β-Lactam Combinations That Exhibit Synergy against Mycobacteroides abscessus Clinical Isolates

Mycobacteroides abscessus (Mab) is an opportunistic environmental pathogen that can cause chronic pulmonary disease in the setting of structural lung conditions such as bronchiectasis, chronic obstructive pulmonary disease, and cystic fibrosis. These infections are often incurable and associated with rapid lung function decline. Mab is naturally resistant to most of the antibiotics available today, and current treatment guidelines require at least 1 year of daily multidrug therapy, which is often ineffective and is associated with significant toxicities. β-Lactams are the most widely used class of antibiotics and have a demonstrated record of safety and tolerability.

ABSTRACT

Mycobacteroides abscessus (Mab) is an opportunistic environmental pathogen that can cause chronic pulmonary disease in the setting of structural lung conditions such as bronchiectasis, chronic obstructive pulmonary disease, and cystic fibrosis. These infections are often incurable and associated with rapid lung function decline. Mab is naturally resistant to most of the antibiotics available today, and current treatment guidelines require at least 1 year of daily multidrug therapy, which is often ineffective and is associated with significant toxicities. β-Lactams are the most widely used class of antibiotics and have a demonstrated record of safety and tolerability. Here, using a panel of recent clinical isolates of Mab, we evaluated the in vitro activities of dual-β-lactam combinations to identify new treatments with the potential to treat infections arising from a wide range of Mab strains. The Mab clinical isolates were heterogeneous, as reflected by the diversity of their genomes and differences in their susceptibilities to various drugs. Cefoxitin and imipenem are currently the only two β-lactams included in the guidelines for treating Mab disease, yet they are not used concurrently in clinical practice. However, this dual-β-lactam combination exhibited synergy against 100% of the isolates examined (n = 21). Equally surprising is the finding that the combination of two carbapenems, doripenem and imipenem, exhibited synergy against the majority of Mab isolates. In the setting of multidrug-resistant Mab disease with few therapeutic options, these combinations may offer viable immediate treatment options with efficacy against the broad spectrum of Mab strains infecting patients today.

INTRODUCTION

Patients with chronic lung diseases such as cystic fibrosis (CF) and bronchiectasis are at high risk for development of recurrent pulmonary infections, largely due to poor clearance of respiratory secretions resulting in persistent colonization with pathogenic bacteria (1–4). Some species of nontuberculous mycobacteria (NTM) are capable of causing invasive infections in this patient population. The prevalence of NTM pulmonary disease in the United States has steadily risen over the past several decades and currently far outweighs the prevalence of tuberculosis in this country (5, 6).

Mycobacteroides abscessus (Mab) (formerly Mycobacterium abscessus) (7) is considered one of the most virulent NTMs, and Mab lung disease is associated with significant decline in pulmonary function (8–10), with cure rates as low as 25 to 40% with antibiotic treatment alone (11, 12). Mab has been described as an “antibiotic nightmare” and “an environmental bacterium turned clinical nightmare” because it is intrinsically resistant to several antibiotic classes (13–16), with acquired resistance further limiting therapeutic options (17). Additionally, Mab subspecies abscessus and bolletii readily develop resistance to macrolide antibiotics upon exposure to these drugs (18), which are considered the cornerstone of treatment of Mab lung disease (8, 19). Current treatment guidelines for pulmonary Mab infections include at least 12 to 18 months of multidrug therapy, with several agents requiring intravenous administration and causing substantial cytotoxicity (20, 21). There are no FDA-approved antibiotics for Mab disease based on clinical trials, and current treatment recommendations were developed from expert consensus and empirical experience (22). β-Lactams are among the antibiotics included in these guidelines and have proved to be useful in treating this disease (22–24).

β-Lactams represent the most widely used class of antibiotics for treatment of bacterial infections (25) and are generally well tolerated, with minimal comparative cytotoxicity. β-Lactams exert their activity via inhibition of the synthesis of peptidoglycan, the exoskeleton of the bacterial cell wall that is essential for cellular growth and survival (26). The final step of peptidoglycan synthesis involves polymerization of disaccharide-peptide monomers to generate a single macromolecule that encapsulates the plasma membrane (27). Unlike most bacteria whose peptidoglycan synthesis requires only d,d-transpeptidases (also known as penicillin-binding proteins), in Mab the majority of monomer polymerization is undertaken by nonclassical peptidases, the l,d-transpeptidases (28). An initial survey of the Mab genome has identified five putative l,d-transpeptidase-encoding genes (29). These l,d-transpeptidases are differentially susceptible to β-lactam subclasses, with carbapenems exhibiting the strongest inhibition, followed by cephalosporins, and only a select few penicillins exhibiting limited inhibition of these enzymes (30–32). Based on these observations, the hypothesis that synergistic activity is likely from a combination of β-lactams, one that optimally inhibits d,d-transpeptidases and another that optimally inhibits l,d-transpeptidases, was proposed (32, 33). Subsequent findings using in vitro (31, 34, 35) and in vivo (36) assessments have supported this hypothesis.

These initial proof-of-concept studies were performed using the Mab reference strain (ATCC 19977). Mab strains isolated from patients exhibit extensive genotypic and phenotypic heterogeneity, resulting in widely variable antimicrobial susceptibility profiles and clinical responses to treatment regimens (37–39); this often necessitates tailoring of treatment regimens based on in vitro drug susceptibility testing. Due to inherent diversity among Mab clinical isolates, the question of whether studies using any single Mab strain can be predictive of treatment success in the clinic is raised. Therefore, preclinical studies including multiple isolates are warranted to better represent the diversity of strains with which patients present. Furthermore, the ATCC 19977 reference strain that was used in the initial proof-of-concept studies was isolated in 1953 from a synovial fluid culture (40) and may not be the most relevant reference to represent a patient population presenting primarily with pulmonary Mab infections occurring in an era of ubiquitous antibiotic use that has resulted in extensive antimicrobial resistance.

Therefore, inclusion of multiple diverse Mab clinical isolates in preclinical studies is rigorous and is likely to generate data that are more representative and predictive of clinical outcomes. Because several β-lactam combinations exhibit synergy against Mab reference strain ATCC 19977 in vitro (34) and in vivo (36), β-lactam combinations present an untapped resource whose further preclinical evaluation could lead to repurposing of these drugs to treat Mab infections. A prior investigation identified synergy between ceftazidime and ceftaroline or imipenem against a panel of Mab clinical isolates (35). Here, we have assessed the activities of 13 dual-β-lactam combinations that have not been previously studied against 21 Mab strains that were recently isolated from pulmonary exudates from CF patients (41). These β-lactams are commercially available and thus have the potential to serve as novel therapeutic options pending efficacy trials. The main objective of this study was to identify dual-β-lactam combinations that exhibit synergy against the largest number of recent Mab clinical isolates. Such combinations are likely to harbor greater potential for clinical efficacy against the broad spectrum of Mab strains infecting patients today.

RESULTS

Mab clinical isolates and genomic diversity.

The clinical strains were obtained through the Johns Hopkins Clinical Mycobacteriology Laboratory and were collected from lung exudate specimens from CF patients at the Johns Hopkins Hospital over the past 15 years (41). The whole-genome sequences of 21 clinical isolates were determined, and six genes were used as genetic markers to identify the isolates to the subspecies level as described (42, 43). This assessment identified 12 strains as belonging to Mab subspecies abscessus and the remaining 9 as belonging to Mab subspecies massiliense (Fig. 1). To verify these subspecies identification results, we also queried the sequence for the erm(41) gene for all strains and observed that the identification of strains as Mab subsp. massiliense by the phylogenetic approach was correct, based on the previously described truncation of the gene (43, 44). Lastly, a recent publication reported that subspecies identification can be achieved by using a single gene, gnD (MAB_0003) (45). We generated a phylogenetic tree using the same approach as described above but with gnD as the marker, and we found this tree to be nearly identical to the tree built with six genes (see Fig. S1 in the supplemental material). Strains that clustered together in the six-gene approach also clustered together in this single-gene approach, successfully validating this method for subspecies identification of Mab strains.

FIG 1.

Phylogenetic tree for clinical Mab isolates based on six marker genes. The genes used were hsp65 (MAB_0650), secA (MAB_2397), rpoB (MAB_3869c), polC (MAB_2696c), hoa (MAB_0626), and ftsZ (MAB_2009). We used the Geneious Tree Builder with the Tamura-Nei genetic distance model and the UPGMA tree-building method bootstrapped to 1,000 iterations. Of the 21 clinical strains, 12 strains were revealed to be Mab subspecies abscessus, while 9 strains were Mab subspecies massiliense.

Based on the phylogenetic tree illustrated in Fig. 1 and Fig. S1, the clinical isolates in this study are genetically diverse, even those that clustered together within the tree. Due to the limitations of short-read sequencing, we mapped the reads to the reference strain ATCC 19977 to compare the patterns of the locations at which each clinical isolate’s reads would map to the reference strain. We found that there were many different patterns, indicating our clinical isolates to be diverse. For further verification, we randomly selected 12 strains for long-read sequencing (PacBio) to attempt to generate de novo a circularized genome for these isolates. Results from the whole-genome alignment showed many insertions/deletions, inversions, and even shuffling of whole sections of the genomes, compared to each other (Fig.S2 and S3).

MICs of β-lactams against Mab clinical isolates.

We began by determining the MICs of 12 β-lactams against all 21 Mab clinical isolates to establish baseline susceptibility (Table 1). The β-lactams evaluated in this study included the cephalosporins cefuroxime, cefadroxil, cefdinir, cefditoren, cefpodoxime, and cefoxitin, the carbapenems ertapenem, imipenem, doripenem, biapenem, and tebipenem, and the penem faropenem. The majority of cephalosporins exhibited high baseline MICs of 256 to 512 μg/ml, with the exception of cefoxitin and cefdinir, which had more variable MICs, ranging from 32 to 256 μg/ml. The MICs of ertapenem, tebipenem, and faropenem were also consistently high, at 256 to 512 μg/ml, for the majority of isolates. Imipenem, doripenem, and biapenem, in contrast, tended to have lower MICs but with high levels of variability between strains, ranging from 8 μg/ml to 256 μg/ml (Table 1). Compared to the Mab reference strain ATCC 19977 (34), the MICs of most cephalosporins, carbapenems, and penem against the clinical isolates were higher, with a few exceptions (Table S1). In addition, the clinical isolates had distinct MIC profiles among themselves. Any two Mab isolates differed in MICs of at least three (for instance, M9501 versus M9502) and to up to seven (M9501 versus M9529) β-lactams. Based on these results, we established that the Mab clinical isolates were intrinsically diverse in their susceptibilities to β-lactams.

TABLE 1.

MICs of each individual β-lactam tested against 21 clinical Mab strains

Mab clinical strain (subspecies)	MIC (μg/ml) ofa:
	Cephalosporins						Carbapenems/penem
	CXM	CFR	CDR	CDN	CPD	FOX	ETP	IPM	DOR	BIA	FAR	TEB
M9501 (abscessus)	512	512	128	512	512	64	512	8	32	16	256	256
M9502 (massiliense)	512	512	64	512	512	128	512	8	16	16	256	256
M9503 (abscessus)	256	512	64	256	512	64	512	8	8	8	256	256
M9504 (massiliense)	256	512	64	512	512	128	256	16	16	16	256	512
M9505 (massiliense)	256	512	64	256	256	32	256	16	16	8	512	256
M9507 (abscessus)	512	512	256	512	512	256	512	32	128	128	512	512
M9509 (massiliense)	256	512	64	512	512	32	256	16	16	16	64	128
M9510 (massiliense)	512	512	256	512	512	256	512	64	128	128	512	256
M9513 (abscessus)	512	512	64	512	512	64	512	16	32	16	512	512
M9514 (massiliense)	512	512	64	512	512	128	512	8	32	32	256	256
M9515 (massiliense)	512	512	256	512	512	256	512	32	256	128	512	512
M9517 (massiliense)	512	512	128	512	512	64	512	16	16	16	256	512
M9521 (massiliense)	512	512	128	512	512	32	256	8	16	16	128	256
M9522 (abscessus)	512	512	256	512	512	64	512	16	32	32	512	512
M9525 (abscessus)	512	512	256	512	512	128	512	32	64	64	512	512
M9526 (abscessus)	512	512	256	512	512	64	512	16	32	32	512	512
M9527 (abscessus)	512	512	128	512	512	128	512	16	64	32	512	512
M9528 (abscessus)	512	512	256	512	512	128	512	32	64	32	512	512
M9529 (abscessus)	512	512	256	512	512	128	512	256	256	128	512	512
M9530 (abscessus)	512	512	128	512	512	64	512	16	64	16	512	512
M9531 (abscessus)	512	256	64	512	512	64	512	16	64	32	256	512

^aBased on CLSI-defined breakpoints, the following extrapolations can be made: cephalosporins: ≤16 μg/ml, susceptible; ≤64 μg/ml, intermediately susceptible; carbapenems: ≤4 μg/ml, susceptible; ≤8 μg/ml, intermediately susceptible. CXM, cefuroxime; CFR, cefadroxil; CDR, cefdinir; CDN, cefditoren; CPD, cefpodoxime; FOX, cefoxitin; ETP, ertapenem; IPM, imipenem; DOR, doripenem; BIA, biapenem; FAR, faropenem; TEB, tebipenem.

Antibiotic synergy against clinical isolates.

Based on a prior study that demonstrated synergy of dual-β-lactam combinations against Mab reference strain ATCC 19977 (34), we determined the activity of 13 distinct dual-β-lactam combinations against the panel of 21 distinct Mab clinical isolates. These combinations were cefoxitin and imipenem, cefuroxime and imipenem, doripenem and imipenem, biapenem and imipenem, cefdinir and imipenem, faropenem and imipenem, cefadroxil and tebipenem, ertapenem and imipenem, cefditoren and imipenem, cefpodoxime and imipenem, cefditoren and tebipenem, cefuroxime and cefditoren, and cefditoren and biapenem. Each combination was tested against the 21 clinical Mab strains in vitro using a checkerboard titration assay, as described in Materials and Methods. The checkerboard assay allows for calculation of a fractional inhibitory concentration index (FICI) for each drug pair against each isolate. The FICI is a mathematical representation of the degree to which each drug in a combination contributes to synergy (46, 47). A stringent interpretation of FICI values was used, with FICIs of ≤0.5 indicating synergy, FICIs of >0.5 to <4 indifference, and FICIs of >4 antagonism (48). The fractional inhibitory concentration (FIC) was used to extrapolate the decrease in the MIC of each drug in a pair as a result of synergy. As may be expected from the MIC results, FICIs of the dual-β-lactam combinations were highly varied among clinical strains (Table 2). Similarly, the number of strains against which each drug combination exhibited synergy also varied, ranging from 14% to 100% of isolates. For example, the combination of cefditoren and biapenem exhibited synergy against only 3/21 isolates, while the combination of cefoxitin and imipenem was synergistic against all 21 strains; this was the only combination that exhibited synergy against 100% of isolates. In general, 7/13 drug pairs exhibited synergy against >50% of the clinical strains (Table 3).

TABLE 2.

Comparison of MICs of single drugs and MICs resulting from dual-β-lactam synergy

Mab clinical strain (subspecies)a	Data forb:
	FOX/IPM	CXM/IPM	DOR/IPM	BIA/IPM	CDR/IPM	FAR/IPM	CFR/TEB	ETP/IPM	CDN/IPM	CPD/IPM	CDN/TEB	CXM/CDN	CDN/BIA
M9501 (abscessus)
MIC (μg/ml) of each drug	64/8	512/8	32/8				512/512		512/8
Extrapolated MIC (μg/ml)	4/1	112/2	7/2				128/128		68/2
FICI	0.292	0.414	0.438				0.500		0.328
M9502 (massiliense)
MIC (μg/ml) of each drug	128/8	512/8					512/256		512/8		512/256	512/512
Extrapolated MIC (μg/ml)	28/2	112/1					96/64		112/2		76/48	78/43
FICI	0.422	0.321					0.438		0.422		0.336	0.235
M9503 (abscessus)
MIC (μg/ml) of each drug	64/8	256/8	8/8	8/8	64/8	256/8	512/256	512/8	256/8	512/8		256/256
Extrapolated MIC (μg/ml)	14/2	51/2	2/2	2/2	14/2	56/1	112/56	83/1	64/2	112/2		56/50
FICI	0.438	0.357	0.422	0.438	0.422	0.360	0.438	0.287	0.500	0.422		0.414
M9504 (massiliense)
MIC (μg/ml) of each drug	64/16	256/16	16/16	8/16	32/16	256/16	512/512	512/16	256/16	512/16
Extrapolated MIC (μg/ml)	10/3	67/2	3/3	3/4	8/3	48/3	86/86	80/2	83/2	112/3
FICI	0.315	0.267	0.360	0.438	0.438	0.360	0.334	0.352	0.297	0.360
M9505 (massiliense)
MIC (μg/ml) of each drug	64/16	256/16	16/16			512/16	256/256	256/16			256/256
Extrapolated MIC (μg/ml)	12/4	40/2	4/4			82/3	67/43	72/4			48/64
FICI	0.438	0.321	0.500			0.286	0.334	0.422			0.375
M9507 (abscessus)
MIC (μg/ml) of each drug	256/32	512/32	128/32	128/32		512/32
Extrapolated MIC (μg/ml)	37/6	128/6	26/6	22/6		128/8
FICI	0.332	0.438	0.378	0.360		0.500
M9509 (massiliense)
MIC (μg/ml) of each drug	32/16	256/16	16/16	16/16	64/16	64/16		256/16	512/16	512/16	512/128
Extrapolated MIC (μg/ml)	4/2	23/7	3/2	3/2	11/2	16/1		48/1	112/1	96/2	98/28
FICI	0.240	0.248	0.320	0.320	0.301	0.290		0.258	0.301	0.289	0.410
M9510 (massiliense)
MIC (μg/ml) of each drug	256/64		128/64	128/64	256/64								512/128
Extrapolated MIC (μg/ml)	43/10		32/10	28/8	56/9								128/32
FICI	0.326		0.407	0.344	0.348								0.500
M9513 (abscessus)
MIC (μg/ml) of each drug	64/16	512/16	32/16		64/16	512/16	512/512				512/512	512/512
Extrapolated MIC (μg/ml)	14/4	112/3	8/3		16/2	112/3	78/72				56/12	96/72
FICI	0.438	0.360	0.438		0.344	0.391	0.291				0.329	0.328
M9514 (massiliense)
MIC (μg/ml) of each drug	128/8	512/8	32/8	32/8			512/256	512/8
Extrapolated MIC (μg/ml)	25/2	83/1	8/2	6/2			112/40	128/2
FICI	0.388	0.279	0.500	0.438			0.375	0.438
M9515 (massiliense)
MIC (μg/ml) of each drug	256/32	512/32	256/32	128/32	256/32	128/32
Extrapolated MIC (μg/ml)	48/5	96/4	35/5	18/6	43/5	80/6
FICI	0.328	0.293	0.284	0.306	0.306	0.438
M9517 (massiliense)
MIC (μg/ml) of each drug	64/16	512/16	16/16	16/16	64/16	256/16	512/512	512/16
Extrapolated MIC (μg/ml)	14/4	86/2	4/4	4/4	16/4	48/3	88/88	102/2
FICI	0.438	0.288	0.500	0.500	0.500	0.375	0.344	0.328
M9521 (massiliense)
MIC (μg/ml) of each drug	32/8	512/8		16/8	128/8	128/8		256/8		512/8	512/256	512/512	512/16
Extrapolated MIC (μg/ml)	5/2	75/1		4/2	24/2	32/2		64/2		112/2	96/48	74/112	96/3
FICI	0.375	0.253		0.500	0.375	0.438		0.438		0.438	0.375	0.364	0.375
M9522 (abscessus)
MIC (μg/ml) of each drug	64/16	512/16	32/16	32/16	256/16	512/16	512/512	512/16		512/16
Extrapolated MIC (μg/ml)	9/3	96/2	8/4	8/4	49/3	112/4	96/72	128/4		128/3
FICI	0.279	0.317	0.414	0.422	0.388	0.438	0.328	0.500		0.500
M9525 (abscessus)
MIC (μg/ml) of each drug	128/32	512/32	64/32	64/32	256/32	512/32	512/512			512/32
Extrapolated MIC (μg/ml)	22/4	112/2	12/6	9/6	56/4	128/0.5	128/128			128/3
FICI	0.306	0.293	0.360	0.293	0.348	0.266	0.500			0.344
M9526 (abscessus)
MIC (μg/ml) of each drug	64/16
Extrapolated MIC (μg/ml)	13/3
FICI	0.396
M9527 (abscessus)
MIC (μg/ml) of each drug	128/16
Extrapolated MIC (μg/ml)	28/4
FICI	0.438
M9528 (abscessus)
MIC (μg/ml) of each drug	128/32	512/32	32/32	32/32				512/32
Extrapolated MIC (μg/ml)	25/4	112/7	16/6	8/8				128/6
FICI	0.328	0.422	0.438	0.500				0.438
M9529 (abscessus)
MIC (μg/ml) of each drug	128/256			256/256					512/256				512/256
Extrapolated MIC (μg/ml)	32/64			33/44					66/51				112/18
FICI	0.500			0.332					0.326				0.293
M9530 (abscessus)
MIC (μg/ml) of each drug	64/16	512/16	64/16	16/16	128/16	512/16	512/512	512/16		512/16	512/512	512/512
Extrapolated MIC (μg/ml)	11/3	96/2	10/3	3/4	64/4	82/2	104/64	112/3		128/1	128/128	83/83
FICI	0.344	0.317	0.301	0.438	0.422	0.262	0.329	0.410		0.321	0.500	0.328
M9531 (abscessus)
MIC (μg/ml) of each drug	64/16	512/16	64/16	32/16	64/16		512/512
Extrapolated MIC (μg/ml)	10/3	86/2	9/3	6/4	16/4		102/98
FICI	0.329	0.310	0.267	0.410	0.500		0.357

^aFor each clinical strain, the top row of data represents the MIC of each individual drug, whereas the middle row represents the extrapolated MIC of each drug when used in combination. The extrapolated MIC of each drug was determined by averaging the concentrations of the drug in each well in which Mab growth was inhibited. These values were used to determine FICIs. FICIs and MICs in combination were extrapolated using data averaged from at least two biological replicates. The bottom row of data represents the FICI for each synergistic combination. The empty cells indicate lack of synergy.

^bFOX, cefoxitin; IPM, imipenem; CXM, cefuroxime; DOR, doripenem; BIA, biapenem; CDR, cefdinir; FAR, faropenem; CFR, cefadroxil; TEB, tebipenem; ETP, ertapenem; CDN, cefditoren; CPD, cefpodoxime.

TABLE 3.

Numbers (and percentage) of strains against which each drug pair exhibited synergy

Drug combinationa	No. (%) of strains for whichb:
	Synergy was exhibited	MIC of ≥1 drug was brought into therapeutic range	MICs of both drugs were brought into therapeutic range
FOX and IPM	21 (100)	21 (100)	19 (90)
CXM and IPM	17 (81)	16 (76)	3 (14)
DOR and IPM	16 (76)	15 (71)	9 (43)
BIA and IPM	15 (71)	14 (67)	10 (48)
CDR and IPM	12 (57)	12 (57)	11 (52)
FAR and IPM	12 (57)	12 (57)	0 (0)
CFR and TEB	12 (57)	0 (0)	0 (0)
ETP and IPM	10 (48)	10 (48)	0 (0)
CDN and IPM	6 (29)	5 (24)	1 (5)
CPD and IPM	6 (29)	6 (29)	0 (0)
CDN and TEB	6 (29)	2 (10)	0 (0)
CXM and CDN	5 (24)	2 (10)	1 (5)
CDN and BIA	3 (14)	1 (5)	0 (0)

^aFOX, cefoxitin; IPM, imipenem; CXM, cefuroxime; DOR, doripenem; BIA, biapenem; CDR, cefdinir; FAR, faropenem; CFR, cefadroxil; TEB, tebipenem; ETP, ertapenem; CDN, cefditoren; CPD, cefpodoxime.

^bExtrapolated MICs were calculated using the FICI for each drug pair and clinical strain. The therapeutic range was defined as MICs at or below the CLSI breakpoints for cephalosporins (≤16 μg/ml, susceptible; ≤64 μg/ml, intermediately susceptible) and carbapenems (≤4 μg/ml, susceptible; ≤8 μg/ml, intermediately susceptible).

Synergistic combinations were further assessed to determine whether the resulting decrease in MICs was sufficient to bring the isolate into the therapeutic range for each drug (Table 3). Although Clinical and Laboratory Standards Institute (CLSI) guidelines regarding MIC breakpoints for Mab strains are not currently available for most of the antibiotics tested, they have been established for cefoxitin and imipenem (49). Therefore, MIC breakpoints for all cephalosporins and carbapenems were assumed to be the same as those for cefoxitin (≤16 μg/ml, susceptible; ≤64 μg/ml, intermediately susceptible) and imipenem (≤4 μg/ml, susceptible; ≤8 μg/ml, intermediately susceptible), respectively.

In general, synergy between agents with very high initial MICs (i.e., 256 to 512 μg/ml) was often not sufficient to bring the resulting MICs within the therapeutic range. For example, the combination of cefadroxil and tebipenem was unable to reach the therapeutic threshold for either drug, despite exhibiting synergy against 12/21 isolates. This was especially true of the carbapenems with high MICs, as their therapeutic window is much smaller than that of the cephalosporins. Indeed, the combinations of cefpodoxime and imipenem, ertapenem and imipenem, faropenem and imipenem, cefadroxil and tebipenem, cefditoren and biapenem, and cefditoren and tebipenem were unable to achieve the therapeutic range for both drugs against any of the clinical strains.

DISCUSSION

Independent studies have reported the phenomenon of synergy between β-lactams against Mab strains using both in vitro and in vivo approaches (31, 34–36, 50, 51). Although the mechanism behind dual-β-lactam synergy is not fully understood, a hypothesis that two β-lactams that inhibit distinct sets of nonredundant enzymes may exhibit synergy in antibacterial activity has been proposed (32), based on differential inhibition of enzymes involved in peptidoglycan synthesis (30, 32, 52). Because β-lactams are widely available and generally well tolerated by patients, they may be an untapped resource in our desperate fight to treat drug-resistant Mab infections.

In this study, we tested the 13 β-lactam combinations against a library of 21 clinical isolates obtained from CF patients in recent years. The rationale for including these specific dual-β-lactam combinations is a prior proof-of-concept screen against a single Mab strain, the reference strain ATCC 19977, in which they exhibited synergy (34). As may be expected based on prior studies (37–39), this study illustrates the significant heterogeneity that exists among Mab clinical strains with regard to drug susceptibility profiles and potential responses to antibiotics. In many instances, the MICs of several of the antibiotics differed by only a 2-fold dilution. However, these differences were observed at higher drug concentrations, where the amount of drug present differed drastically between the dilutions, and the results were highly reproducible. It also supports our position that study of a single Mab reference strain provides an inadequate assessment of novel treatments. Several clinical isolates harbored much higher MICs for certain drugs, compared to the reference strain, and we observed significant variations in the degree of synergy exhibited by β-lactam pairs.

For example, the MIC of imipenem for ATCC 19977 is 8 μg/ml. Only 5 of the 21 clinical strains had an imipenem MIC of 8 μg/ml; the other MICs ranged from 16 to 256 μg/ml, and the strains were considered resistant to imipenem, based on CLSI breakpoints. This is particularly relevant because imipenem is one of the cornerstone antibiotics used to treat Mab lung infections (21). It is one of only two β-lactams included in the Cystic Fibrosis Foundation and European Cystic Fibrosis Society treatment guidelines, along with cefoxitin (21). In our study, the combination of imipenem and cefoxitin was the only pair that exhibited synergy against 100% of isolates and was capable of reducing MICs to within the therapeutic range for all but one strain (Table 3). However, these two agents are never used concurrently against Mab in clinical practice. This is likely due to the fact that current treatment paradigms were developed using the historical model of β-lactam activity, which was predominantly influenced by observations in Gram-negative organisms that considered only one target in the peptidoglycan synthesis pathway, the d,d-transpeptidases (26). An important and underappreciated reason for this was the availability of only the penicillin and cephalosporin subclasses at the time, which were used to probe enzymes inhibited by β-lactams (53). These two subclasses preferentially bind to d,d-transpeptidases and did not readily permit identification of l,d-transpeptidases (54). According to this historical model, d,d-transpeptidases represented the only enzyme class targeted by β-lactams. Therefore, concurrent use of two β-lactams was considered redundant. The recent discovery of l,d-transpeptidases (55), their dominant role in peptidoglycan synthesis in Mab (28), and their preferential inhibition by the carbapenem subclass of β-lactams (31) have called this historical dogma into question.

Several of the currently recommended first-line agents for treatment of Mab infections have substantial side effect profiles, and regimens often require adjustment due to adverse events or toxicity. For example, amikacin is associated with significant nephrotoxicity and ototoxicity, the latter of which is dose dependent and irreversible. Use of this drug for a prolonged period, as is required for treatment of Mab infections, can eventually cause deafness (56–59). Based on these reports, use of this drug as first-line treatment for Mab disease in CF patients, especially children, is undesirable. Therefore, there is a real potential for dual β-lactams to replace existing first-line agents, which could be a major innovation for our current and future patients.

Despite this, use of dual β-lactams remains a radical approach to treatment and has also raised concern regarding the risk of adverse events. However, a recent meta-analysis comparing dual β-lactams to β-lactam plus aminoglycoside regimens for treatment of Gram-negative infections showed significantly lower rates of nephrotoxicity and ototoxicity among dual β-lactams, with both treatment efficacy and rates of other adverse events being largely equal between the two study arms (60).

The combination of cefoxitin and imipenem exhibited synergy against all 21 of the clinical strains tested, maintaining efficacy despite vast differences in drug resistance profiles. These agents are already used separately in the treatment of Mab lung disease. Given our current deficit of viable therapeutic options against this pathogen, it may be reasonable to consider concurrent use of both agents as an alternative approach in the setting of extensive drug resistance.

Although none of the other β-lactam pairs performed as well as cefoxitin and imipenem, there were three additional combinations that exhibited previously unexpected synergistic activities; these were doripenem and imipenem, biapenem and imipenem, and cefdinir and imipenem, which exhibited synergy against 76%, 71%, and 57% of the isolates, respectively. These agents tended to have slightly lower initial MICs and thus exhibited sufficient synergy to bring the MICs of both drugs to within the therapeutic range for roughly one-half of the isolates (43%, 48%, and 52%, respectively). Both cefdinir and doripenem are commercially available in the United States, and cefdinir’s oral formulation makes it logistically appealing for patients. Biapenem (61) is not currently FDA approved in the United States, but it has been used commercially in Asia for nearly 20 years and has shown efficacy against Mab in vivo (36, 62) and against Mycobacterium tuberculosis in vitro and in vivo (63–66). Given the abysmal cure rates of 25 to 40% with current regimens (11, 12) and our urgent need for tolerable long-term therapies against Mab pulmonary disease, these combinations may be worth consideration in certain clinical cases.

The combination of cefuroxime and imipenem exhibited synergy against the second largest number of clinical strains at 17/21 strains (81%). However, the therapeutic ranges for both drugs were achieved in only 14% of isolates due to the very high initial MIC of cefuroxime. This trend was observed with several of the β-lactam pairs, in which synergy was achieved with a number of isolates but the therapeutic window was not, thus limiting the potential for clinical use of these combinations (Table 3).

As mentioned above, the current Mab reference strain (ATCC 19977), which was derived nearly 70 years ago from a synovial fluid sample, may not be representative of modern trends in Mab pulmonary disease, especially with regard to drug resistance profiles. We speculate that our fully sequenced library of clinical strains may contain potential candidates for one or more updated reference strains. In a recent study, we evaluated six of these clinical isolates (M9505, M9513, M9521, M9522, M9526, and M9529) in a murine model of Mab pulmonary disease to determine whether they were capable of causing invasive infections in vivo (62), as would be required of any candidate reference strain for the purpose of drug treatment studies. All of the clinical strains exhibited similar lung implantation burdens, and three of the isolates (M9513, M9521, and M9529) caused invasive infections, leading to ≥3-log₁₀ unit increases in the bacterial burden at 3 weeks, similar to results seen with ATCC 19977 under the same conditions. Although additional in vivo studies are needed to further characterize these strains, they may have the potential to broaden our current repertoire of reference strains.

This study was limited by the fact that the dual-β-lactam combinations tested were initially identified as exhibiting synergy against the Mab reference strain; therefore, it is likely that a percentage of the additional 107 β-lactam combinations that were previously deemed nonsynergistic against ATCC 19977 (34) might have exhibited synergy against one or more clinical strains. Additionally, this study was performed in vitro, which does not necessarily correlate with clinical responses in patients with Mab infections. Further studies are needed to assess the in vivo efficacy of these novel combinations in animal models of Mab disease or in preclinical models that approximate certain aspects of drug activity in humans, such as the hollow-fiber model that has been used to model the pharmacokinetics/pharmacodynamics of antibacterials against Mab (67).

In conclusion, Mab pulmonary disease can have devastating effects on patient health, and current treatments are prolonged, poorly tolerated, and too often ineffective. In this study, several dual-β-lactam combinations exhibited synergistic efficacy against a genetically varied library of clinical strains that are resistant to several drugs currently used to treat Mab disease. In particular, cefoxitin and imipenem exhibited synergy against 100% of the isolates, suggesting their potential for widespread clinical efficacy, as these agents are already used separately against Mab. Given the urgent need for novel therapeutics against this pathogen, dual-β-lactam regimens may offer viable immediate treatment options with efficacy against drug-resistant Mab infections.

MATERIALS AND METHODS

Bacterial strains and in vitro growth conditions.

Clinical Mab isolates utilized in this study were obtained by Nicole Parrish from CF patients seen at the Johns Hopkins Hospital between 2004 and 2018 and were biobanked by the Johns Hopkins Clinical Microbiology Laboratory. Strains were grown at 37°C in Middlebrook 7H9 broth (Difco) supplemented with 0.5% glycerol, 10% albumin-dextrose-catalase enrichment, and 0.05% Tween 80, with constant shaking at 220 rpm in an orbital shaker. All drugs were obtained from the following commercial vendors: ertapenem, Toronto Research Chemicals; imipenem, doripenem, biapenem, faropenem, tebipenem, and all cephalosporins, Sigma-Aldrich. To assess the quality of these compounds, a few were randomly selected and assessed by liquid chromatography-mass spectrometry. The purity of compounds ranged from 95% to 99%.

MICs.

The MIC of each drug against each clinical isolate of Mab was determined using the standard broth dilution method (68, 69) in accordance with CLSI guidelines specific for this organism (49). In summary, powdered drug stocks were reconstituted in either dimethyl sulfoxide (DMSO) or sterile deionized water (for imipenem and biapenem, given their poor solubility in DMSO), and 2-fold serial dilutions were prepared in Middlebrook 7H9 broth to obtain final drug concentrations ranging from 512 μg/ml to 2 μg/ml in 96-well plates in a final volume of 200 μl. A total of 10⁵ CFU of Mab from an exponentially growing culture was added to each well. Mab culture without drug and 7H9 broth alone were included in each plate as positive and negative controls, respectively. Plates were incubated at 30°C for 72 h according to CLSI guidelines. Growth of Mab or lack thereof was determined using a Sensititre Manual Viewbox, and the MIC for each drug was recorded as the lowest concentration at which Mab growth was not observed. All MIC assessments were repeated to verify results.

Checkerboard titration assay.

The checkerboard titration assay is a modified broth dilution assay and was performed as described previously (46, 47). To determine the degree of synergy, two drugs were added to Middlebrook 7H9 broth in a 96-well plate, each starting at 2× MIC and serially diluted 2-fold up to 1/64× MIC; therefore, all possible 2-fold dilution combinations from 2× to 1/64× MIC were assayed. A total of 10⁵ CFU of Mab was inoculated into each well. Plates were incubated at 30°C and evaluated for Mab growth by visual inspection at 72 h using a Sensititre Manual Viewbox. The FIC of each drug in combination was determined as described (46, 47). The FIC of a drug in a sample is calculated as the concentration of the drug divided by the MIC of the drug when used alone. The FICI is the sum of the FICs of two drugs in a sample. The FICI was calculated for each combination of drugs that inhibited Mab growth at less than one-half the MIC of each individual drug. FICIs of ≤0.5 were interpreted as synergy, FICIs of >0.5 to 4 as indifference, and FICIs >4 as antagonism, according to the most stringent interpretation recommended (48). As an internal control, the MIC of each individual drug was also assessed via broth microdilution within each plate. All combinations with FICIs of ≤0.5 were tested in duplicate to confirm reproducibility, and an average FICI was calculated and reported here.

DNA extraction, whole-genome sequencing, and assembly.

To extract genomic DNA from Mab, we used the phenol-chloroform extraction method optimized for mycobacteria (70). Briefly, for each strain, 30 ml of culture was centrifuged, resuspended in Tris-EDTA (TE) buffer containing lysozyme and RNase A, and incubated overnight at 37°C. SDS (10%) with proteinase K was added, and the mixture was incubated for 10 min at 65°C; 5 M NaCl was then added together with cetyltrimethylammonium bromide (CTAB)/NaCl solution, and the mixture was incubated again for 10 min at 65°C. A solution of 25:24:1 phenol/chloroform/isoamyl alcohol was then added to the suspension and gently mixed before centrifugation for 10 min. The supernatant was transferred to a new tube, which contained a solution of 24:1 chloroform/isoamyl alcohol, and was centrifuged again for 10 min. Next, the top layer of the suspension was added to a tube containing ice-cold isopropanol and was incubated for 20 min at –20°C. This suspension was then centrifuged, the supernatant was discarded, 70% ethanol was added, and the mixture was centrifuged for another 5 min. The supernatant was discarded, the tubes containing the DNA were air dried overnight, and the DNA was resuspended with 300 μl of deionized water the following day.

The genomic DNA obtained was then sequenced using the Illumina PE150 platform (Novogene, CA, USA) and the PacBio Sequel platform (Genewiz, NJ, USA). To identify to the subspecies level the clinical strains at our disposal, we used Geneious v11.1.5 (Biomatters) to de novo assemble the reads into contigs using SPAdes (bundled in Geneious). Subsequently, the Geneious read-mapping algorithm was used to produce longer contigs. We then used the “map to reference” feature in Geneious to map the six different genetic markers that were validated previously (42, 43) for subspecies identification of Mab strains; these genes were hsp65 (MAB_0650), secA (MAB_2397), rpoB (MAB_3869c), polC (MAB_2696c), hoa (MAB_0626), and ftsZ (MAB_2009). Once we obtained the sequence of each of the six genes for each of our strains, we concatenated them in the same order to create a single sequence for each strain, including the reference strain ATCC 19977 for Mab and the reference strain CCUG 48898 for Mab subspecies massiliense. Each sequence was aligned using MUSCLE to create a phylogenetic tree using Geneious’ own tree builder. The parameters set for the tree building included the Tamura-Nei genetic distance model with the UPGMA tree-building method, bootstrapped with 1,000 iterations (43). To compare the genomes generated de novo, we used the Mauve aligner in Geneious to generate an interactive map that allowed for direct comparison of the genomes.