The Combination of Doripenem and Colistin Is Bactericidal and Synergistic against Colistin-Resistant, Carbapenemase-Producing Klebsiella pneumoniae

Abstract

We tested two-drug combinations of doripenem, colistin, gentamicin, and doxycycline against 12 carbapenemase-producing Klebsiella pneumoniae (KPC) isolates by time-kill. The combination of doripenem and colistin reduced the starting inocula by 2 logs for each isolate (range, 2.02 to 6.01 log₁₀) and was bactericidal and synergistic against 75 and 50%, respectively. Among colistin- and pan-drug-resistant isolates, synergy was identified in 60 and 67%, respectively. All other combinations were inferior. We are currently evaluating the combination of doripenem and colistin as a frontline therapy for KPC infection.

TEXT

Infections with carbapenemase-producing Klebsiella pneumoniae (KPC) are emerging causes of morbidity and mortality at institutions worldwide (2, 11, 14–16). KPC isolates are defined by resistance to carbapenems and are generally resistant to most other antimicrobial classes. Salvage agents that may retain activity against KPC isolates include colistin, gentamicin, doxycycline, and tigecycline (8, 9, 18). Resistance to these agents is increasing (1, 6), however, and treatment failures are well recognized with each (8). Given the dearth of effective antimicrobial agents and poor clinical outcomes among patients with KPC infections, combination therapy has been utilized. Unfortunately, in vitro synergy studies have yielded conflicting results and left clinicians with no clear direction in selecting combination regimens. At our center, we have utilized the combination of doripenem and colistin to treat patients with KPC infections, a decision that was extrapolated from our encouraging preliminary experience in treating extensively drug-resistant (XDR) Acinetobacter baumannii infections (17). The primary objective of this study was to determine if this combination is active against KPC isolates recovered from our patients. In particular, we focused on colistin-resistant isolates determined in vitro by broth dilution. We also sought to test other two-drug combinations that include commonly employed salvage agents.

Toward these ends, we determined the in vitro activity of doripenem, colistin, gentamicin, and doxycycline, alone and in combination, against 12 KPC isolates collected over a 3-year period from unique patients at our center. We included doxycycline rather than tigecycline because it is more commonly used at our center, particularly as part of the treatment of KPC urinary tract infections.

We confirmed the presence of KPC for each isolate by a modified Hodge test, and we determined MICs by standard broth microdilution methods (4). One hundred percent (12/12), 83% (10/12), 83% (10/12), and 58% (7/12) of isolates were resistant to doripenem, colistin, gentamicin, and doxycycline, respectively; 50% (6/12) were resistant to all agents. The corresponding MIC₉₀s were >128, 16, 64, and 16 μg/ml, respectively. Using time-kill assays (12), we found that colistin was bactericidal (≥3 log₁₀ decrease from the starting inoculum) against 33% (4/12), including both colistin-susceptible strains. Doripenem and gentamicin were bactericidal against one isolate each, whereas doxycycline was not bactericidal against any isolate (Table 1).

Table 1.

MICs for KPC isolates^a

Strain	MIC (μg/ml)				Log Δ from the starting inoculum (log10 CFU/ml)b
	Doripenem	Colistin	Gentamicin	Doxycycline	Doripenem	Colistin	Gentamicin	Doxycycline
1c	64	8	16	8	3.67	−3.76	3.78	−1.40
18	16	64	32	4	−4.96	−2.86	3.75	2.73
82	16	8	8	4	3.64	3.64	3.22	3.95
124c	16	16	64	16	2.99	−1.89	−2.35	2.87
133c	128	16	64	16	3.83	−3.60	3.62	3.70
136	128	4	>64	4	3.53	−2.05	2.78	3.24
141	>128	0.5	1	4	4.44	−3.19	−2.74	3.90
145c	>128	16	64	16	2.66	−1.11	−0.03	2.12
167	32	1	16	8	3.75	−4.74	4.43	3.49
180c	32	8	32	16	4.06	−0.47	3.53	3.69
182c	32	8	64	16	2.66	−0.47	3.93	3.98
183	64	16	1	2	4.39	−0.79	−3.70	3.52

^aThe following susceptibility breakpoints were used: doripenem and colistin, ≤2 μg/ml, and gentamicin and doxycycline, ≤4 μg/ml.

^bLog Δ = Final inoculum − starting inoculum (log₁₀ CFU/ml). Negative numbers represent a decrease from the starting inoculum, and positive numbers represent an increase. Fixed concentrations of doripenem (8 μg/ml), colistin (1 μg/ml), gentamicin (2 μg/ml), and doxycycline (2 μg/ml) were used for all isolates.

^cIndicates isolates resistant to all agents tested.

We next assessed the activity of two-drug combinations by time-kills using 50-ml flasks containing 9.9 ml of cation-adjusted Mueller-Hinton broth, 0.1 ml of organism (starting inoculums of 1 × 10⁶ CFU/ml), and fixed drug concentrations alone and in combination (doripenem, 8 μg/ml; colistin, 1 μg/ml; gentamicin, 2 μg/ml; and doxycycline, 2 μg/ml). Flasks were incubated with agitation at 37°C, and viable bacterial colonies were enumerated at 0, 4, 8, 12, and 24 h as previously reported (17). Antimicrobial concentrations were representative of achievable serum levels. Synergy and antagonism were defined as ≥2-log₁₀ greater and lesser kills, respectively, in combination treatments versus the most active single agent (21). The combination of doripenem and colistin was bactericidal against 75% (9/12) of isolates (Table 2) and synergistic against 50% (6/12) (Table 3). Log kills ranged from 2.02 to 6.01 log₁₀ CFU/ml (median, 3.95 log₁₀ CFU/ml) (Table 2), and this was the only combination that reduced the starting inocula of all strains. Moreover, doripenem plus colistin was synergistic against 60% (6/10) or 67% (4/6) of isolates that were resistant to colistin (MICs, 4 to 64 μg/ml) or resistant to all agents, respectively.

Table 2.

Log change (log₁₀ CFU/ml) from the starting inoculum and most active single agent after 24 h of incubation

Combination	Log Δ (log10 CFU/ml)a for isolate:
	1	18	82	124	133	136	141	145	167	180	182	183
Colistin + doripenem	−6.01	−4.12	−2.09	−5.99	−3.57	−3.95	−3.94	−4.65	−5.76	−2.02	−2.73	−3.90
	−2.25	0.84	−5.73	−4.10	0.03	−1.90	−0.75	−3.54	−1.02	−1.55	−2.26	−3.11
Colistin + gentamicin	−4.56	−3.07	−3.20	−5.87	−2.07	−2.22	−5.60	1.70	−2.05	−0.60	−0.56	−5.00
	−0.80	−0.21	−6.84	−3.52	1.53	−0.17	−2.41	1.73	2.69	−0.13	−0.09	−1.30
Colistin + doxycycline	−1.21	−0.38	0.29	−0.57	−1.06	−0.85	−3.37	−0.24	−5.48	0.02	−0.60	−1.15
	2.55	2.48	−3.35	1.35	2.54	1.20	−0.18	0.87	−0.74	0.49	−0.13	−0.36
Doripenem + gentamicin	3.69	−5.95	−2.40	3.85	4.11	0.99	−4.43	2.63	3.88	3.94	3.40	−5.04
	0.02	−0.99	−5.62	6.20	0.49	−1.79	−1.69	2.66	0.13	0.41	0.74	−1.34
Doripenem + doxycycline	4.47	−0.88	3.90	−3.03	4.13	3.21	3.95	2.70	4.49	−0.43	4.06	−1.99
	5.87	4.08	0.26	−5.90	0.89	−0.03	0.05	0.58	1.00	−4.12	1.40	−5.51
Gentamicin +doxycycline	−1.61	−0.54	−5.98	4.01	2.93	−1.38	−5.65	2.74	4.51	3.42	3.73	−5.73
	−0.21	−3.27	−9.20	6.36	−0.69	−4.16	−2.91	2.77	1.02	−0.11	−0.20	−2.03

^aLog Δ is defined as follows. Nonshaded rows: final inoculum − starting inoculum (log₁₀ CFU/ml). Negative numbers represent a decrease from the starting inoculum, and positive numbers represent an increase. Shaded rows: final inoculum of the most active single drug − final inoculum of the two-drug combination (log₁₀ CFU/ml). Negative numbers represent greater killing with two-drug combinations versus a single drug. Fixed concentrations of doripenem (8 μg/ml), colistin (1 μg/ml), gentamicin (2 μg/ml), and doxycycline (2 μg/ml) were used for all isolates. Bold numbers indicate synergy.

Table 3.

Interactions between two drugs tested in combination

Backbone agent	Agent tested in combination	Synergy, n (%)	Indifference, n (%)	Antagonism, n (%)
Colistin	Doripenem	6 (50)	6 (50)	0
	Gentamicin	3 (25)	8 (67)	1 (8)
	Doxycycline	1 (8)	8 (67)	3 (25)
Doripenem	Gentamicin	1 (8)	9 (75)	2 (17)
	Doxycycline	3 (25)	7 (58)	2 (17)
Gentamicin	Doxycycline	5 (42)	5 (42)	2 (17)

Other combinations were less active (Table 3). Colistin plus gentamicin was bactericidal and synergistic against 50% (6/12) and 25% (3/12) of isolates, respectively. The next most active combination, gentamicin plus doxycycline, was bactericidal and synergistic against 25% (3/12) and 42% (5/12) of isolates, respectively. Synergy was achieved against isolates that were doxycycline susceptible; bactericidal activity or synergy was not identified among doxycycline-resistant strains. Doripenem plus gentamicin was also bactericidal against 25% (3/12), but synergy was evident against only 8% (1/12). The least active combinations were colistin plus doxycycline and doripenem plus doxycycline, which were bactericidal against 17% (2/12) and 8% (1/12) of isolates, respectively. Doripenem plus doxycycline was synergistic against 25% (3/12) of isolates, whereas colistin plus doxycycline was synergistic against 8% (1/12). Antagonism was identified with all combinations except doripenem plus colistin (Table 3). The combination of colistin and doxycycline was antagonistic in 25% (3/12).

Doripenem plus colistin and doripenem plus gentamicin were associated with the lowest and highest median areas under the curve (AUCs), respectively (Table 4). Indeed, the median AUC for colistin plus doripenem was lower than those for the other combinations (P = 0.004), including the individual combinations of gentamicin plus doxycycline (P = 0.04) and doripenem plus doxycycline (P < 0.0001) (Fig. 1) (GraphPad Prism Software, La Jolla, CA).

Table 4.

AUCs for two-drug combinations after 24 h of incubation

Isolate	AUCa
	Colistin			Doripenem		Gentamicin plus doxycycline
	Plus doripenem	Plus gentamicin	Plus doxycycline	Plus gentamicin	Plus doxycycline
1	62.34	88.02	NA	209.5	NA	106.6
18	69.6	103.9	NA	81.7	NA	134.2
82	74.8	57.4	141.0	62.8	150.6	52.9
124	47.8	37.46	132.5	NA	128.2	NA
133	87.9	110.2	NA	176.5	168.5	206.0
136	84.8	121.4	138.5	202.9	213.6	143.4
141	59.4	29.1	99.9	22.4	156.7	50.8
145	81.5	139.0	157.4	NA	195.3	NA
167	33.9	NA	51.0	207.6	207.4	220.1
180	118.3	132.4	146.7	155.4	138.7	203.7
182	112.3	137.0	135.5	183.8	164.8	168.2
183	82.1	10.0	132.1	15.9	107.5	26.4
Median (range)	78.2 (33.9–118.3)	103.9 (10.0–139.0)	135.5 (51.0–157.4)	165.9 (15.9–209.5)	160.8 (107.5–213.6)	138.8 (26.4–220.1)

^aNA, not applicable; area under the curve was not calculated for antagonistic combinations.

Fig 1.

Representative area under the curve (AUC) for isolates alone and in combination for the following drug combinations: (a) colistin (dark gray) plus doripenem (light gray); (b) gentamicin (light gray) plus doxycycline (dark gray); and (c) doripenem (light gray) plus doxycycline (dark gray). Black shading represents the AUC for two-drug combinations.

Taken together, our data demonstrate that doripenem plus colistin was the most active two-drug combination against KPC isolates. Doripenem plus colistin attained the highest levels of bactericidal activity and synergy, and it was the only combination that did not show any antagonism. Moreover, the combination was effective at clinically achievable concentrations of both drugs (3, 7) and it was particularly active against colistin- and pan-drug-resistant isolates. In this regard, our findings are similar to those we reported against XDR Acinetobacter baumannii isolates from our center (17). Furthermore, our results extend our understanding of in vitro synergy against KPC isolates, which have been studied in colistin-susceptible strains (10, 20, 22) or by checkerboard broth microdilution (5) previously.

At present, the treatment options for patients with KPC infections are limited, as evident by crude mortality rates that are reported to range from 42 to 69% (13, 14, 19, 24). Mortality rates are higher among patients infected with colistin-resistant isolates (24). Investigators have reported that combination therapy may improve outcomes of KPC infections compared to those achieved with monotherapy (8, 23), but optimal regimens are not defined and none have been shown to be effective against colistin-resistant isolates. Of note, we reported promising preliminary data in a pilot clinical study using doripenem plus colistin to treat XDR A. baumannii infections (17). Based on these data and the in vitro findings of the current study, we are currently using the combination as the recommended front-line regimen against KPC infections.

The other combinations tested in this study demonstrated inferior bactericidal activity and less synergy. Furthermore, the second most potent combination in vitro (colistin plus gentamicin) is limited in clinical settings by the potential for additive toxicity. Our data suggest that doxycycline plus gentamicin may be useful against isolates that retain susceptibility to the former agent, but at our center such isolates are in the minority. In the end, this combination was synergistic against only 25% (3/12) of isolates and may be unsuitable for the treatment of severe or life-threatening KPC infections. At present, we reserve doxycycline-containing regimens for the treatment of uncomplicated cystitis at our center. The remaining combinations were even less effective in vitro, and we do not employ them in the clinic. Finally, it is important to note that the number of strains evaluated here is limited and future studies are needed. Toward this, the exact mechanisms underlying the synergistic effects between doripenem and colistin have not been identified and merit further investigation. It is possible that colistin permeabilizes the outer cell membrane of KPC isolates, allowing high concentrations of doripenem to access penicillin-binding proteins in the cytoplasmic membrane where the drug acts.

In conclusion, it is important to stress that these results were obtained against KPC isolates from a single center. As such, the results may not be applicable to isolates from other centers or geographic locations. It is advisable for centers to test isolates recovered from patients at their institution. Having introduced doripenem plus colistin as the recommended therapy for KPC infections at our center, we are in the process of accumulating our clinical experience and assessing the impact of the regimen on preventing the emergence of colistin resistance.