Comparative efficacy and safety of treatment options for MDR and XDR Acinetobacter baumannii infections: a systematic review and network meta-analysis

Abstract

Objectives

To comprehensively compare and rank the efficacy and safety of available treatment options for patients with MDR and XDR Acinetobacter baumannii (AB) infection.

Methods

We searched PubMed, Embase and the Cochrane register of trials systematically for studies that examined treatment options for patients with MDR- and XDR-AB infections until April 2016. Network meta-analysis (NMA) was performed to estimate the risk ratio (RR) and 95% CI from both direct and indirect evidence. Primary outcomes were clinical cure and microbiological cure. Secondary outcomes were all-cause mortality and nephrotoxic and non-nephrotoxic adverse events.

Results

A total of 29 studies with 2529 patients (median age 60 years; 65% male; median APACHE II score 19.0) were included. Although there were no statistically significant differences between treatment options, triple therapy with colistin, sulbactam and tigecycline had the highest clinical cure rate. Colistin in combination with sulbactam was associated with a significantly higher microbiological cure rate compared with colistin in combination with tigecycline (RR 1.23; 95% CI 1.03–1.47) and colistin monotherapy (RR 1.21; 95% CI 1.06–1.38). No significant differences in all-cause mortality were noted between treatment options. Tigecycline-based therapy also appeared less effective for achieving a microbiological cure and is not appropriate for treating bloodstream MDR- and XDR-AB infections.

Conclusions

Combination therapy of colistin with sulbactam demonstrates superiority in terms of microbiological cure with a safety profile similar to that of colistin monotherapy. Thus, our findings support the use of this combination as a treatment for MDR- and XDR-AB infections.

Introduction

MDR and XDR Acinetobacter baumannii (MDR- and XDR-AB) are serious causes of healthcare-associated infections worldwide.^1,2 The common presentations of these particular pathogens include hospital-acquired pneumonia and bloodstream infection.^3,4 MDR- and XDR-AB infections have become more difficult to treat with the emergence of isolates resistant to commonly available prescribed antibiotics.² This results in poor prognosis, high treatment failure and significantly increased mortality.^5,6

Currently, a consensus recommendation for the optimal treatment of these infections has not been established. Several antimicrobials have been used to treat MDR- and XDR-AB infections, including colistin, sulbactam and tigecycline as monotherapy or combination therapy.⁷ Colistin is currently used as the backbone therapy, despite its nephrotoxic effects.^7,8 Sulbactam and tigecycline show in vivo activity against a broad spectrum of MDR-AB.^9,10 The use of colistin with sulbactam or tigecycline as a combination therapy could bring about additional benefits to increase treatment success, but their use is still controversial.^8,11

Previous pairwise meta-analyses were performed to evaluate the efficacy and safety of common backbone treatments for MDR-AB.^8,12–14 However, most of these studies considered only direct comparisons of the intervention and their results remain inconclusive. Network meta-analysis (NMA) is a methodology suited to assessing multiple interventions using indirect comparisons. This allows comparisons of treatment options for which there have been no head-to-head comparison.¹⁵ We therefore performed a systematic review (SR) and NMA to comprehensively compare and rank the efficacy and safety of available treatment options for treating MDR- and XDR-AB-infected patients.

Methods

Study design

An SR and NMA were conducted following the protocol registered with PROSPERO, number CRD42016042494.¹⁶ This study was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement extension for NMA.¹⁷

Participants

Patients with MDR-AB infections such as hospital-acquired pneumonia (HAP), ventilator-associated pneumonia (VAP) and bloodstream infection (BSI) were included. These infections were defined according to IDSA¹⁸ and US CDC¹⁹ criteria.

Search strategy and selection criteria

We searched PubMed, Embase, the Cochrane Central Register of Controlled Trials and Cumulative Index to Nursing and Allied Health Literature (CINAHL) for relevant articles published from inception to 18 April 2016, using the keywords ‘Acinetobacter*’, ‘*drug resistant’, ‘cure’, ‘response’, ‘mortality’, ‘nephrotoxicity’ and ‘adverse events’. We also screened Clinicaltrials.gov, grey literature reports and published systematic reviews for additional relevant studies. No language, publication date or publication status restrictions were imposed.

We included studies evaluating the efficacy and safety of the following antibiotic regimens: colistin, tigecycline, sulbactam or other antibiotics in monotherapy and combination or triple therapy for treating MDR, XDR or pandrug-resistant (PDR) A.baumannii infection. Details of antibiotic regimens are shown in Table S1 (available as Supplementary data at JAC Online). However, studies consisting of patients with mixed Gram-negative bacilli, single-arm studies, abstracts presented at conferences, editorials, reviews, systematic reviews and meta-analyses were excluded. Although observational studies are not as good as randomized controlled trials (RCTs), including observational studies can help offset the limitations of analysing limited RCT data.^20,21

Study selection and data extraction

Two investigators (K. K. and K. K. H.) independently screened the titles and abstracts of retrieved studies to identify potentially eligible ones. Any discrepancies were resolved by consensus within the review team (K. K., K. K. H., S. S., A. A. and N. C.). Each of the potentially relevant studies was accessed in a full-text manner and reviewed by the same investigators against the inclusion criteria. We extracted study design, study size, details of intervention (such as treatment regimens, dose and treatment durations) and patient characteristics [including mean age, gender, type of drug resistance, site of infection and mean Acute Physiological Assessment and Chronic Health Evaluation II (APACHE II) score]. All extracted data were independently checked by two investigators (S. S. and K. K. H.).

Quality assessment

Risk of bias of individual studies was assessed by three investigators (K. K., K. K. H. and S. S.) using the Cochrane Risk-of-bias Tool for RCTs²² and the Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I) tool for observational studies.²³

Outcomes and definitions

Primary outcomes were clinical cure and microbiological cure. Clinical cure was defined as a resolution of signs and symptoms of infection at the end of treatment, while microbiological cure was defined as a negative culture after initial treatments.¹⁸ Secondary outcomes were all-cause mortality, nephrotoxicity (defined according to KDIGO 2012)²⁴ and other adverse events (neurotoxicity, hepatotoxicity and skin rash). MDR was defined as acquired non-susceptibility to at least one agent in three or more antimicrobial categories. XDR was defined as non-susceptibility to at least one agent in all but two or fewer antimicrobial categories. Additionally, the causative pathogens had to be resistant to carbapenems. PDR was defined as non-susceptibility to all agents in all antimicrobial categories.²⁵

Data synthesis and analysis

Pairwise meta-analysis was performed using the DerSimonian and Laird random-effects model to estimate pooled risk ratio (RR) with 95% CI.²⁶ Heterogeneity in each pairwise comparison was assessed using the I² statistic.²⁷ We also performed a random-effects NMA to combine direct and indirect evidence of all treatment effects using STATA 14 (StataCorp)^28,29 with the methods of NMA described by Lu and Ades.³⁰ Since we decided to include observational studies in this analysis, a two-stage approach was used to perform evidence synthesis. Initially, only data from RCTs were analysed. Then, data from both RCTs and observational studies were incorporated and analysed.²⁰

Inconsistency between direct and indirect sources of evidence was systematically assessed.^31,32 We used a loop-specific approach to identify consistency within every closed triangular or quadratic loop. We then considered the inconsistency factor (IF) to determine whether there is a difference between direct and indirect estimates for a specific treatment comparison in the loop.³¹ The node splitting method was used to explore within-network inconsistency by separating evidence into direct and indirect evidence.³³ We used the surface under the cumulative ranking (SUCRA), which estimates the probability of the ranking of each treatment in a treatment hierarchy.³⁴ We assessed small-study effects by using a comparison-adjusted funnel plot of treatments to detect the presence of any publication bias in NMA.^28,34

To determine whether the results were affected by study characteristics, we performed subgroup NMA on primary outcomes according to the type of drug resistance (MDR, XDR or PDR) and the site of infection (pneumonia, BSI or mixed infection). Multiple sensitivity analyses were performed to assess the robustness of the findings. These were based on: (i) analysing only RCTs; (ii) analysing only high-quality studies; (iii) excluding studies that did not clearly define drug resistance; (iv) excluding studies published before 2010; and (v) excluding studies with small sample size (at the 25th percentile).³⁵ Statistical testing was two-sided, with P < 0.05 indicating statistical significance.

Results

Overall, 5483 records were identified through database searches, and 38 records were identified through other sources (Figure 1). Of 95 potentially relevant articles, 29 studies were included in the systematic review,^6,36–63 and 26 studies were incorporated in NMA.^{6,36,38–42,44,46–63} The full details of our literature search are shown in Table S2.

Figure 1.

Flow diagram and references of included studies.

Characteristics and quality of included studies

The studies included in the SR and NMA are summarized in Table 1. Among 29 studies involving 2529 patients, 4 were RCTs (374 patients),^38,48,49,58 while the other 25 were cohort studies (2155 patients).^{6,36,37,39–47,50–57,59–63} Within the cohort studies, 4 were prospective^6,57,59,60 and 21 were retrospective.^{36,37,39–47,50–56,61–63} The median age of study participants was 60 years (IQR 57–66) and 65.5% of participants were male. The median of average APACHE II score of patients was 19.0 (IQR 17.5–22.5). Seventeen studies (58.5%) were performed on pneumonia (12 studies in VAP alone^{6,38,40–43,45,48,54,60,61,63} and 5 studies on VAP or HAP),^{44,46,53,55,62} whereas the other 12 studies examined various infections, including BSIs, pneumonia and intra-abdominal infections.^{36,37,39,47,49–52,56–59} Most studies (13 studies, 45.0%) included XDR-AB patients,^{37,38,42,46,49,51,52,54–56,59,61,63} while the remaining studies included MDR-AB (12 studies, 41.0%),^{6,39–41,43–45,47,48,50,53,58} mixed MDR- and XDR-AB (2 studies, 7.0%)^57,62 and PDR-AB patients (2 studies, 7.0%).^36,60 The full details of study characteristics are given in Tables S3 to S6. In terms of study quality, the risk of bias in all RCTs was unclear,^38,48,49,58 while for observational studies the risk of bias was moderate for 14 of them (56.0%),^{39,40,42–45,47,50,53,55,59,60,62,63} serious for 10 (40.0%),^{6,37,41,46,51,52,54,56,57,61} and critical for 1 (4.0%)³⁶ (Figure S1).

Table 1.

Characteristics of included studies

Study	Location	Study design	Infection	Resistance	Participants	Treatment regimens	Treatment details	Treatment duration (days)	Mean age (years)	Males (%)	Mean APACHE II score
Betrosian (2007)38	Greece	RCT	VAP	XDR	28	COL; SUL	colistin 3 MIU q8h; ampicillin/sulbactam 9 g q8h	9.2 ± 1.5 vs 9.9 ± 2.6	70.0	50.5	14.0
Aydemir (2013)48	Turkey	RCT	VAP	MDR	43	COL; COL + OTH	colistin 300 mg/day; colistin 300 mg/day + rifampicin 600 mg/day	9.3 ± 3.2	61.0	70.0	19.0
Durante-Mangoni (2013)49	Italy	RCT	mixed (VAP 70%, BSI 20%, HAP 9%)	XDR	209	COL; COL + OTH	colistin 2 MIU q8h; colistin 2 MIU q8h + rifampicin 600 mg q12h	12.5 (8–17)	62.0	65.5	NR
Sirijatuphat (2014)58	Thailand	RCT	mixed (pneumonia 75%, BSI 5%)	MDRa	94	COL; COL + OTH	colistin 5 mg/kg/day; colistin 5–10 mg/kg/day + fosfomycin 4 g q12h	12 (3–15)	68.0	47.0	22.5
Garnacho- Montero (2003)6	Spain	PCS	VAP	MDR	35	COL; OTH	colistin 2.5–5 mg/kg/day; imipenem 2–3 g/day	10 (8–12)	60.0	76.5	NR
Lee (2005)36	Taiwan	RCS	mixed	PDR	89	SUL + OTH; OTH	sulbactam + other antibiotics; amikacin or cephalosporins or fluoroquinolones	NA	71.0	54.0	NR
Holloway (2006)37	USA	RCS	mixed (VAP 50%, BSI 43%)	XDR	37	COL; OTH	colistin 2 MIU q8h; doxycycline 100 mg q12h	NA	41.0	78.0	19.0
Gounden (2009)39	South Africa	RCS	mixed (pneumonia 70%, BSI 30%)	MDR	64	COL; OTH	colistin 2 MIU q8h; tobramycin 5–6 mg/kg/day	8 (5–13); 7 (6–10)	44.0	NA	14.5
Jang (2009)40	Korea	RCS	VAP	MDR	41	COL; COL + SUL	colistin 2.5 mg/kg q12h; colistin 2.5 mg/kg q12h + sulbactam	12.3 ± 6.8; 19.1 ± 11.2	60.0	71.5	27.0
Chan (2010)41	USA	RCS	VAP	MDRa	14	COL; SUL; TGC + OTH; COL + SUL + TGC	colistin 2.5–5 mg/kg/day; ampicillin/sulbactam 2 g q6h; tigecycline 50 mg q12h + other antibiotics; colistin + ampicillin/sulbactam + tigecycline	13 ± 4	40.0	73.0	NR
Nakwan (2011)42	Thailand	RCS	VAP	XDR	15	COL + OTH; OTH	colistin 4mg/kg q12h; other antibiotics	9 (4–14)	NAb	67.0	NR
Pérez–Pedrero (2011)43	Spain	RCS	VAP	MDR	54	COL; COL; COL	colistin iv 1.5 MIU q12h; colistin nb 2–3 MIU per day; colistin iv with nb 2–3 MIU per day	NA	57.0	71.7	11.5
Tasbakan (2011)44	Turkey	RCS	HAP, VAP	MDR	72	TGC; TGC + OTH	tigecycline 50 mg q12h; tigecycline + imipenem 500 mg q6h	10.7 ± 4.8	65.0	61.0	19.3
Kalin (2012)45	Turkey	RCS	VAP	MDRa	45	COL; COL; COL	colistin high dose; 2.5 mg/kg q6h; colistin normal dose; 2.5 mg/kg q12h; colistin low dose; renal dose adjustment	NA	49.0	70.0	22.0
Shields (2012)46	USA	RCS	HAP, VAP	XDR	37	COL + OTH; OTH	colistin + other antibiotics; other antibiotics	15 (5–44)	56.0	63.0	19.0
Shin (2012)47	Korea	RCS	mixed (VAP 55%, BSI 30%)	MDR	27	TGC; TGC + OTH	tigecycline 50 mg q12h; tigecycline 50 mg q12h + other antibiotics	14.5 ± 7.5	66.0	71.0	27.0
Garnacho–Montero (2013)50	Spain	RCS	mixed (VAP 85%, BSI 15%)	MDRa	57	COL; COL + OTH	colistin 2 MIU q8h; colistin 2 MIU q8h + vancomycin 2 g/day	13 ± 7 vs 13 ± 8	58.0	61.5	17.5
Lee (2013)51	Taiwan	RCS	mixed	XDR	386	TGC; SUL + OTH	tigecycline 50 mg q12h; sulbactam 1 g q6h + imipenem 500 mg	10 (7–14)	71.5	69.0	21.0
Batirel (2014)52	Turkey	RCS	BSI	XDR	250	COL; COL + OTH; COL + SUL	colistin; colistin + other antibiotics; colistin + sulbactam	NA	59.0	63.3	18.0
Chuang (2014)53	Taiwan	RCS	HAP, VAP	MDR	168	COL; TGC	colistin 2.5–5 mg/kg/day; tigecycline 50 mg q12h	14 ± 11.2 vs 12.9 ± 9.3	64.0	76.0	21.0
Kalin (2014)54	Turkey	RCS	VAP	XDR	82	COL; COL + SUL	colistin 5–10 mg/kg/day; colistin 5–10 mg/kg/day + sulbactam 3 g q8h	14 (2–22)	57.5	59.0	24.5
Khawcharoenporn (2014)55	Thailand	RCS	HAP, VAP	XDR	166	COL + SUL; COL +  TGC; COL + OTH	colistin 150 mg q12h + sulbactam 6g/day; colistin 150 mg q12h + tigecycline 50 mg q12h; colistin 150 mg q12h + imipenem or meropenem or doripenem 1 g q8h	3 (1–10) vs 3 (0–8) vs 4 (0–8)	75.0	54.3	19.0
Kwon (2014)56	Korea	RCS	mixed (pneumonia 70%, BSI 20%)	XDR	55	COL; TGC	colistin 75–300 mg/day; tigecycline 50–100 mg/day	NA	60.0	59.0	NR
López–Cortés (2014)57	Spain	PCS	mixed (pneumonia 60%)	MDR, XDR	84	COL; OTH; TGC; SUL; COL + TGC; TGC + OTH; COL + TGC + OTH	NA	12 (8–17)	60.0	62.0	NR
Cheng (2015)59	Taiwan	PCS	mixed (pneumonia 40%, BSI 10%)	XDR	55	COL + OTH; COL + TGC	colistin 2.5–5 mg/kg/day + imipenem or meropenem; colistin 2.5–5 mg/kg/day + tigecycline 50 mg q12h	9 (4–14) vs 10 (7–15)	62.0	65.5	NR
Yilmaz (2015)61	Turkey	RCS	VAP	XDR	70	COL; COL + OTH; COL + SUL	colistin 150 mg q12h; colistin 150 mg q12h + meropenem 1 g q8h; colistin 150 mg q12h + sulbactam 1 g q8h	12.3 ± 3.2 vs 11.7 ± 5.6 vs 10.8 ± 4.2	60.0	46.7	NR
Jean (2016)60	Taiwan	PCS	VAP	PDR	84	TGC + OTH; SUL + OTH	tigecycline 50 mg q12h + imipenem 500 mg q6h; sulbactam 1.5–2 g q6h + imipenem 500 mg q6h	NA	76.0	53.5	29.5
Kim (2016)62	Korea	RCS	HAP, VAP	MDR, XDR	70	COL + OTH; TGC + OTH	colistin 150 mg q12h + other antibiotics; tigecycline 50 mg q12h + other antibiotics	12 (9–19) vs 11 (7–15)	70.0	77.5	NR
Zalts (2016)63	Israel	RCS	VAP	XDR	98	COL; SUL	colistin 2 MIU q8h; ampicillin/sulbactam 3 g q6h	NA	53.5	75.0	17.5

PCS, prospective cohort study; RCS, retrospective cohort study; COL, colistin; SUL, sulbactam; TGC, tigecycline; OTH, other antibiotics; iv, intravenously; MIU, million international units; nb, nebulizer; NA, not available; NR, not recorded; vs, versus.

a

Carbapenem resistant and unclear definitions.

b

Neonate population.

Network consistency

The network of eligible comparisons for individual treatment outcomes is shown in Figure 2. The graphical representations of all networks are shown in Figure S2. There was no evidence of global inconsistency in treatment networks except for some loops of all-cause mortality (Table S7), which led us to perform network meta-analysis using an inconsistency model.

Figure 2.

Network comparisons of studies included in the analyses. The size of the nodes and the thickness of the lines are weighted according to the number of studies assessing each treatment and direct comparison. Numbers above and below the lines indicate studies and patients respectively. COL, colistin; SUL, sulbactam; TGC, tigecycline; OTH, other antibiotics.

Treatment outcomes

Treatment effects in pairwise meta-analyses from direct evidence are shown in Table S8. No evidence of statistical heterogeneity was identified in pairwise comparisons for the primary outcomes, as well as for nephrotoxicity and other adverse events.

Clinical cure

The NMA of 17 studies involving 1476 patients evaluated the clinical cure of nine options for treating MDR- and XDR-AB (Figure 2a). No treatment options significantly increased clinical cure rate. The triple therapy consisting of colistin, sulbactam and tigecycline had the highest rank among all treatments compared with colistin in combination with sulbactam (RR 1.17, 95% CI 0.56–2.44), colistin monotherapy (RR 1.28, 95% CI 0.63–2.61) and tigecycline monotherapy (RR 1.38, 95% CI 0.63–3.04) (Figure 3a). The overall findings were consistent with the synthesized data from RCTs (Figure S3). The ranking of treatments based on SUCRAs are shown in Figure S4.

Figure 3.

Network estimated risk ratios (95% CIs) of treatment options for (a) clinical cure and (b) microbiological cure. Treatments are reported in order of efficacy ranking according to SUCRA. Comparisons should be read from left to right. The estimate is located at the intersection of the column-defining treatment and the row-defining treatment. For clinical cure and microbiological cure, an RR value >1 favours the column-defining treatment. Significant results are in bold and underlined. COL, colistin; SUL, sulbactam; TGC, tigecycline; OTH, other antibiotics.

Microbiological cure

Twenty studies involving 1863 patients evaluated the treatment effectiveness related to microbiological cure (Figure 2b). Our NMA revealed that colistin in combination with other antibiotics was significantly associated with a higher microbiological cure rate versus colistin monotherapy (RR 1.21, 95% CI 1.10–1.34), tigecycline in combination with other antibiotics (RR 2.47, 95% CI 1.32–4.62) and tigecycline monotherapy (RR 3.15, 95% CI 1.67–5.94). No significant differences were noted for colistin in combination with other antibiotics versus colistin in combination with sulbactam (RR 1.00, 95% CI 0.89–1.13) and sulbactam in combination with other antibiotics (RR 1.05, 95% CI 0.57–1.93) (Figure 3b). In colistin combination analysis, colistin in combination with rifampicin, sulbactam or other antibiotics, such as carbapenems, was associated with a significantly higher microbiological cure rate compared with colistin monotherapy (Figure S5). Tigecycline in monotherapy or in combination with other antibiotics was associated with a significantly lower microbiological cure rate versus other treatment options (Figure 3b).

All-cause mortality

The NMA of 25 studies involving 2379 patients observing all-cause mortality was included in the analysis (Figure 2c). The combination of colistin with other antibiotics was associated with a significantly decreased all-cause mortality compared with sulbactam in combination with other antibiotics (RR 0.58, 95% CI 0.35–0.96) (Figure 4a). No significant differences were found in the other comparisons.

Figure 4.

Network estimated risk ratios (95% CIs) of treatment options for (a) all-cause mortality, (b) nephrotoxic adverse events and (c) non-nephrotoxic adverse events. Treatments are reported in order of safety ranking according to SUCRA. Comparisons should be read from left to right. The estimate is located at the intersection of the column-defining treatment and the row-defining treatment. For all-cause mortality, nephrotoxic or non-nephrotoxic adverse events, an RR value <1 favours the column-defining treatment. Significant results are in bold and underlined. COL, colistin; SUL, sulbactam; TGC, tigecycline; OTH, other antibiotics.

Nephrotoxic and non-nephrotoxic adverse events

Fifteen studies involving 1342 patients were evaluated for nephrotoxic events (Figure 2d). Tigecycline (RR 0.25; 95% CI, 0.10–0.65) and tigecycline in combination with other antibiotics (RR 0.13; 95% CI, 0.04–0.44) were associated with a significantly lower risk of nephrotoxicity than colistin monotherapy (Figure 4b). No statistically significant difference was noted between treatment options and their effects on non-nephrotoxic adverse events (Figure 4c).

Subgroup and sensitivity analyses and publication bias

The results from subgroup analyses related to type of drug resistance were not different from those of our primary analysis. Colistin in combination with sulbactam or other antibiotics was still associated with a significantly lower microbiological cure rate for all different types of drug resistance. In subgroup analyses related to type of infection, the results of clinical cure were similar to our primary results (Table S9). No treatment options significantly increased clinical cure rate, although the triple therapy of colistin, sulbactam and tigecycline had the highest rank among all treatment options. Results related to microbiological cure of some treatments showed a few differences. Tigecycline monotherapy and tigecycline in combination with other antibiotics had a significantly lower microbiological cure rate than colistin monotherapy in patients with bloodstream infections, whereas no statistically significant differences were found in patients with pneumonia (Table S10). Prespecified sensitivity analyses showed no differences from the main analysis (Table S11). Comparison-adjusted funnel plots showed no sign of asymmetry (Figure S6).

Discussion

This is the first SR and NMA comparing the efficacy and safety of current treatment options for MDR- and XDR-AB-infected patients. The study has several key findings. First, regarding the clinical cure, although no treatment option is significantly more effective than colistin monotherapy, the triple therapy of colistin, sulbactam and tigecycline is ranked highest among all treatment options. Second, the combination of colistin with sulbactam or other antibiotics such as carbapenems is associated with increased microbiological success; however, colistin-induced nephrotoxicity is still a major concern. Third, the most surprising finding from our NMA is that MDR- and XDR-AB-infected patients treated with a combination of colistin and other antibiotics have significantly lower all-cause mortality compared with those patients treated with a combination of sulbactam and other antibiotics. These findings indicate that the combination of colistin with sulbactam or other antibiotics may be the optimal treatment option for MDR- and XDR-AB infection. Finally, although tigecycline-based therapy results in lower nephrotoxicity compared with colistin monotherapy, it is less effective in terms of microbiological cure rate. Therefore, tigecycline should not be used for treating MDR- and XDR-AB infection, especially in patients with BSI.

Similar to previous meta-analyses, our findings suggest that although colistin-based combination therapy does not have better clinical success or all-cause mortality, it offers significantly higher microbiological response than colistin monotherapy.^8,14 Previous in vitro and in vivo studies have also indicated that this combination possesses synergistic effects, which can potentially help to prevent the emergence of colistin-resistant AB strains.^10,11,64,65 Therefore, we support the possibility that colistin-based combination therapy may be used as a first-line regimen for this indication.

Triple therapy with colistin, sulbactam and tigecycline was ranked first in clinical success. However, this result is not statistically significant, probably due to a limited number of studies resulting in insufficient statistical power. High-quality RCTs are needed to confirm this finding.

Tigecycline monotherapy demonstrates a significantly lower microbiological cure rate, even when used in combination with other antibiotics.¹² This is probably due to inadequate tigecycline plasma concentration.⁶⁶ Our findings confirm that tigecycline-based regimens have no additional benefit for treating MDR- or XDR-AB-infected patients with bacteraemia.

Currently, there is still no evidence-based guideline recommending which treatment options clinicians should choose to treat MDR- and XDR-AB infections. In deciding which therapy should be used in combination with colistin, we suggest clinicians consider all clinically relevant factors, including site of infection, patients’ renal and hepatic functions, and comorbidities. In addition, local context issues such as the pattern of drug resistance should also be taken into consideration for choosing the appropriate therapy.

A major strength of our study is the inclusion of a substantially greater number of studies than any previous review. Consequently, this work is the largest completed evaluation of treatment options on efficacy outcomes and adverse events to date. Furthermore, NMA makes comparisons among multiple treatment options possible, even if there is only a small number of available studies. Our review incorporated all eligible studies including RCTs and observational studies, which provide treatment effects in real-world practice and help to improve the generalizability of the overall findings.²⁰

This study has some limitations. First, considering the low prevalence of MDR- and XDR-AB infection and ethical concerns, it is very difficult to conduct prospective RCTs to explore this issue. Therefore, the majority of studies incorporated in this NMA are observational studies, not RCTs. Confounding factors could not be eliminated owing to the inherent nature of observational studies. Thus, the interpretation of our findings should be done with caution. The results do not materially change when restricting the analysis to include only RCTs and high-quality observational studies. Second, the definitions of MDR- and PDR-AB are notably different across studies, especially those published before 2010. To harmonize this, we reclassified the definitions of drug resistance of all included studies with the standard reference.²⁵ Based on sensitivity analysis using a harmonized definition of drug resistance and excluding studies published before 2010, the results are considerably robust across analyses. Third, as we included all studies that evaluated the efficacy of treatment options for MDR- and XDR-AB at any site of infection, it renders the overall treatment effect less clinically applicable. Hence, we separately analysed treatment effects for each primary site of infection. Our findings revealed that the microbiological cure obtained with tigecycline is low in patients with bacteraemia. Finally, we are not able to determine the most effective dose, route of administration and duration of therapy for each treatment owing to small sample size causing insufficient statistical power.

In conclusion, our NMA summarizes the effects of treatment options on clinical and microbiological cures, all-cause mortality and adverse events. The results suggest that there is no significant difference in clinical cure outcomes among treatment options for MDR- and XDR-AB infections. However, colistin-based combination therapy, especially with sulbactam, demonstrates a microbiological benefit with a comparable safety profile to colistin monotherapy. Our research should be regarded as crucial evidence to help formulate clinical decisions in choosing a treatment regimen for treating MDR- and XDR-AB infections. Since the majority of evidence is from observational studies, high-quality RCTs are still needed to confirm our findings.

Acknowledgements

We are grateful to Eric Baehr (The University of Illinois at Chicago College of Pharmacy) for proofreading and English language review. We also thank Phongsathorn Phlaisaithong (School of Pharmaceutical Sciences, University of Phayao) for his technical assistance in the preparation of figures.

Funding

This study was carried out as part of our routine work.

Transparency declarations

None to declare.

Author contributions

K. Kengkla, N. C., A. A. and S. S. conceived and designed the study. K. Kengkla and K. Kongpakwattana selected the articles and extracted the data. K. K. and S. S. analysed the data. K. Kengkla and K. Kongpakwattana wrote the first draft of the manuscript. All authors critically revised the article for important intellectual content and approved the final version. All authors read and met the ICMJE criteria for authorship and agreed with the results and conclusion of this article.

Supplementary data

Supplementary Figure S1 to S6 and Tables S1 to S11 are available as Supplementary data at JAC Online.