What is the first-line treatment for carbapenem-resistant Acinetobacter baumannii?

Published 2026-09-05 · Updated 2026-09-05 · Last checked against sources 2026-09-05

Before I looked it up

I have been hearing talk of durlobactam for a few years now, but unfortunately it has not come into the country. For MRAB treatment we mostly refer to infectious diseases and the antibiotics get used that way, but in the case of MRAB pneumonia I use high-dose ampicillin-sulbactam, colistin nebuliser, and minocycline PO. This regimen is a combination an infectious diseases doctor recommended a few years ago, and at that time the patient got considerably better, and the memory of it still remains vivid. But two or three years have already passed, so I took a look at the current knowledge.

What I found

The short answer is that the guidance moved, and it moved in a direction that makes the drug you cannot get more important rather than less.

• Preferred: sulbactam-durlobactam plus imipenem or meropenem, for invasive CRAB infection.
• Everything else is now bridging therapy, not an alternative. High-dose ampicillin-sulbactam, cefiderocol, minocycline and polymyxin B are positioned as what you give until sulbactam-durlobactam can be started, or when resistance rules it out.
• Nebulised antibiotics are no longer a routine add-on for CRAB pneumonia.
• The general question of combination therapy has been dropped from the guidance altogether.
• Colistin is not the polymyxin that gets named — polymyxin B is, on pharmacokinetic grounds.

The guidance I had was out of date

My note was written against the IDSA 2024 guidance. The current document is the IDSA 2026 Guidance on the Treatment of Antimicrobial-Resistant Gram-Negative Infections, published online in August 2026 and current as of March 2026. Its abstract states plainly that it replaces earlier versions. So the 2023 and 2024 documents are no longer the reference.

One wording change in it is worth noticing before the drug names: the CRAB section now says invasive infection throughout. That is not tidying. CRAB is most often grown from respiratory and wound cultures, where telling colonisation from infection in a ventilated or badly burned patient is the hard part. The suggestions below are for when you have decided it is an infection.

First line: sulbactam-durlobactam plus a carbapenem

Sulbactam-durlobactam 1 g + 1 g IV q6h, infused over 3 hours
  + imipenem or meropenem

Recommendation strength: IDSA 2026 preferred. Evidence is one randomised trial plus in vitro and surveillance data.

What the ATTACK trial actually showed

ATTACK was a non-inferiority trial — phase 3, 59 sites, 16 countries, 181 randomised, 125 with laboratory-confirmed carbapenem-resistant isolates in the primary analysis. Both arms received imipenem-cilastatin as background therapy, so the comparison is sulbactam-durlobactam plus imipenem against colistin plus imipenem.

This is not a 13.2% mortality reduction, and it is worth being precise about that — the confidence interval crosses zero, so superiority was not shown. What the trial established is non-inferiority plus a real difference in nephrotoxicity.

IDSA is candid about the rest of the reasoning: the comparator, colistin plus imipenem, is itself neither preferred nor alternative therapy for CRAB, and survival in other CRAB trials is frequently under 60%. Higher survival against that background is what moved sulbactam-durlobactam to preferred.

Why the carbapenem is added

Sulbactam binds PBP1a/1b and PBP3; imipenem preferentially binds PBP2. The targets do not overlap, and durlobactam protects both agents from hydrolysis. In vitro the combination lowers sulbactam MICs one- to two-fold and produces more than 2-log reductions in time-kill assays — activity not seen with sulbactam-durlobactam alone.

That benefit is still mechanistic, not clinical. IDSA says so and suggests the combination anyway. For infections needing prolonged therapy such as CRAB osteomyelitis, stopping the carbapenem after sustained clinical improvement is described as possibly reasonable, with no data either way.

Where durlobactam stops

Durlobactam is a diazabicyclooctane inhibitor related to avibactam. It inhibits class A, C and D. It has no activity against class B enzymes — the metallo-β-lactamases, NDM among them. NDM-producing CRAB is uncommon in the United States but increasingly reported in South Asia and the Middle East.

• If sulbactam-durlobactam resistance (MIC ≥16/4 µg/mL) or an MBL gene is identified, IDSA prefers non-sulbactam-based regimens — combinations of cefiderocol, minocycline, polymyxin B or tigecycline.
• Over 95% of CRAB isolates in United States surveillance are susceptible, which is the basis for starting empirically while awaiting susceptibility results. That 95% is a United States figure. Local susceptibility should be checked before it is borrowed.

When you cannot get it: bridging, not alternatives

This is the change that matters most if the drug is not stocked where you work. The agents below are not “what you use instead of sulbactam-durlobactam” — they are what you use until it can be started. Unless resistance rules it out, the target regimen stays sulbactam-durlobactam plus a carbapenem.

High-dose ampicillin-sulbactam

Total daily sulbactam 9 g, given as either:
  ampicillin-sulbactam 9 g (ampicillin 6 g + sulbactam 3 g) IV q8h, infused over 4 hours
  OR ampicillin-sulbactam 27 g/day (ampicillin 18 g + sulbactam 9 g) as a 24-hour continuous infusion

Recommendation strength: IDSA 2026 bridging therapy. Evidence is two meta-analyses of mostly observational studies, plus PK/PD and animal models.

The 9 g given every 8 hours is ampicillin-sulbactam, not sulbactam. Of that dose, 3 g is sulbactam. The daily sulbactam total is 9 g, not 27 g. Ampicillin-sulbactam is formulated 2:1, and susceptibility is reported in the same fixed ratio, so a reported MIC of 8/4 µg/mL corresponds to a sulbactam MIC of 4 µg/mL.

Combine it with one of cefiderocol, minocycline or polymyxin B. Ampicillin-sulbactam carries no inhibitor to protect sulbactam from the β-lactamases CRAB produces, so hydrolysis is expected.

Why the dose is pushed this high:

• Only about 10% of CRAB isolates are sulbactam susceptible in surveillance data.
• In neutropenic murine lung models, 4 g/day gave >90% probability of target attainment up to sulbactam MIC ≤4 µg/mL; 9 g/day extended that to MIC 8 µg/mL.
• Susceptibility testing for ampicillin-sulbactam in CRAB misclassifies when anything other than reference broth microdilution is used. Given no safety signal at the higher dose, IDSA favours high-dose prolonged infusion even for isolates reported susceptible.

The supporting clinical evidence is softer than the dosing detail suggests. Two meta-analyses covering more than 20 studies and over 2,000 patients associate sulbactam-containing regimens with better outcomes, but they are dominated by observational studies with heterogeneous regimens. Of five trials with a sulbactam-containing arm, only one showed a statistically significant survival benefit; the rest were numerically higher.

The closest thing to local data. A retrospective cohort at a university-affiliated hospital in South Korea compared a colistin-based regimen against a high-dose ampicillin-sulbactam-based regimen — sulbactam 9 g/day — in 179 patients with CRAB nosocomial pneumonia.

Read the design before the effect size. The two regimens were used in sequential periods — colistin from mid-2021 to mid-2022, ampicillin-sulbactam from late 2022 to late 2023 — not concurrently and not randomised. Anything else that changed between those periods is inside the comparison, and a 61% mortality figure in the colistin arm is high even for this population. Treat it as the direction local practice moved and how it went, not as an effect estimate.

What it does establish is that the sulbactam 9 g/day regimen is deliverable outside a trial, in a setting where sulbactam-durlobactam was not an option.

Cefiderocol

Cefiderocol 2 g IV q8h, infused over 3 hours
  + one of high-dose ampicillin-sulbactam / minocycline / polymyxin B

Recommendation strength: IDSA 2026 alternative, in combination. Trial results remain unresolved.

Over 90% of CRAB isolates are susceptible in vitro, falling to about 60% for NDM producers. The clinical picture is genuinely split:

• CREDIBLE-CR, CRAB subgroup of 54: 28-day survival 51% (20/39) with cefiderocol versus 82% (14/17) with best available therapy — which was mostly polymyxin-based, and is not the same thing as a colistin control arm. Small subgroup, unbalanced at baseline.
• A trial in 47 patients with CRAB pneumonia: 14-day survival 78% (18/23) versus 83% (20/24) with high-dose extended-infusion meropenem. Comparable to an agent with no meaningful CRAB activity is not a reassuring result.
• A trial in 25 patients with CRAB bacteraemia: 30-day survival 55% (6/11) versus 50% (7/14).
• Against that, two meta-analyses of mostly observational data found higher 30-day survival with cefiderocol (58% vs 40%; 62% vs 37%) — but no patient in those comparator arms received sulbactam-durlobactam, and heterogeneity was substantial.

No study has compared cefiderocol with sulbactam-durlobactam directly.

Minocycline, and where tigecycline sits

Minocycline 200 mg IV/PO q12h
  + one of high-dose ampicillin-sulbactam / cefiderocol / polymyxin B

Recommendation strength: IDSA 2026 alternative, in combination. No clinical trials — observational data only.

• Under the revised 2025 CLSI breakpoint (≤1 µg/mL), fewer than 50% of CRAB isolates are susceptible.
• Population PK modelling says 200 mg q12h reliably reaches bacteriostasis targets up to that breakpoint but not 1-log killing targets.
• Nausea and other gastrointestinal intolerance in roughly 20–25%.
• Rapid tissue distribution means low serum and urinary concentrations.

For tigecycline, the dose quoted in the CRAB setting is not the licensed one:

Tigecycline 200 mg IV loading dose, then 100 mg IV q12h

Neither CLSI nor FDA has a tigecycline breakpoint for A. baumannii. That absence is exactly why minocycline is the preferred tetracycline here — not superior data, but interpretable susceptibility criteria. Omadacycline is not suggested unless neither of the other two is available.

Polymyxin B, not colistin

Polymyxin B — dose per the international polymyxin consensus guidelines
  + one of high-dose ampicillin-sulbactam / cefiderocol / minocycline

Recommendation strength: IDSA 2026 alternative, in combination. Reserved on grounds of uncertain effectiveness and toxicity.

• About 85% of CRAB isolates have polymyxin MIC ≤2 µg/mL, but CLSI has no susceptible category for polymyxins against A. baumannii — even ≤2 µg/mL is intermediate.
• Polymyxin B has a more favourable pharmacokinetic profile than colistin, which is why it is the one named.
• Activity in pulmonary epithelial lining fluid is poor. Intravenous polymyxin is not a lung drug.
• The therapeutic window is essentially closed: the concentration needed for 1-log killing, around 2 µg/mL, is also the nephrotoxicity threshold.
• Survival with colistin monotherapy across six trials of roughly 500 patients ranged from 27% to 57%.

The colistin–carbapenem–rifampicin triple is not supported. In 210 patients with extensively drug-resistant A. baumannii, adding rifampicin to colistin did not change 30-day mortality (P = .95); the authors concluded rifampicin should not be routinely combined with colistin. Microbiological eradication was higher, mortality was not. Rifampicin does not appear in the 2026 CRAB section at all.

Nebulised antibiotics are no longer a routine add-on

Recommendation strength: IDSA 2026 explicitly does not suggest routine use.

If you carry “CRAB pneumonia means add aerosolised colistin” as a rule, that rule has been withdrawn.

• Three placebo-controlled randomised trials of adjunctive nebulised therapy — nebulised colistin in 100 patients (65% with A. baumannii pneumonia), nebulised amikacin/fosfomycin in 142 (20%), nebulised amikacin in 508 (29%) — none improved survival, including in the drug-resistant subgroups.
• Subsequent systematic reviews and meta-analyses found no survival benefit either; some reported modest improvement in clinical response.
• The proposed reasons are mechanical: poor penetration into consolidated lung, uneven distribution within infected airways, off-label use of parenteral formulations, and delivery through nebulisers not designed for pulmonary drug delivery.

What remains is a narrow, deliberate use. Selective use may be reasonable where resistance to sulbactam-durlobactam is documented, or where the response to systemic therapy is inadequate — weighed against bronchoconstriction, laryngeal injury and hypersensitivity pneumonitis.

Resistance mechanisms, and what Korea looks like

The defining feature of A. baumannii is that the mechanisms accumulate rather than replace one another.

Enzymatic — ADC cephalosporinases; OXA-type carbapenemases (OXA-23, OXA-24/40, OXA-58); metallo-β-lactamases such as NDM, uncommon in the United States but rising elsewhere.

Non-enzymatic — loss or alteration of porins (CarO, OmpA) limiting carbapenem entry; increased outer membrane hydrophobicity; efflux pump overexpression (AdeABC, AdeIJK); aminoglycoside-modifying enzymes or 16S rRNA methyltransferases, which usually take out the whole class.

In Korea, OXA-23 is effectively the default assumption. In 817 Acinetobacter isolates collected across 16 regions of South Korea from long-term care facilities and general hospitals in 2022–2023, the carbapenem resistance rate among 659 A. baumannii was 87.4% (576 isolates), and every one of those CRAB isolates produced blaOXA-23. Clonal complex 92 predominated.

Resistance to sulbactam-durlobactam itself is poorly characterised — PBP1a/b or PBP3 mutations combined with efflux pump overexpression are suggested. There are no clinical data to guide treatment of sulbactam-durlobactam-resistant CRAB.

Practical points

That last row is the practical form of the guidance dropping its combination-therapy question. Nothing in the 2026 document supports escalating drug count as a default response to failure.

Drugs in development

Preclinical or early clinical, and not decision-grade information.

• OMN6 — an antimicrobial peptide, not a β-lactamase inhibitor. It inhibited growth across multidrug-resistant clinical isolates regardless of resistance mechanism, no resistance emerged on serial passage, and it was effective in murine lung and bloodstream models.
• BWC0977 — broad-spectrum agent in clinical trials.
• Eravacycline — a fluorocycline; IDSA lists a dose (1 mg/kg IV q12h) but makes no CRAB recommendation.

Hypervirulent A. baumannii is accumulating case reports as virulence and resistance converge, but the definition and clinical implications are not settled and no separate treatment strategy is recommended.

Meanwhile CRAB remains a critical priority pathogen on the WHO bacterial priority pathogens list updated in 2024, alongside third-generation cephalosporin-resistant and carbapenem-resistant Enterobacterales.

What I take from it

1. Check the edition first. This area turns over every couple of years, and the 2026 document says outright that it replaces the earlier ones.
2. “9 g sulbactam q8h” does not exist. The 9 g q8h figure is ampicillin-sulbactam; 3 g of it is sulbactam.
3. The nebuliser is no longer routine — but it was not abolished either. It survives for inadequate response to systemic therapy.
4. Rifampicin can come out. It did not change mortality.
5. Susceptibility testing still decides, even though United States surveillance supports starting sulbactam-durlobactam empirically. That surveillance is not our surveillance.

Where this leaves me

So the first-line treatment is sulbactam-durlobactam + carbapenem. But then again, since there is no telling when it will come in.. for the time being I will have to use the ampicillin-sulbactam + minocycline combination. (Colistin, in the IDSA 2026, is written as something to use when the response to systemic therapy is inadequate..)

Sources

Doses, thresholds and rates above are traceable to the following. Anything not tied to one of these is background and is written as background.

1. Tamma PD, Bonomo RA, Heil EL, Justo JA, Satlin MJ, Mathers AJ. Infectious Diseases Society of America 2026 Guidance on the Treatment of Antimicrobial-Resistant Gram-Negative Infections. Clin Infect Dis. 2026 (online ahead of print). doi:10.1093/cid/ciag481 — CRAB section 5; dosing in Table 1. Full text and tables: https://www.idsociety.org/practice-guideline/amr-guidance/
2. Kaye KS, Shorr AF, Wunderink RG, et al. Efficacy and safety of sulbactam-durlobactam versus colistin for the treatment of patients with serious infections caused by Acinetobacter baumannii-calcoaceticus complex (ATTACK): a multicentre, randomised, active-controlled, phase 3, non-inferiority clinical trial. Lancet Infect Dis. 2023;23(9):1072–1084. doi:10.1016/S1473-3099(23)00184-6 — mortality 19% vs 32%, difference −13.2% (95% CI −30.0 to 3.5); nephrotoxicity 13% vs 38%.
3. Bassetti M, Echols R, Matsunaga Y, et al. Efficacy and safety of cefiderocol or best available therapy for the treatment of serious infections caused by carbapenem-resistant Gram-negative bacteria (CREDIBLE-CR): a randomised, open-label, multicentre, pathogen-focused, descriptive, phase 3 trial. Lancet Infect Dis. 2021;21(2):226–240. doi:10.1016/S1473-3099(20)30796-9 — the comparator was best available therapy, not colistin.
4. Durante-Mangoni E, Signoriello G, Andini R, et al. Colistin and rifampicin compared with colistin alone for the treatment of serious infections due to extensively drug-resistant Acinetobacter baumannii: a multicenter, randomized clinical trial. Clin Infect Dis. 2013;57(3):349–358. doi:10.1093/cid/cit253 — n = 210; 30-day mortality unchanged (P = .95).
5. Bae IK, Hong JS. The distribution of carbapenem-resistant Acinetobacter species and high prevalence of CC92 OXA-23-producing Acinetobacter baumannii in community hospitals in South Korea. Infect Drug Resist. 2024;17:1633–1641. doi:10.2147/IDR.S459739 — 817 isolates, 16 regions; 87.4% carbapenem resistance; all CRAB carried blaOXA-23.
6. Lee J, Lee I, Lee K-B, Lee SS. Comparative effectiveness and safety of colistin-based versus high-dose ampicillin/sulbactam-based combination therapy for nosocomial pneumonia caused by carbapenem-resistant Acinetobacter baumannii. Antimicrob Agents Chemother. 2025;69(6):e0188024. doi:10.1128/aac.01880-24 — retrospective, sequential periods; 28-day mortality 20% vs 61%, aRR 0.16 (0.08–0.32).
7. Tsuji BT, Pogue JM, Zavascki AP, et al. International consensus guidelines for the optimal use of the polymyxins. Pharmacotherapy. 2019;39(1):10–39. doi:10.1002/phar.2209 — the polymyxin dosing reference the IDSA table defers to.
8. Michaeli J, Mandel S, Maximov S, et al. In vitro and in vivo antimicrobial activity of the novel peptide OMN6 against multidrug-resistant Acinetobacter baumannii. Antibiotics (Basel). 2022;11(9):1201. doi:10.3390/antibiotics11091201 — OMN6 is an antimicrobial peptide.
9. Wu N, Ma X, Ni W. Hypervirulent Acinetobacter baumannii (hvAB): the convergence of virulence and multidrug resistance. Antibiotics (Basel). 2025;14(6):551. doi:10.3390/antibiotics14060551 — review; definition not yet settled.
10. World Health Organization. WHO bacterial priority pathogens list, 2024. https://www.who.int/news/item/17-05-2024-who-updates-list-of-drug-resistant-bacteria-most-threatening-to-human-health — carbapenem-resistant A. baumannii retained as critical priority.