The relationship between the alveolar epithelial lining fluid concentration of intravenous plus inhaled polymyxin B and clinical efficacy in patients with pneumonia caused by carbapenem-resistant gram-negative bacilli: a prospective cohort study

Abstract

Background

Nebulization combined with intravenous polymyxin B (PMB) for carbapenem-resistant gram-negative bacilli (CRGNB) pneumonia still has a number of failures that may be related to insufficient alveolar epithelial lining fluid (ELF) concentration of PMB. This study aimed at determining the relationship between the alveolar ELF concentration of PMB and clinical efficacy after intravenous (IV) plus inhaled (IH) PMB.

Methods

Seventy-five patients with pneumonia caused by CRGNB were treated with IH plus IV PMB. Alveolar lavage fluid was collected before and after nebulization, and ELF was calculated according to the urea dilution equation. Differences in clinical outcomes between patients with high and low peak and trough concentrations of ELF concentration of PMB were compared separately according to the ROC curve grouping strategy. The primary outcome was favorable clinical outcome. The secondary outcomes included microbiological outcome and time to bacterial eradication, time to renormalize body temperature, CRGNB-related and all-cause mortality, 28-day survival, length of hospitalization, inflammation marker levels and side effects related to PMB.

Results

Clinical efficacy and bacterial clearance were higher in the high ELF peak concentration group than in the low one (total efficacy: 94.44% vs. 48.72%, total bacterial clearance: 63.89% vs. 38.46%, both P < 0.05). The clinical effective rate was higher in the high ELF trough concentration group than in the low one (88.89% vs. 43.33%, P < 0.05). The 28-day survival rate was higher in the high ELF peak and trough concentration group than in the low one (peak: 86.11% vs. 38.46%, trough: 75.56% vs. 40.00%, both P < 0.05).

Conclusions

Patients with high PMB alveolar ELF concentrations demonstrated more favorable clinical outcomes than those with low concentrations.

Trial registration

This trial was registered. The Chinese trial registration number is ChiCTR2100044087. The date of registration is 9-3-2021. The registered name is that clinical study on intravenous drip and aerosol inhalation of polymyxin B for the treatment of pneumonia due to multidrug-resistant gram-negative bacteria.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-025-11515-z.

Background

In today’s clinical healthcare, the emergence and spread of drug-resistant strains of bacteria has become a global public health problem with the generalization of antimicrobial drug use [1]. Currently in China, gram-negative bacilli have become the main causative organisms causing infections, among which the detection rate of carbapenem-resistant gram-negative bacilli (CRGNB), such as Acinetobacter baumannii, Klebsiella pneumoniae, or Pseudomonas aeruginosa has shown a significant increase. The high level of resistance to multiple antimicrobial drugs not only limits treatment options, but also significantly increases healthcare costs and patient mortality [2]. Pneumonia caused by CRGNB infections has become a significant cause of morbidity and mortality globally over the past decade [3, 4]. In 2016, the global associated mortality rate was estimated to be 700,000 cases per year, and this is projected to rise significantly by 2050 [5]. Given its significant public health impact, both the World Health Organization and the Centers for Disease Control have prioritized the risk level of CRGNB [6]. In this context, the search for effective therapeutic strategies has become a research priority.

According to the data released by the China Bacterial Drug Resistance Monitoring Network in the first half of 2023, polymyxin B sulfate (PMB) has a low rate of resistance to carbapenem-resistant Klebsiella pneumoniae, Pseudomonas aeruginosa, and Acinetobacter baumannii, and is highly sensitive. In the face of the increasing rate of CRGNB infection in recent years and the limitations of the effectiveness of traditional antimicrobial drugs, PMB has been regarded as the “last line of defense” against CRGNB infection [7–9].

When PMB is injected intravenously for the treatment of CRGNB pneumonia, it binds to the phospholipid molecules on the outer membrane of Gram-negative bacteria, especially to the phosphate ester portion of lipopolysaccharide, and has a mechanism of “spontaneous uptake”, which destroys the integrity of the bacterial cell membrane and increases the permeability of the cell membrane, leading to the leakage of the cell contents and the death of the bacteria [10]. In addition, the binding of PMB to the cell membrane can also disturb the polarity and fluidity of the membrane, further disrupting the life activities of bacteria; due to the disruption of the cell membrane and the leakage of the contents, PMB can also indirectly inhibit the normal growth and metabolic processes of bacteria [11]. In our previous clinical study, we found that the combination of PMB intravenous (IV) plus inhaled (IH) administration was more effective than IV alone in the treatment of multidrug-resistant G- bacterial pneumonia without increasing the risk of renal damage [12]. This may be related to alveolar epithelial lining fluid (ELF) concentrations. It is well known that alveolar ELF is the main site of infection for most disease-causing bacteria, and the concentration of a drug in alveolar ELF directly affects its therapeutic efficacy for lung infections [13]. However, the presence of an alveolar capillary barrier and the physicochemical properties of macromolecular hydrophilic divalent cations of PMB make it difficult to effectively penetrate the tight cellular barrier of the lungs by IV administration only, resulting in relatively low drug concentrations in the alveolar ELF. It has been shown that drug concentrations in ELF are significantly lower than those in plasma when using IV PMB alone to treat CRGNB pneumonia [14]. Pharmacokinetic/pharmacodynamic studies in a mouse model of lung infection by Klebsiella et al. showed low penetration of PMB into the lung parenchyma with IV use [15].

Nebulized inhalation, as a mode of drug delivery, can turn drugs into a mist through a high-speed airflow and act directly on the lungs, increasing the concentration of drugs in the lungs and reducing systemic adverse effects, thus potentially improving therapeutic efficacy [16–19]. Therefore, clinicians are increasingly using nebulized antimicrobial drugs as a therapeutic option for respiratory tract infections in mechanically ventilated patients [20]. Currently, several domestic and international guidelines and consensus consider that the potential benefits of nebulized inhaled PMB outweigh the risks, and recommend nebulized inhaled PMB as one of the important therapeutic approaches for CRGNB-induced pneumonia [21, 22].

However, whether nebulized PMB can be used as an adjunct or alternative strategy for the treatment of CRGNB pneumonia remains controversial. This is due to the fact that nebulization combined with intravenous PMB for CRGNB pneumonia still has a number of failures that may be related to insufficient alveolar ELF concentration of PMB. Moreover, too high a concentration of PMB in alveolar ELF may cause pulmonary toxicity and too low a concentration may induce bacterial resistance [23, 24]. Although several studies have now shown that nebulized inhaled PMB improves clinical outcomes in patients with CRGNB pneumonia, however, the relationship between its concentration in the alveolar ELF and clinical outcomes is unclear.

Therefore, this study included patients with CRGNB pneumonia who underwent IV combined IH PMB treatment, observed the relationship between drug concentration in alveolar ELF and clinical efficacy, and explored the role of nebulized inhalation PMB in the treatment of CRGNB pneumonia, aiming to provide a theoretical basis for the implementation of the nebulized inhalation PMB treatment strategy in patients with CRGNB pneumonia.

Materials and methods

Study design and patient population

This was a prospective observational cohort study also a post hoc analysis of an RCT that was performed at the Intensive Care Unit (ICU) in Fujian Medical University Union Hospital between Feb 2022 and Feb 2024 (clinical trial registration number: ChiCTR2100044087). A total of 75 eligible adult patients were included. The protocol was approved by the Ethics Committee of Fujian Medical University Union Hospital (No.2020KY0136/2022YF033-03). Informed consent was obtained from all participating patients or their relatives, and all aspects of the study complied with the Declaration of Helsinki.

Identified patients were included in this study if they met the following inclusion criteria: (1) the diagnosis of pneumonia due to mono-microbial infection following the Chinese Thoracic Society guidelines [25]; (2) CRGNB in sputum culture that was sensitive to PMB; (3) at least 3 days of intravenous combined nebulized PMB; and (4) patients with artificial airways. Patients were excluded if they (1) were judged to be imminently dying; (2) suffered from severe hepatic (Child-Pugh score C) and renal (creatinine clearance < 30 mL/min) insufficiency; (3) Combination of mixed infections at sites other than the lungs, such as intracranial or abdominal sites; (4) cystic fibrosis; (5) lung transplantation; (6) less than 3 days of PMB use due to death or automatic discharge.

All enrolled patients were treated with IV combined IH PMB for around two weeks or until death (less than 2 weeks). IV PMB was administered at a dose of 1.25 mg/kg every 12 h after a loading dose of 2.0 mg/kg over 1 h, and IH PMB was administered at 25 mg every 12 h. Simultaneous administration of IH and IV PMB was at 12-hour intervals. All patients were administered this regimen. The standard nebulization procedures were implemented. Thirty minutes before nebulization, β2 receptor agonist was used to induce bronchodilation, and the sputum was exhausted in the airway. PMB was delivered every 12 h via a vibrating mesh nebulizer, installed upstream of the inspiratory Limb, for 20 min or until the nebulized solution container was empty, and a SIMV(PC) + PSV mode with 8 ml/kg of constant inspiratory flow, 18 times/min of respiratory rate, 1:2 of inhaling-to-breathing ratio, and 6 cm H₂O of positive end expiratory pressure was selected. A 20% duty cycle end-inspiratory pause was set, and the humidification system was turned off during inhalation and turned back on at the end of nebulization. The expiratory filter was changed after each nebulization.

Grouping

The study used a grouping strategy based on the receiver operating characteristic curve (ROC curve) to categorize patients into high ELF peak concentration, low ELF peak concentration, high ELF trough concentration, and low ELF trough concentration based on ELF peak concentration and ELF trough concentration.

Endpoints

The primary outcome was favorable clinical outcome after PMB medication course. Microbiological outcome and time to bacterial eradication, time to renormalize body temperature, CRGNB-related and all-cause mortality, 28-day survival, length of hospitalization, inflammation marker levels and side effects related to PMB were selected as secondary endpoints.

Sampling and date collection

Ethylenediaminetetraacetic acid-treated blood samples (3 mL) and bronchoalveolar lavage fluid (BALF) were collected at 1 h and 12 h after nebulization when PMB was administrated at least four doses. Because, based on Boisson’s study, ELF reaches peak concentration 1 h after nebulizing polymyxin, and ELF reaches trough concentration before the next nebulization [26]. Before executing alveolar lavage, the patient’s CT report was checked and the diseased Lung segment for lavage was selected. Lavage should be conducted at the same site with the saline lavage of 20 ml three times. The first tube of lavage fluid including mucous sputum should be discarded, and the combined volume of the second and third tubes of alveolar lavage fluid should be obtained [27]. Blood and BALF samples were centrifuged (4 °C, 3000 rpm × 10 min). The supernatant plasma and BALF were stored at −80 °C until further analysis.

Information was collected from electronic medical records, including demographic data, clinical data (primary diagnosis, microbiological diagnosis of CRGNB-induced pneumonia, comorbidities, Acute Physiology, and Chronic Health Evaluation (APACHE II) score, Sequential Organ Failure Assessment (SOFA) score, Comorbidity with sepsis, duration of mechanical ventilation, clinical outcomes, microbiological outcomes, CRGNB-related and all-cause mortality after ICU admission, 28-day survival, length of hospitalization), laboratory tests including inflammatory indicators, body temperature, pathogens and antimicrobial susceptibility, PMB administration information, PMB concentrations in ELF and plasma, concomitant use of other antibiotics (within 7 days of PMB) and immunosuppressive therapy (within 28 days of PMB), and drug-related side effects.

Bioanalysis

PMB concentrations in plasma and bronchoalveolar lavage were determined using ultrafast liquid chromatography (Jasper™)-tandem mass spectrometry (AB SCIEX Triple Quad™4500MD) method as previously published [28]. Briefly, protein precipitation was conducted by adding 0.1% formic acid-acetonitrile. The sample was subsequently separated on a Shim-pack GSP-HPLC C₁₈ column and detected within 4 min by the mass spectrometer in the positive mode coupled with multiple reaction monitoring.

Actual PMB concentrations in ELF (PMB ELF) were estimated from BAL concentrations (PMB BALF) using urea as an endogenous marker of ELF dilution as described previously [13, 29]: PMB ELF = PMB BALF × urea dilution index (urea plasma/urea BALF). An automated analyzer (AU 5800, Beckman Coulter, Indianapolis, IN, USA) was used to measure the levels of urea concentrations in plasma and BALF.

Definition

Clinical outcome was classified as clinical cure (the clinical symptoms, signs, and laboratory test results of patients were returned to normal or pre-infection status after treatment), clinical improvement (the clinical symptoms, signs, and laboratory test results of patients were improved after treatment), clinical failure (the clinical symptoms, signs, and laboratory test results of patients did not improve significantly, and even aggravated or switched to other treatment protocols after treatment). Clinical cure or clinical improvement was considered favorable clinical outcomes, while clinical failure was considered unfavorable clinical outcomes [14]. Clinical efficacy % = (number of clinical cured cases + number of clinical improvement cases)/total number of cases × 100%. The data were independently evaluated by two physicians who were blinded to the group. If there were discrepancies, a third investigator was consulted to resolve differences and reach an agreement.

The presence of microbial infection was considered in two consecutive BALF specimens with a colony count of more than 10⁴ CFU/mL at 24 h intervals prior to drug administration. Quantitative microbiological culture of BALF from the same lesion sampled by fiberoptic bronchoscopy alveolar lavage about every 2 days from the start of drug administration. The microbiological outcome was defined as bacterial eradication (sputum culture-confirmed eradication of the pathogen for three consecutive times), bacterial replacement (sputum culture positive switching from original pathogen to other pathogens for three consecutive times), bacterial persistence (sputum culture-confirmed persistent existence of the original pathogen with progressive or persistent symptoms and signs of infection), bacterial recurrence (regrowth of the same pathogen with symptoms and signs of infection during hospitalization). Total bacterial clearance % = (number of bacterial eradication cases + number of bacterial replacement cases)/total number of cases × 100%.

All-cause mortality was assessed during a follow-up period of up to 28 days. CRGNB-related mortality was identified as death closely associated with pneumonia, respiratory failure, or septic shock [30].

Nephrotoxicity was defined as an episode of acute kidney injury (AKI) in this study using the criteria provided by the Kidney Disease Improving Global Guidelines, where AKI was defined as an increase in serum creatinine levels of ≥ 0.3 mg/dL within 48 h; or 1.5 - < 2-fold over the baseline value within 7 days (grade 1), 2 - < 3-fold (grade 2), and 2 - < 3-fold baseline (grade 3); or a urine volume of < 0.5 mL/kg/h for 6 h [31]. The darkening of skin was assessed by the investigator using the Fitzpatrick scale [32]. For inhalational administration of treatment, significant bronchospasm should be considered if respiratory wheezes were observed and defined as ≥ 15% decrease in absolute forced expiratory volume with local irritation and coughing during the IH PMB treatment period.

To adjust for the effect of dose adjustments, concentration data were dose-normalized for subsequent analysis. Serum levels of inflammatory cytokines, such as procalcitonin (PCT), interleukin-6 (IL-6) and C-reactive protein (CRP), were recorded before and after administration of PMB. Time to renormalize body temperature referred to the time from enrollment to fever resolution (body temperature ≤ 37.2℃ for ≥ 24 h). Time to bacterial eradication was calculated as the time of the inability to culture the original pathogen isolated from the patients.

Statistics analysis

The sample size was determined using PASS 2021 (version: 21.0.9, NCSS, LLC. Kaysville, Utah, USA. www.ncss.com). SPSS 25.0 software was used for statistical analysis of the data. Spearman’s rank correlation coefficient was used to assess the linear relationship between drug concentration and efficiency, and then a binary logistic regression analysis model was further used to determine that the PMB ELF concentration was an independent factor affecting the clinical efficiency, and the ROC of PMB ELF concentration versus clinical efficiency was plotted to determine the sensitivity and specificity at different concentration cutoffs. The actual number of total effective cases and the number of ineffective cases were then tested for consistency with the predicted number of total effective cases and the number of ineffective cases. An optimal PMB ELF concentration cutoff point was selected based on the shape of the ROC curve and the predefined study objectives. This cut-off point was defined as the point that maximizes the sum of sensitivity and specificity.

Continuous quantitative data were tested for normality using the Shapiro Wilks test. Quantitative information that conformed to normal distribution was expressed as mean ± standard deviation ( ± S), and comparisons between groups were performed using the independent samples t test. Data that did not fit the normal distribution were expressed as median and upper and lower quartiles [M (P25, P75)], and comparisons between groups were made using the Wilcoxon signed rank sum test for two independent samples. Categorical variables were expressed as frequencies or percentages and were compared using the chi-square test or Fisher’s exact probability method. Variables suggesting statistically significant differences in the comparison of clinical characteristics between the two groups were subjected to binary logistics regression analysis to screen for independent influences on clinical efficacy and 28-day survival. Differences were considered statistically significant when P < 0.05.

Results

Patient demographics and baseline clinical characteristics

The primary outcome measure was favorable clinical outcome. According to the proportions sample size calculation method, which is a test of superiority, with α = 0.025 and 1-β = 0.9 test level, a test was performed according to 1:1 grouping, taking into account the shedding rate of about 10%, a total of 36 patients need to be included in the two groups of ELF peak concentration, 18 cases in each group; a total of 40 patients need to be included in the two groups of ELF trough concentration, 20 patients in each group. Seventy-five cases were included in this study and the sample size was sufficient.

The study used a grouping strategy based on the ROC curve to categorize patients into high ELF peak concentration (ELF > 19.01 µg/mL, n = 36), low ELF peak concentration (ELF ≤ 19.01 µg/mL, n = 39), high ELF trough concentration (ELF > 1.275 µg/mL, n = 45), and low ELF trough concentration (ELF ≤ 1.275 µg/mL, n = 30). Detailed information is demonstrated in Table 1 and Fig. 1.

Table 1.

Evaluation of the efficacy of predicting clinical efficacy with ELF peak concentrations and ELF trough concentrations

	ELF Peak Concentration	ELF Trough Concentration
AUC	0.922	0.921
P-value	0.029*	0.030*
95% Confidence Interval	0.865–0.980	0.862–0.979
Sensitivity	72.92%	85.42%
Specificity	96.30%	85.19%
Youden Index	69.21%	70.60%
Optimal Threshold	19.01	1.275
Positive Likelihood Ratio	19.69	5.77
Negative Likelihood Ratio	0.2813	0.1712
Positive Predictive Value	97.22%	91.11%
Negative Predictive Value	66.67%	76.67%

*P < 0.05 indicates statistically significant differences

Fig. 1.

ROC curves for predicting clinical efficacy with ELF peak concentrations and ELF trough concentrations

Descriptive characteristics are summarized in Tables2,3,4and5 and Figure S1 ~ 2. Only the differences in SOFA scores and APACHE-II scores were statistically significant between the high and low ELF peak concentration groups (P < 0.05) (Table 2), whereas the differences in gender, age, cause of admission to the ICU, major pathogens of infections, comorbid sepsis, duration of PMB use, mean PMB blood concentration, history of use of combined glucocorticoids, and duration of mechanical ventilation were not statistically significant (P > 0.05) (Tables 2and3). There was no significant difference in the antibiotics combined between the two groups (Fisher’s exact test: P = 0.142) (Figure S1).

Table 2.

Comparison of clinical characteristics between two groups based on ELF peak concentrations

Variable	High ELF Peak Concentration Group (n = 36)	Low ELF Peak Concentration Group (n = 39)	P-value
Age (years)	63.94 ± 17.19	65.15 ± 15.55	0.750
Gender			0.083
Male (n)	27(75.00%)	36(92.31%)
Female (n)	9(25.00%)	3(7.69%)
Reason for ICU Admission
Acute respiratory failure	28(77.78%)	25(64.10%)	0.194
Multiple trauma	1(2.78%)	3(7.69%)	0.616
Postoperative	9(25.00%)	8(20.51%)	0.643
Other	3(8.33%)	9(23.08%)	0.082
Comorbidities
Cardiovascular diseases	21(58.33%)	25(64.10%)	0.608
Cerebrovascular diseases	15(41.67%)	13(33.33%)	0.456
Chronic pulmonary diseases	4(11.11%)	5(12.82%)	> 0.999
Chronic liver diseases	2(5.56%)	0(0%)	0.227
Diabetes	12(33.33%)	10(25.64%)	0.465
Chronic kidney disease	5(13.89%)	5(12.82%)	> 0.999
Solid tumor	10(27.78%)	12(30.77%)	0.776
Hematologic malignancy	1(2.78%)	2(5.13%)	> 0.999
SOFA score	5.61 ± 2.17	7.77 ± 3.94	0.005*
APACHE-II score	17.61 ± 6.30	22.15 ± 6.91	0.004*
Sepsis	20(55.56%)	28 (71.79%)	0.143
PMB Usage Duration (days)	13.00(10.00,21.00)	11.00(7.00,16.00)	0.115
Mean PMB Blood Concentration (mg/L)	2.62(1.30,3.00)	2.27(1.35,2.97)	0.501
PMB Concentration Range 2 ~ 4 mg/L (n)	22(61.11%)	23(58.97%)	0.850
PMB Concentration < 2 or > 4 mg/L (n)	14(38.89%)	16(41.03%)	0.850
History of glucocorticoid use (n)	6(16.67%)	10(25.64%)	0.343
Duration of Mechanical Ventilation (days)	9.50(5.00,13.50)	10.00(6.00,12.00)	0.920

*P < 0.05 indicates statistically significant differences

Table 3.

Distribution and related analysis of main pathogens in relation to ELF peak concentrations

Bacterial Type	Total (Total Bacteria = 112)	High ELF Peak Concentration Group (Total Bacteria = 55)	Low ELF Peak Concentration Group (Total Bacteria = 57)	P-value
Carbapenem-resistant Klebsiella pneumoniae (n)	32(28.57%)	17(30.91%)	15(26.32%)	0.591
Carbapenem-resistant Escherichia coli (n)	1(0.89%)	0(0%)	1(1.75%)	> 0.999
Carbapenem-resistant Pseudomonas aeruginosa (n)	39(34.82%)	18(32.73%)	21(36.84%)	0.648
Carbapenem-resistant Acinetobacter baumannii (n)	40(35.71%)	20(36.36%)	20(35.09%)	0.888

*P < 0.05 indicates statistically significant differences

Table 4.

Comparison of clinical characteristics between two groups based on ELF trough concentrations

Variable	High ELF Trough Concentration Group (n = 45)	Low ELF Trough Concentration Group (n = 30)	P-value
Age (years)	66.00(55.50,77.50)	68.00(54.00,76.25)	0.832
Gender			0.108
Male (n)	35(77.8%)	28(93.33%)	-
Female (n)	10(22.22%)	2(6.67%)	-
Reason for ICU Admission
Acute respiratory failure	33(73.33%)	20(66.67%)	0.534
Multiple trauma	1(2.22%)	3(10.00%)	0.295
Postoperative	11(24.44%)	6(20.00%)	0.652
Other	6(13.33%)	6(20.00%)	0.526
Comorbidities
Cardiovascular diseases	28(62.22%)	18(60.00%)	0.846
Cerebrovascular diseases	18(40.00%)	10(33.33%)	0.559
Chronic pulmonary diseases	4(8.89%)	5(16.67%)	0.470
Chronic liver diseases	2(4.44%)	0(0%)	0.514
Diabetes	15(33.33%)	7(23.33%)	0.351
Chronic kidney disease	5(11.11%)	5(16.67%)	0.508
Solid tumor	12(26.67%)	10(33.33%)	0.534
Hematologic malignancy	1(2.22%)	2(6.67%)	0.560
SOFA score	5.91 ± 2.73	7.97 ± 3.88	0.009*
APACHE-II score	18.40 ± 6.69	22.33 ± 6.80	0.016*
Sepsis	27(60.00%)	21(70.00%)	0.377
PMB Usage Duration (days)	12.00(10.00,21.00)	10.50(6.75,14.50)	0.109
Mean PMB Blood Concentration (mg/L)	2.62(1.30,3.00)	2.27(1.35,2.97)	0.165
PMB Concentration Range 2 ~ 4 mg/L (n)	28(62.22%)	17(56.67%)	0.630
PMB Concentration < 2 or > 4 mg/L (n)	17(37.78%)	13(43.33%)	0.630
History of glucocorticoid use (n)	9(20.00%)	7(23.33%)	0.730
Duration of Mechanical Ventilation (days)	10.00(5.00,14.50)	9.00(5.75,11.00)	0.461

*P < 0.05 indicates statistically significant differences

Table 5.

Distribution and related analysis of main pathogens in relation to ELF trough concentrations

Bacterial Type	Total (Total Bacteria = 112)	High ELF Trough Concentration Group (Total Bacteria = 45)	Low ELF Trough Concentration Group (Total Bacteria = 67)	P-value
Carbapenem-resistant Klebsiella pneumoniae (n)	32(28.57%)	15(33.33%)	17(25.37%)	0.361
Carbapenem-resistant Escherichia coli (n)	1(0.89%)	0(0%)	1(1.49%)	> 0.999
Carbapenem-resistant Pseudomonas aeruginosa (n)	39(34.82%)	13(28.89%)	26(38.81%)	0.280
Carbapenem-resistant Acinetobacter baumannii (n)	40(35.71%)	17(37.78%)	23(34.33%)	0.709

*P <0.05 indicates statistically significant differences

 Only the differences in SOFA scores and APACHE-II scores were statistically significant between the high and low ELF trough concentration groups (P < 0.05) (Table 4), whereas the differences in gender, age, cause of admission to the ICU, major pathogens of infections, comorbid sepsis, duration of PMB use, mean PMB blood concentration, history of use of combined glucocorticoids, and duration of mechanical ventilation were not statistically significant (P > 0.05) (Tables 4 and 5).There was no significant difference in the antibiotics combined between the two groups (Fisher’s exact test: P = 0.674) (Figure S2).

Outcomes and adverse events in the ELF peak concentration groups

Favorable clinical outcome and logistics regression analysis

The total effective rate was 94.44% in the high ELF peak concentration group and 48.72% in the low ELF peak concentration group, and the total effective rate in the high ELF peak concentration group was significantly higher than that in the low one (P < 0.05) (Table S1). The SOFA score was 5.61 ± 2.17 in the high ELF peak concentration group and 7.77 ± 3.94 in the low ELF peak concentration group, and the APACHE-II score was 17.61 ± 6.30 in the high ELF peak concentration group and 22.15 ± 6.91 in the low ELF peak concentration group. Only SOFA score and APACHE-II score differences were statistically significant in the comparison of baseline data between the two groups (P < 0.05). Clinical efficacy (cure + effective, ineffective) as the dependent variable, ELF peak concentration (high and low concentration) and APACHE-II score, SOFA score as the independent variables were analyzed in binary logistic regression analysis, and the results suggested that the ELF peak concentration (high and low concentration) was an independent factor affecting the clinical efficacy (P < 0.05), and the APACHE-II score, SOFA score were not statistically significant (P > 0.05) (Table S2).

Microbiological outcome and time to bacterial eradication

The total clearance rate was 63.89% in the high ELF peak concentration group and 38.46% in the low ELF peak concentration group. The total clearance rate in the high ELF peak concentration group was significantly higher than that in the low one, and the difference between the two groups was statistically significant (P < 0.05), but the difference in the bacterial clearance/replacement time between the two groups was not statistically significant (P > 0.05) (Table S3).

Time to renormalize body temperature

In the high ELF peak concentration group, there were 14 cases (38.89%) of no fever, 12 cases (33.33%) of temperature recurrence after drug administration, and 10 cases (27.78%) of temperature non-recovery after drug administration, of which the average time of recurrence was 6.33 ± 4.16 in the 12 cases of temperature recurrence; and in the low ELF peak concentration group, there were 6 cases (15.38%) of no fever, and 21 cases (53.85%) of body temperature recurrence after drug administration, and 12 cases (30.77%) of temperature not recovered after drug administration, of which the average time of temperature recovery in the 21 cases was 7.76 ± 5.55. The number of cases of no fever in the high ELF peak concentration group was greater than that in the low ELF peak concentration group, and the difference between the two groups was statistically significant (P<0.05), and the differences in the number of cases of temperature recovery, the number of cases of temperature not recovered, and the time of temperature recovery between the two groups were not statistically significant (P > 0.05) (Table S4).

Mortality, 28-day survival and length of hospitalization

The all-cause mortality rate was 25.00% in the high ELF peak concentration group and 61.54% in the low ELF peak concentration group, and the all-cause mortality rate was significantly lower in the high ELF peak concentration group than in the low one, and the difference between the two groups was statistically significant (P < 0.05). The difference in CRGNB-related mortality between the high ELF peak concentration group and the low ELF peak concentration group was not statistically significant (P > 0.05). The 28-day survival rate was 86.11% in the high ELF peak concentration group and 38.46% in the low ELF peak concentration group, and the 28-day survival rate in the high ELF peak concentration group was significantly higher than that in the low one, and the difference between the two groups was statistically significant (P < 0.05). The difference in hospitalization time and ICU stay between the high ELF peak concentration group and the low ELF peak concentration group was not statistically significant (P > 0.05, partly because of the unavailability of beds in the general ward) (Table S5 and Fig. 2).

Fig. 2.

28-day survival curves of two patient groups categorized by ELF peak concentrations

Binary logistic regression analysis of 28-day survival rate

The difference between the baseline data of the two groups was statistically significant only for SOFA score and APACHE-II score (P < 0.05). The 28-day survival status (survival, death) as the dependent variable, ELF peak concentration (high and low concentration) and APACHE-II score, SOFA score as the independent variables were analyzed in binary logistic regression analysis, and the results suggested that the ELF peak concentration (high and low concentration), SOFA score were the independent factors affecting the 28-day survival rate (P < 0.05), while APACHE- II score was not statistically significant (P > 0.05) (Table S6).

Inflammation marker levels

There was no statistically significant difference in the comparison of serum PCT levels before and after PMB treatment in the high ELF peak concentration group (P > 0.05), and the levels of IL-6 and CRP after PMB treatment in the high ELF peak concentration group were significantly reduced compared with those before treatment (both P < 0.05); the comparison of serum PCT, IL-6, and CRP levels before and after PMB treatment in the low ELF peak concentration group were not statistically significant (P > 0.05) (Table S7).

Side effects related to PMB

The difference in the incidence of bronchospasm, nephrotoxicity and skin pigmentation during PMB treatment between the high ELF peak concentration group and the low ELF peak concentration group was not statistically significant (P > 0.05) (Table S8).

Outcomes and adverse events in the ELF Valley concentration groups

Favorable clinical outcome and logistics regression analysis

The total effective rate was 88.89% in the high ELF trough concentration group and 43.33% in the low ELF trough concentration group, and the total effective rate was significantly higher in the high ELF trough concentration group than in the low one (P < 0.05) (Table S9). The SOFA score was 5.91 ± 2.73 in the high ELF trough concentration group and 7.97 ± 3.88 in the low ELF trough concentration group, and the APACHE-II score was 18.40 ± 6.69 in the high ELF trough concentration group and 22.33 ± 6.80 in the low ELF trough concentration group. The differences in baseline data between the two groups were statistically significant only in SOFA score and APACHE-II score (P < 0.05). Clinical efficacy (cure + effective, ineffective) as the dependent variable, ELF trough concentration (high and low concentration) and APACHE-II score, SOFA score as the independent variables were analyzed by binary logistic regression analysis, and the results suggested that the ELF trough concentration (high and low concentration) was an independent factor affecting the clinical efficacy (P < 0.05), and the APACHE-II score and SOFA score were not statistically significant (P > 0.05) (Table S10).

Microbiological outcome and time to bacterial eradication

The total clearance rate was 57.78% in the high ELF trough concentration group and 40.00% in the low ELF trough concentration group, and the difference in total clearance rate and bacterial clearance/replacement time between the two groups was not statistically significant (P > 0.05) (Table S11).

Time to renormalize body temperature

In the high ELF trough concentration group, there were 13 cases (28.89%) of no fever, 18 cases (40.00%) of temperature recurrence after drug administration, and 14 cases (31.11%) of temperature non-recovery after drug administration, of which the mean time of temperature recurrence was 7.22 ± 5.55 among the 18 cases; and in the low ELF trough concentration group, there were 7 cases (23.33%) of no fever, and 15 cases (50.00%) of body temperature recurrence after drug administration, and 8 cases (26.67%) of body temperature non-recovery after drug administration, in which the average time of recovery was 7.27 ± 4.62 in 15 cases of body temperature recovery. The differences in the number of cases of non-febrile cases, number of cases of temperature recovery, number of cases of body temperature non-recovery, and the time of temperature recovery between the two groups were not statistically significant (P > 0.05) (Table S12).

Mortality, 28-day survival and length of hospitalization

The all-cause mortality rate was 33.33% in the high ELF trough concentration group and 60.00% in the low ELF trough concentration group. The all-cause mortality was lower in the high ELF trough concentration group than in the low one, and the difference between the two groups was statistically significant (P < 0.05). The difference in CRGNB-related mortality between the high ELF trough concentration group and the low ELF trough concentration group was not statistically significant (P > 0.05). The 28-day survival rate was 75.56% in the high ELF trough concentration group and 40.00% in the low ELF trough concentration group. The 28-day survival rate was significantly higher in the high ELF trough concentration group than in the low one (P < 0.05). The difference in hospitalization time and ICU stay between the high ELF trough concentration group and the low ELF trough concentration group was not statistically significant (P > 0.05, partly because of the unavailability of beds in the general ward) (Table S13 and Fig. 3).

Fig. 3.

28-day survival curves of two patient groups categorized by ELF trough concentrations

Binary logistic regression analysis of 28-day survival rate

The differences in baseline data between the two groups were statistically significant only in SOFA score and APACHE-II score (P < 0.05). The 28-day survival status (survival, death) as the dependent variable, ELF trough concentration (high and low concentration) and APACHE-II score, SOFA score as the independent variables were analyzed in binary logistic regression analysis, and the results suggested that the ELF trough concentration (high and low concentration), SOFA score were the independent factors affecting the 28-day survival rate (P < 0.05), while APACHE- II score was not statistically significant (P > 0.05) (Table S14).

Inflammation marker levels

There was no statistically significant difference in the comparison of serum PCT levels before and after PMB treatment in the high ELF trough concentration group (P > 0.05), and the levels of IL-6 and CRP after PMB treatment in the high ELF trough concentration group were significantly reduced compared with those before treatment (both P < 0.05); the comparison of serum PCT, IL-6, and CRP levels before and after PMB treatment in the low ELF trough concentration group were not statistically significant (P > 0.05) (Table S15).

Side effects related to PMB

The difference in the incidence of bronchospasm, nephrotoxicity and skin pigmentation during PMB treatment between the high ELF trough concentration group and the low ELF trough concentration group was not statistically significant (P > 0.05) (Table S16).

Discussion

Due to the fact that IH combined with IV PMB for CRGNB pneumonia still has a number of failures that may be related to insufficient alveolar ELF concentration of PMB, our study collected and analyzed clinical data from 75 patients with CRGNB pneumonia treated with IV and IH of PMB, and explored the relationship between the concentration of PMB in the alveolar ELF and its clinical efficacy. In this analysis, we found that (1) The PMB ELF high-concentration group resulted in better clinical outcomes, higher bacterial clearance and 28-day survival without an increased risk of adverse events; (2) PMB ELF concentrations were positively correlated with clinical outcomes and were independent predictors of clinical outcomes; (3) PMB ELF peak and trough concentrations, and SOFA score were independent predictors of 28-day survival.

The global emergence of CRGNB is a challenging clinical and public health problem that is increasing mortality and medical costs [3, 4]. As a classical peptide antibiotic, PMB has been re-recognized and applied to pneumonia caused by some refractory CRGNB in recent years [33]. Clinically, intravenous drip is the common mode of administration of PMB, however, the mean blood concentration of PMB does not directly affect its clinical efficacy during the treatment of CRGNB pneumonia, based on the fact that alveolar ELF is a common site of infection for most extracellular pathogens such as Klebsiella pneumoniae, Pseudomonas aeruginosa, Acinetobacter baumannii, Escherichia coli, and so on [13]. Therefore, attention should be paid to the concentration of PMB in alveolar ELF when treating CRGNB pneumonia, because the concentration in alveolar ELF is a better indicator of the anti-infective effect than the blood concentration [34]. Due to the high molecular weight of PMB, it is difficult for the drug to effectively penetrate the air-blood barrier, resulting in lower than expected concentrations in the alveolar ELF [35]. And nebulized inhalation, as a special way of drug delivery, can directly deliver the drug to the lungs, so that the drug can reach the ideal bactericidal concentration in the target site, thus improving the effect of drug treatment [36, 37].

In recent years, several clinical studies have also confirmed that the combination dosing strategy of PMB (IV combined with IH) has demonstrated superior therapeutic efficacy compared to single IV dosing [38–42].The International Consensus Guidelines for the Optimal Use of Polymyxins [22], the Infectious Disease Society of America and the American Thoracic Society [43], and Multi-Disciplinary Expert Consensus on the Optimal Clinical Use of Polymyxins in China [21] have developed evidence-based guidelines for aerosol treatment of PMB. The results of this study showed that on the premise that the difference in blood concentrations between the two groups of high and low ELF concentrations was not statistically significant (P > 0.05), the total clinical effective rate and total bacterial clearance rate of the high ELF peak concentration group were significantly higher than that of the low ELF peak concentration group (P < 0.05); the total effective rate of the high ELF trough concentration group was significantly higher than that of the low ELF trough concentration group (P < 0.05); further regression analysis was used to troubleshoot the effects of confounding factors suggesting that that the ELF concentration (high and low concentration) was an independent factor affecting the clinical efficacy. Although the time of temperature recovery between the two groups were not statistically significant (P > 0.05), the levels of IL-6 and CRP after PMB treatment in the high ELF concentration group were significantly reduced compared with those before treatment (both P < 0.05). These findings led to the conclusion that there is a direct relationship between higher PMB concentration in alveolar ELF and better clinical efficacy. This is consistent with the findings of Prof. Li [36]. His study confirmed that PMB IH is capable of achieving prolonged and extensive drug exposure in the lungs, which is crucial for PMB, a concentration-dependent antimicrobial drug. This not only validates the principle that high drug concentrations can improve clinical efficacy, but also demonstrates the potential of IH as a drug delivery modality to improve the efficacy of respiratory tract infections. In this way, the topical application of PMB optimizes its antimicrobial activity, reduces systemic toxicities, and provides an effective and safe strategy for the treatment of respiratory tract infections.

But PMB IH combined with IV therapy for CRGNB pneumonia is not always effective. The efficiency of nebulization is closely related to the size of the nebulized particles, the method of inhalation, the frequency of treatment, etc [44]. Ensuring that the drug reaches the site of infection in the lungs and achieves effective drug concentrations in the ELF by maximizing consideration of drug deposition in the lungs may be an important factor leading to effective treatment. In addition, in patients with CRGNB pneumonia, sputum is yellowish and sticky, and the airway is blocked by sputum embolus or the lung aeration is poor in consolidated infected lung parenchyma, etc [45]. Most of the nebulized drugs are unable to reach the target location, which results in the insufficient concentration of ELF and affects the therapeutic efficacy. Moreover, ELF variability in mechanically ventilated patients stems mainly from differences in ventilator parameters and uneven alveolar reexpansion, which need to be minimized by standardizing inspiratory flow, breath-holding time, and correcting for lavage recovery. To further explore the effect of ELF concentration on prognosis, we found that the high ELF concentration group exhibited a significant survival advantage compared with the low ELF concentration group in terms of 28-day survival (P < 0.05), suggesting the importance of high ELF concentration in improving the 28-day survival of patients. Binary logistic regression analysis revealed that ELF peak concentration, ELF trough concentration and SOFA score were independent factors affecting 28-day survival (P < 0.05). This result demonstrates the potential importance of monitoring and optimizing ELF concentrations to improve 28-day survival in patients with CRGNB pneumonia, and also confirms the value of the SOFA score in assessing patient prognosis. However, the lower concentration of alveolar ELF measured in our study compared to the concurrent study by Ding P [46] may be related to our inclusion of patients with CRGNB pneumonia whose poor local aeration in the lung segments affected the efficiency of nebulization, the dilution of the drug concentration due to the need for more lavage fluid flushing because of the high number of sputum plugs in the small airways, and the loss of material during sample handling and analysis. Given the importance of ELF monitoring on the efficacy and prognosis of CRGNB pneumonia, there should be subsequent multicenter, large-scale clinical studies to determine the optimal range of peak and trough ELF concentrations, with an emphasis on the use of ELF concentrations to guide the rational use of clinical PMB.

From the perspective of PMB adverse effects [47–51], we found the difference in the incidence of bronchospasm, nephrotoxicity and skin pigmentation during PMB treatment between the high ELF concentration group and the low ELF concentration group was not statistically significant (P > 0.05). This suggests that increased ELF concentrations do not cause an increase in the adverse effects of airway spasm, nephrotoxicity, and incidence of skin pigmentation. It is safe to increase ELF concentrations with rational nebulized dosing. Previous literature has reported possible airway adverse effects with the use of nebulized inhaled polymyxins [50], and the possible cause of bronchospasm is choline in vitro leading to mast cell degeneration [51]. The incidence of airway spasm was low in the present study because all the patients in the present study had an artificial airway, had prophylactic application of bronchodilators prior to nebulization of the PMB, and the use of optimized under-ventilator nebulization of PMB was administrated, hence the incidence of airway spasm was low. Neurotoxicity was not discussed in this study because all patients received mechanical ventilation with sedation, which made assessing neurotoxicity difficult.

The novelty of this study is introduced of PMB ELF concentration as a new index for assessing the therapeutic efficacy of CRGNB pneumonia and the demonstration of a direct relationship between higher PMB concentration in ELF and better clinical efficacy. This finding provides a new perspective on the evaluation of the efficacy of PMB-based therapy for CRGNB pneumonia. In fact, microdialysis has been shown to be a reliable method for measuring alveolar antibiotic concentrations in animal studies due to its ability to directly and dynamically monitor interstitial fluid concentrations [52], avoiding the dilution effect and spatial heterogeneity of ELF sampling. However, its invasiveness and technical complexity limit clinical translation. Future optimization of microdialysis probe design and validation of its correlation with clinical outcomes in human pneumonia models are needed. At this stage, ELF drug concentration measurement is easy to perform clinically and can still be used as a complementary tool to assess drug concentration compliance in diseased lung segments. Nevertheless, this study is a single-center observational study with a small sample size, and the PMB nebulization dose and nebulization modality used in the included patients were approximately the same, so it is not possible to judge the optimal nebulization dose and nebulization modality of PMB on this basis; therefore, large-sample, multicenter clinical trials are needed in the future to further investigate the clinical efficacy of different nebulization doses and modalities of PMB in the treatment of CRGNB pneumonia, the concentration of ELF, and the changes in adverse effects.

Limitations to our study exist. Excluding patients who discontinued treatment early may introduce selection bias, particularly since the reasons for early discontinuation could be treatment-related. To address this gap, we plan to explore the effects of early PMB ELF and plasma concentrations on clinical efficacy in future research. This represents an important area of investigation that we will examine in subsequent studies.

Conclusions

These findings suggest that achieving higher PMB ELF concentrations may enhance therapeutic efficacy, but further prospective studies are needed to confirm these results and establish causal relationships. In addition, PMB alveolar ELF concentration was an important independent predictor of clinical outcome and survival, highlighting the value of monitoring drug concentration in alveolar ELF to adjust clinical PMB use. These results provide important insights into PMB therapy for pneumonia treatment. Future studies should explore the specific effects of different ELF drug concentrations on therapeutic efficacy and how to achieve optimal drug concentrations by adjusting the nebulized inhalation strategy to improve the success of treatment.

Abbreviations

APACHE II

Acute Physiology, and Chronic Health Evaluation

BALF

Bronchoalveolar lavage fluid

CRGNB

Carbapenem-resistant gram-negative bacteria

CRP

C-reactive protein

ELF

Epithelial lining fluid

ICU

Intensive care unit

IH

Inhaled

IL-6

Interleukin-6

IV

Intravenous

PCT

Procalcitonin

PMB

Polymyxin B

ROC

Receiver operating characteristic

SOFA

Sequential Organ Failure Assessment

Authors’ contributions

Conceptualization, Lili Zhou, and Hui Zhang; methodology, Lili Zhou, Danjie Wang, and Hui Zhang; software, Danjie Wang; validation, Yiqin Lin, Yu Cheng, and Xueyong Li; formal analysis, Qinyong Weng; investigation, Danjie Wang; resources, Lili Zhou, Danjie Wang, Yiqin Lin, and Xuanxi Huang; data curation, Danjie Wang and Xueyong Li; writing—original draft preparation, Danjie Wang; writing—review and editing, Lili Zhou; visualization, Danjie Wang; supervision, Lili Zhou and Hongqiang Qiu; project administration, Lili Zhou, Wenwei Wu, Hongqiang Qiu, and Hui Zhang; funding acquisition, Lili Zhou, Wenwei Wu, and Hui Zhang. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by Science and Technology Guiding Project for Social Development of Fujian Science and Technology Plan in 2022 (Appropriation No.2022Y0020), Funding for Top Hospital and Specialty Excellence of Fujian Province [Appropriation No.2022(884)], The Fourth Batch of Hospital Key Discipline Construction Project (Appropriation No.0260001#), Fujian Guiding Project of Science and Technology (Grant No. 2021Y0019), the National Natural Science Foundation of China (Appropriation No. 82070377), and Joint Funds for the innovation of science and Technology, Fujian province (Appropriation No. 2021Y9048).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. All data generated during the project will be made freely available via the ResMan Research Manager (ChiCTR2100044087) and Union Hospital, Fujian Medical University’s Research Data Repository.

Declarations

Ethics approval and consent to participate

All patients or their legally authorized representatives provided written informed consent. The study was approved by the Ethics Committee of Union Hospital Affiliated to Fujian Medical University. The committee’s reference number was 2020KY0136/2022YF033-03. Chinese clinical trial registration number is ChiCTR2100044087. All procedures involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.