [1]. Anaerobic bacteria are also opportunistic pathogens that could be involved in various types of human infections in association with aerobic bacteria, such as brain abscesses, intraabdominal, skin, or pelvic infections. They can also cause monomicrobial infections such as bacteremia, deep tissue infections, and bone and joint infections. These infections are associated with severe morbidity and a high rate of mortality ^[2][3][2,3]. In addition to the well-known anaerobes such as Bacteroides spp. or Clostridium perfringens, new genera and species are regularly described through improvements in culture and identification techniques and implicated in severe human infections. Clinically relevant Gram-negative anaerobic bacteria include Bacteroides fragilis group, Prevotella spp., Fusobacterium spp., and Veillonella spp. [4]. The main Gram-positive bacilli isolated from clinical samples are Cutibacterium spp., especially C. acnes (formerly known as Propionibacterium acnes), which is involved in chronic bone and joint infections. Additionally, Clostridium spp. (apart from Clostridioides difficile) are responsible for many types of severe infections, such as gas gangrene. Actinomyces spp., isolated in deep tissue infections, is also noteworthy ^[5][6][7][5,6,7]. Finegoldia magna and Parvimonas micra are the two most isolated Gram-positive anaerobic cocci (GPAC).

Due to technical and financial constraints associated with the identification and culture of anaerobic bacteria, microbiology and antibiotic susceptibility testing (AST) of anaerobes isolates are rarely routinely performed in clinical microbiology laboratories [1]. Therefore, the treatment of anaerobic infections has long been empirical, which has led to therapeutic failures and the emergence of resistance [8]. Over the last two decades, a growing number of studies around the world have focused on describing the epidemiology of resistance in anaerobic bacteria, with a worldwide increase despite differences between countries [9]. However, AST has been performed differently in laboratories according to countries, most of the time without following CLSI and EUCAST methods [4]. For example, Asian laboratories mostly used micro-dilution-based techniques, while the majority of laboratories in Europe and the US employed gradient diffusion strips or agar dilution, which is the reference method. Breakpoints for AST interpretation also sometimes differ between EUCAST and CLSI guidelines. This diversity of practices could lead to great variability in results and make studies difficult to compare. Moreover, there is a scarcity of recent comprehensive epidemiological data with a significant number of strains.

β-lactam antibiotics are considered the drugs of choice in the management of anaerobic infections. This is due to their broad spectrum of activity, low toxicity, and continued efficacy against almost all anaerobic species, especially when used in combination with β-lactam/β-lactamase inhibitors (BL/BLI) or carbapenems. Among the anaerobes, Bacteroides and Parabacteroides species are of greatest concern considering their higher resistance rates. In the 1990s, in Europe, a large multicenter study in 15 different countries reported a prevalence of 1%, 3%, and 0.3% for amoxicillin/clavulanate (AMC), cefoxitin, and imipenem among the B. fragilis group (n = 1289) [10]. Over the past 20 years, nearly 10% have become resistant to AMC and piperacillin/tazobactam (PTZ), while 17% and 1% are resistant to cefoxitin and carbapenems, respectively [11]. In Canada, an increase in resistance to AMC was also observed between 1992 and 2010–2011 (from 0.8% to 6.2%), while a slight decrease in cefoxitin resistance was reported (26% vs. 15%), potentially related to reduced use of cefoxitin [12]. In the US, an increase in the resistance rate of ampicillin/sulbactam (from 4% to 6%) and PTZ (from 2% to 7%) was observed among Bacteroides and Parabacteroides isolates between 2007–2009 and 2010–2012 ^[13][14][13,14]. An increase in carbapenem resistance was also reported, such as in Poland, where imipenem resistance increased between 2007–2012 and 2013–2017 in the Bacteroides fragilis group (0.5% to 2.2%), especially in non-fragilis Bacteroides (1.4% to 3.7%) [15]. A decrease in susceptibility to meropenem among the B. fragilis group was also reported in Japan between 2010 and 2018–2019 (98% to 90%) [16]. In recent studies, AMC resistance ranges from 2 to 9% in B. fragilis, except in Spain, where higher rates were reported (29%) ^{[15][17][18][19][20]}[15,17,18,19,20]. PTZ resistance remains lower, with a resistance rate varying between 1 and 3%, while the resistance rate reaches 5% in Korea and Greece ^{[17][21][22][23]}[17,21,22,23]. Higher AMC and PTZ resistance rates were reported in the B. fragilis group excluding B. fragilis, especially in B. thetaiotaomicron and Phocaeicola vulgatus ^[17][18][23][17,18,23]. In B. fragilis, the rate of resistance to carbapenems ranges from 0 to 5% for imipenem and from 2 to 5% for meropenem ^{[15][17][18][19][20][21][23][24]}[15,17,18,19,20,21,23,24].

Among Prevotella spp., a slight increase in penicillin-resistant isolates was described in Belgium between 1993–1994 (52%) and 2011–2012 (65%), while a higher increase was observed in Bulgaria between 2003–2004 (15%) and 2007–2009 (61%) ^[13][25][13,25]. In recent studies, most isolates are resistant to penicillin, with a prevalence of 60–80% in European countries, except for Germany, where Wolf et al. noted a lower rate (36%) ^{[13][19][26][27]}[13,19,26,27]. Over the world, resistance rates were similar in the US (65%) and Canada (63.5%), while a higher resistance rate was noted in Korea (91%) ^[21][22][28][21,22,28]. However, most strains remain susceptible to BL/BLI combinations and carbapenems, except in Canada and Spain, where a few strains resistant to PTZ and AMC were reported ^[19][22][19,22]. Among Fusobacterium spp., the rate of resistance to penicillin range between 5 and 17%, while a higher prevalence was reported in Ireland (50%) ^{[17][19][21][22][27]}[17,19,21,22,27]. BL/BLI combination and carbapenems still have excellent activity and only some resistant isolates have been sporadically reported ^[17][19][22][17,19,22]. Veillonella spp. have high rates of penicillin resistance, ranging from 29 to 55%, except in Korea, where Buyn et al. reported 100% of resistance (n = 11), in contrast to Ali et al., who reported no resistant strains in Ireland (n = 9) ^{[17][19][21][22][27][28]}[17,19,21,22,27,28]. Among Veillonella spp., high levels of resistance to TZP (MIC ≥ 128 mg/L) were observed ^[17][21][22][17,21,22].

In Gram-positive anaerobic bacteria, most isolates of Propionibacterium spp., Cutibacterium spp., Finegoldia magna, Peptoniphilus spp., Anaerococcus spp., and Parvimonas micra are susceptible to β-lactams ^{[18][19][22][27][28][29]}[18,19,22,27,28,29]. Penicillin resistance in Peptostreptococcus anaerobius appears to be more common and ranges from 5 to 25%, although a higher rate of resistance to ampicillin has been reported in France by Guérin et al. (55%, 5/11) ^[29][30][31][29,30,31]. In Clostridium spp., penicillin resistance is higher, varying between 11–30% worldwide, and only a few strains are resistant to BL/BLI combinations and carbapenems. C. perfringens exhibits a lower resistance rate, ranging from 0 to 5% ^{[13][21][22][22][27][28][28][30][31]}[13,21,22,22,27,28,28,30,31]. In Eggerthella lenta, resistance to penicillin was commonly recovered (13–98%), while low susceptibility levels have been observed for TZP, with MIC₅₀ ranging between 16 and 32 mg/L ^[21][22][32][21,22,32]. The ranges of MIC, MIC_50, and MIC₉₀ are synthetized for AMC (Gram-negative) and penicillin (Gram-positive) in Table 1 and Table 2.

Carbapenem resistance in B. fragilis is primarily promoted by the class-B metallo-carbapenemase CfiA, encoded by a chromosomal gene recovered in some strains. In terms of intra-species diversity, B. fragilis can be classified into two subgroups based on the presence or absence of the cfiA and cepA genes. These subgroups are referred to as division I (cfiA-) and division II (cfiA+). Subgroup division is achieved by DNA–DNA hybridization and ribotyping, while detection is now available by matrix-assisted laser desorption/ionization time-of-flight mass spectroscopy ^[39][40][41][39,40,41]. In a retrospective study, Ferløv-Schwensen et al. observed an increase in division II bacteroides among clinical isolates (2.8% vs. 7.8%) between 1973–1991 and 2002–2015, potentially due to the overuse of carbapenems [42]. Recent studies have reported a similar prevalence of cfiA in Europe, ranging between 8 and 16% ^[24][34][43][24,34,43]. In Asia, close rates have been reported, with a prevalence of 15% and 22% in Japan and China, respectively ^[33][44][33,44]. Curiously, a higher prevalence of division II (cfiA+) has been reported in bloodstream infections compared to other clinical isolates [43]. The gene cfiA is not always correlated with phenotypic resistance related to low-level expression. Insertion sequences upstream of the gene, mainly belonging to the IS1380 family, are the cornerstone to induce their overexpression, which leads to phenotypic resistance to β-lactams ^[38][45][46][38,45,46]. It should be noted that the use of meropenem (not imipenem) +/− EDTA allows the detection of cfiA+ strains with low-level expression [47]. In the non-fragilis Bacteroides group, only cfxA is generally recovered, and an unknown mechanism provides resistance to β-lactams in non-cfxA isolates [34]. However, in recent studies, Soki et al. identified in B. xylanisolvens, a non-fragilis Bacteroides species, the crxA gene coding for a metallo-B-carbapenemase close to cfiA that confers resistance to carbapenem, while Wallace et al. identified putative class A β-lactamases among non-fragilis Bacteroides ^[45][48][45,48]. In Prevotella spp., cfxA variants (cfxA2, cfxA3, cfxA6, and cfxA7) are associated with ampicillin resistance, with a prevalence ranging between 51–78% ^{[34][49][50][51]}[34,49,50,51]. The cfxA2 gene differs from the cfxA of P. vulgatus by an amino acid change, while cfxA3, cfxA6, and cfxA7 differ by two [49].

β-lactamases in other anaerobes are less studied, but penicillinases are outlined, mainly by phenotypic approaches in Fusobacterium spp., Porphyromonas spp., and Clostridium spp. ^{[52][53][54][55]}[52,53,54,55]. Target modification by alteration of penicillin-binding proteins (PBPs) promotes cefoxitin resistance in the Bacteroides fragilis group. In fact, the modification of PBPA or PBP3 appears to play a greater role than hydrolysis by CfxA ^[37][56][57][37,56,57]. A decrease in imipenem susceptibility was also associated with PBP2Bfr modifications [58]. In C. perfringens, PBP alterations following β-lactams exposure result in decreased affinity of Β-lactams for PBP1 but an overproduction of PBP6 related to phenotypic resistance to penicillin G and ceftriaxone ^[59][60][59,60].

In Veillonella spp., PBP modification leads to high-level resistance to TZP (MIC >128 mg/L) in β-lactamase-negative isolates, whereas ampicillin remains active (MIC = 0.5–4 mg/L) due to a retained affinity for PBP [61]. In B. fragilis, derepression of bmeABC coding for a RND efflux pump can trigger the extrusion of ampicillin, cefoperazone, and cefoxitin [62]. Resistance induced by porin loss remains poorly studied, while in B. thetaiotaomicron, it has been suggested that resistance to AMC may be related to a defect in the expression or absence of a porin [63]. Moreover, loss of porin associated with PBP alteration may be co-induced following cefoxitin exposure leading to ampicillin and cephalosporin resistance in B. thetaiotaomicron [57].