Trimethoprim-sulfamethoxazole as de-escalation in ventilator-associated pneumonia: a cohort study subanalysis.

PURPOSE: This is a subanalysis of a previous study which compared the effectiveness of trimetoprim-sulfametoxazole (TMP-SMX) with all other regimens for treatment of ventilator-associated pneumonia (VAP). Aim of the current study was to focus on the effectiveness of a strategy based on TMP-SMX as de-escalation from β-lactam including regimens.
METHODS: Retrospective cohort study including patients who were hospitalized for VAP from 2011 to 2019. Patients were distributed in two groups: NO SWITCH TO TMP-SMX group, including patients who received β-lactams for all treatment duration, and SWITCH TO TMP-SMX group, which included patients who switched to TMP-SMX from a β-lactam including regimen after microbiology diagnosis. Three clinical outcomes were analyzed: mortality at 30 days from the start of the antibiotic treatment (T30), mortality at the end of treatment (EoT), and acquisition of multidrug-resistant bacteria during hospitalization in intensive care unit.
RESULTS: Overall, 70 patients were included in the current study, 32/70 (45.7%) in NO SWITCH TO TMP-SMX group and 38/70 (54.3%) in SWITCH TO TMP-SMX group, 37/70 (52.8%) had been already included in the previous study. No significant differences in clinical outcomes and patient's characteristics were found when the two groups were compared.
CONCLUSIONS: De-escalation to TMP-SMX for VAP treatment was not associated with higher mortality at EoT and T30 than standard treatment with β-lactam. Monotherapy with TMP-SMX as de-escalation from broad-spectrum empirical regimens is a β-lactam sparing strategy worthy to be further investigated in either multicenter cohort studies or randomized clinical trials.

Introduction

β-lactam sparing strategies are strongly recommended to reduce the selection of extended spectrum β-lactamase (ESBL) and carbapenemase-producing bacteria [1]. Nevertheless, β-lactams are largely used for treatment of ventilator-associated pneumonia (VAP), as empirical treatment to cover potential infection sustained by gram negative bacilli. They are often associated with other molecules which cover methicillin-resistant Staphylococcus aureus (MRSA), as recommended by international guidelines [2, 3]. Broad-spectrum regimens are potentially not appropriate, and quick de-escalation is recommended when microbiology data are available to adapt treatment to antibiotic susceptibility [4]. Moreover, de-escalation allows reducing broad-spectrum antibiotic use in intensive care unit (ICU), and it is not associated with increased mortality [5]. However, few data are available about the best molecules to use for targeted treatment of VAP, and no specific recommendations are given by international guidelines [2, 3].

Trimethoprim-sulfametoxazole or co-trimoxazole (TMP-SMX) is usually active on the most frequent bacteria associated with VAP, notably S. aureus (including MRSA) and Gram negative bacilli (Enterobacteriaceae) with the exception of nonfermenting Gram negative bacilli, Pseudomonas aeruginosa, and Acinetobacter baumannii [6]. In a previous study, we demonstrated that treatment of VAP with TMP-SMX was not associated with higher mortality and multidrug-resistant (MDR) bacteria acquisition than treatment not including TMP-SMX. However, that study compared patients receiving TMP-SMX, as first-line therapy or as de-escalation, with patients not receiving TMP-SMX treated with β-lactams or other broad-spectrum antibiotics, such as fluoroquinolones. That study did not focus on TMP-SMX as de-escalation from β-lactam including regimens for VAP treatment [7].

Aim of this study was to perform a subanalysis to verify the effectiveness of a de-escalation strategy from β-lactams to TMP-SMX for treatment of VAP.

Materials and methods

We conducted a monocentric retrospective cohort study in an ICU of a 350 acute-care-bed hospital in the Ile de France region in France. The entire cohort of patients identified in the previous study (2011–2017) plus all patients with a diagnosis of VAP from January 1, 2018, to December 31, 2019, were considered for inclusion [7]. From this new cohort, only patients with VAP treated with broad-spectrum penicillins plus/minus β-lactamase inhibitors as first-line regimen and who switched or not to TMP-SMX were included. Exclusion criteria were (i) absence of isolates at cultures from lower respiratory tract samples; (ii) positivity at cultures from lower respiratory tract samples for TMP-SMX naturally resistant bacteria (P. aeruginosa and A. baumannii); (iii) positivity at cultures from lower respiratory tract samples for naturally β-lactam-resistant bacteria (Stenotrophomonas maltophilia); (iv) first-line treatment with molecules different from broad-spectrum penicillins; (v) switch to alternative molecules to broad-spectrum penicillins and TMP-SMX.

The study was conducted in accordance with Declaration of Helsinki and national and institutional standards. Approval by the local ethic committee was not demanded because a noninterventional research was conducted, according to French law. A written consent form was not proposed to patients because the noninterventional nature of the study required only the absence of patients’ opposition, according to the French law [8, 9].

VAP was defined as infection of pulmonary parenchyma developed after at least 48 h of mechanical ventilation [10]. For all patients, first-line treatment always included a broad-spectrum penicillin plus/minus β-lactamase inhibitor (amoxicillin, piperacillin, amoxicillin/clavulanic acid and piperacillin/tazobactam). No antibiotic for MRSA coverage was systemically added because of the extremely low rate of MRSA in our hospital (5%). De-escalation to TMP-SMX was systemically proposed when antibiotic susceptibility test for microbiologic isolates from lower respiratory tract samples was available not later than 48–72 h from the beginning of the antibiotic treatment. All patients with susceptible isolates switched to TMP-SMX, regardless of clinical response and even in case of clinical worsening. This decision was justified by the priority of limiting selection of ESBL. Patients continued or switched to broad-spectrum penicillin plus/minus β-lactamase inhibitor when antibiotic susceptibility test was not available at 48–72 h from the beginning of the antibiotic treatment or it documented TMP-SMX-resistant isolates. TMP-SMX was prescribed at a maximal dose of 160/800 mg tid or qid and adapted to estimated glomerular filtration rate (eGFR) and patient’s weight [11]. All patients received a 7-day-long antibiotic treatment, with broad-spectrum penicillins plus/minus β-lactamase inhibitors with or without switch to TMP-SMX.

Patients were retrospectively pooled in two groups: (i) NO SWITCH TO TMP-SMX group, which included patients who started a β-lactam as first-line treatment and maintained a β-lactam based regimen even after microbiologic diagnosis, and (ii) SWITCH TO TMP-SMX group, which included patients who switched to TMP-SMX from a β-lactam including regimen.

For collection of patients’ characteristics, laboratory analysis and clinical outcomes, the following software were used: Sillage v17.2.4.5 and CGM Lab channel 1.20.33686. Patients’ characteristics included age, gender, body mass index (BMI), co-morbidities, antibiotic treatment before the onset of VAP, simplified acute physiology score II (SAPS-II), early VAP, shock, bacteria isolates from lower respiratory tract samples, antibiotic susceptibility and first-line molecules. Co-morbidities included diabetes, heart disease (acute myocardial infarcts and heart failure), lung disease (chronic obstructive pulmonary disease or chronic restrictive diseases), liver disease, solid or hematologic neoplasia, severe acute or chronic kidney disease. Severe kidney disease was defined for eGFR<30 ml/min [12]. An onset of VAP ≤96 h from the start of mechanical ventilation was considered for the definition of early VAP [13]. Shock was defined by the needing of vasopressors to maintain a mean arterial pressure ≥65 mmHg at VAP onset [14].

All positive isolates from lower respiratory tract samples were considered (endotracheal aspiration or bronchoalveolar lavage). Bronchial secretions were analyzed when they presented <25 squamous epithelial cells and >25 leukocytes per low power field. Bacterial culture threshold was 105 for endotracheal aspirate and 104 for bronchoalveolar lavage.

Primary outcome was mortality at 30 days after antibiotic treatment initiation (T30). Secondary outcomes were (i) mortality at the end of treatment (EoT); (ii) acquisition of MDR bacteria during hospitalization in ICU; (iii) severe allergy occurrence (defined as any allergic event causing antibiotic treatment interruption or switch to other molecules); (iv) Clostridium difficile disease occurrence (defined as the presence of binary toxin in stools). Nasopharyngeal and rectal swabs were systemically realized at admission and discharge, and their results were considered for definition of MDR bacteria. Other samples obtained during the hospitalization according with patient’s clinical evolution were also considered for definition of MDR acquisition. MDR research included MRSA, ESBL and carbapenemase-producing bacteria.

The two groups of patients were compared (NO SWITCH TO TMP-SMX vs. SWITCH TO TMP-SMX). The following statistic tests were performed: χ2 test for qualitative variables and Student’s t test for quantitative variables. Quantitative variables were presented in the text as mean values.

R, the language for statistical computing (Vienna, Austria, https://www.r-project.org/), was used to perform all statistical analysis. Nominal statistical significance was set at p < 0.050.

Results

Overall, 199 cases of VAP were identified from January 1, 2011, to December 31, 2019. Among them, 126/199 (63.3%) had been already identified for the previous study (2011–2017) while 73/199 (36.7%) were newly detected (2018–2019). A total of 70/199 (35%) patients were included in the current study, among them, 37/70 (52.8%) had been already included in the previous study. According to exclusion criteria, the following patients were excluded: 10/199 (5%) patients had not microbiology diagnosis, 48/199 (24%) patients had P. aeruginosa and Acinobacter baumannii isolates, 10/199 (5%) patients had Stenotrophomonas maltophilia isolates, 52/199 (26%) patients were excluded because their first-line treatment included molecules different than broad-spectrum penicillins and 8/199 (4%) patients were excluded because they switched towards a molecule different than TMP-SMX or broad-spectrum penicillins.

Overall, 32/70 (45.7%) patients were included in NO SWITCH TO TMP-SMX group and 38/70 (54.3%) in SWITCH TO TMP-SMX group. All patients started a monotherapy with a β-lactam as first-line treatment with the exception of two patients who started a dual therapy with piperacillin/tazobactam plus aerosolized amikacin and amoxicillin/clavulanic acid plus linezolid, respectively. The following β-lactam molecules were prescribed as first-line regimen: piperacillin/tazobactam (34/70, 48.6%), amoxicillin/clavulanic acid (27/70, 38.6%), amoxicillin (7/70, 10%) and piperacillin (2/70, 2.8%). Switching to another molecule was prescribed in 48/70 (68.5%) as monotherapy (45/48, 93.7%) or dual therapy (3/48, 6.3%). The following β-lactams were prescribed for de-escalation in 10/48 (20.8%) patients: amoxicillin/clavulanic acid (4/10, 40%), piperacillin/tazobactam (3/10, 30%) and amoxicillin (3/10, 30%). Aerosolized amikacin was associated with β-lactams in 2/32 (6.25%) patients. A total of 38/70 (54.3%) patients switched to TMP-SMX. Among them, only one patient switched to a dual therapy with TMP-SMX and aerosolized amikacin (1/38, 2.6%) while the rest of patients switched to a monotherapy (37/38, 97.4%).

Comparison of NO SWITCH TO TMP-SMX and SWITCH TO TMP-SMX groups did not show any significant differences (Table 1) in terms of patient’s characteristics and clinical outcomes (mortality at T30, mortality at EoT and acquisition of MDR bacteria). Analysis of safety profile showed no case of severe allergy in NO SWITCH TO TMP-SMX and SWITCH TO TMP-SMX groups. Clostridium difficile disease occurred only in a patient who switched to TMP/SMX from amoxicillin.

Table 1

Characteristics of the population

Parameters	NO SWITCH TO TMP-SMX	SWITCH TO TMP-SMX	p value
	(n = 32)	(n = 38)
Biological parameters
Age years [mean (SD)]	60 (17.6)	66 (11.8)	0.06
Male gender [n (%)]	27 (84.4)	29 (76.3)	0.40
BMI (SD) [mean (SD)]	28.50 (6.3)	28.52 (5.4)	0.99
Co-morbidities
Diabetes [n (%)]	8 (25)	16 (42.1)	0.13
Heart disease [n (%)]	17 (53.1)	22 (57.9)	0.69
Lung disease [n (%)]	13 (40.6)	12 (31.6)	0.43
Liver disease [n (%)]	6 (18.8)	8 (21.1)	0.81
Cancera [n (%)]	6 (18.8)	5 (13.2)	0.52
eGFR<30 ml/min [n (%)]	4 (12.5)	7 (18.4)	0.50
Clinical parameters
SAPS-II [mean (SD)]	47 (19.5)	49 (16.7)	0.65
Antibiotic treatment before VAP [n (%)]	25 (78.1)	31 (81.6)	0.72
Early VAP [n (%)]	8 (25)	15 (39.5)	0.20
Shock [n (%)]	17 (53.1)	23 (60.5)	0.53
Bacterial isolates from lower respiratory tract samples
Enterobacteriacae [n (%)]	19 (59)	32 (84)	0.14
Haemophilus influenzae [n (%)]	6 (19)	3 (8)
Gram positive bacteria [n (%)]	7 (22)	3 (8)
Antibiotic susceptibility
Amoxicillin susceptible [n (%)]	7 (21.87)	6 (15.78)	0.51
Amoxicillin/clavulanic acid susceptible [n (%)]	23 (71.87)	12 (31.57)	0.0007
Piperacillin/tazobactam susceptible [n (%)]	26 (81.25)	27 (71.05)	0.32
Third generation cephalosporin susceptible [n (%)]	25 (78.12)	30 (78.94)	0.93
Fluoroquinolone susceptible [n (%)]	24 (75)	36 (94.73)	0.02
TMP-SMX susceptible [n (%)]	18 (56.25)	38 (100)	<0.0001
First-line molecules
Amoxicillin [n (%)]	5 (16)	2 (5)	0.17
Amoxicillin/clavulanic acid [n (%)]	10 (31)	17 (45)
Piperacillin [n (%)]	2 (6)	0 (0.0)
Piperacillin/tazobactam [n (%)]	15 (47)	19 (50)
Clinical outcomes
EoT mortality [n (%)]	6 (18.8)	9 (23.7)	0.62
T30 mortality [n (%)]	12 (37.5)	16 (42.1)	0.69
MDR bacteria acquisition during hospitalizationb [n (%)]	3 (9.4)	5 (13.2)	0.62

BMI body mass index; eGFR estimated glomerular filtration rate; MDR multidrug-resistant; SAPS-II simplified acute physiology score II; SD standard deviation; TMP-SMX trimethoprim-sulfamethoxazole; VAP ventilator-associated pneumonia

aIncludes both solid and haematological cancer.

bAccording to nasopharyngeal and rectal swab screening (at admission and discharge) and other samples obtained during the hospitalization.

Discussion

Results of this study suggest that TMP-SMX may represent a valid alternative to β-lactams in case of VAP with microbiology diagnosis. Indeed, no differences were found in terms of clinical outcomes (mortality at T30 and EoT, and MDR bacteria acquisition) when NO SWITCH TO TMP-SMX and SWITCH TO TMP-SMX groups were compared. Because of lack of activity on nonfermenting Gram-negative bacteria (P. aeruginosa and A. baumannii), the best use of TMP-SMX on VAP is likely de-escalation after microbiology diagnosis. First-line empirical treatment with TMP-SMX should be avoided.

Although quick de-escalation is recommended by international guidelines [2, 3], no study was performed to compare the effectiveness of different regimen for de-escalation treatment of VAP. This is the first study which analyses the effectiveness of de-escalation from a β-lactam regimen toward a regimen not including β-lactams.

De-escalation therapy for VAP should always been encouraged because it is associated with some advantages. At first, de-escalation reduces the exposition to β-lactamase inhibitor and carbapenems and the risk of selection of ESBL and carbapanemase producing bacteria [15, 16]. Secondly, it allows reducing cost and duration of hospitalization without affecting clinical outcomes, notably mortality [17-20]. Third, it reduces the number of molecules administered, and consequently, it enhances the safety profile of antibiotic treatment [21].

This study is monocentric, and it is limited by retrospective design, small population and absence of sample size calculation. Consequently, results are not definitive and need to be reproduced in bigger studies. However, aim of this study was to perform an exploratory analysis which could justify new multicenter studies, either retrospective cohort studies or randomized clinical trials (RCT). In particular, RCT are strongly required because they represent the most suitable way to compare drug efficacy and effectiveness. We retain that for its pharmacodynamic and pharmacokinetic characteristics, TMP-SMX is worthy to be further explored as de-escalation treatment for VAP. Indeed, its higher lung concentration and intestinal absorption make it suitable for a quick and rapid switch from an intravenous treatment after microbiology diagnosis [13, 14, 22]. Moreover, its activity versus the most common agents of VAP, namely, Enterobacteriacae, MRSA and vancomycin intermediate or resistant S. aureus, makes it an interesting molecule even in settings with high prevalence of β-lactam and glycopeptide-resistant bacteria [23-25]. Further studies should investigate whether these theoretical benefits produce a clinical advantage or not. In particular the impact of de-escalation to TMP-SMX on clinical outcomes (mortality and MDR bacteria acquisition) should be defined.

In conclusion, this study reported encouraging results about the use of TMP-SMX as de-escalation for treatment of VAP. Monotherapy with TMP-SMX as de-escalation from broad-spectrum empirical regimes is a β-lactam sparing strategy worthy to be further investigated in either multicenter cohort studies or RCT.