*Article* **Heterogeneity of Antibiotics Multidrug-Resistance Profile of Uropathogens in Romanian Population**

**Răzvan-Cosmin Petca 1,2,† , Silvius Negoit,ă 1,3,†, Cristian Mares, 1,2,\*, Aida Petca 1,4,\* , Răzvan-Ionut, Popescu 1,2 and Călin Bogdan Chibelean 5,6**


**Abstract:** Urinary tract infections (UTIs) are a leading cause of morbidity for both males and females. The overconsumption of antibiotics in general medicine, veterinary, or agriculture has led to a spike in drug-resistant microorganisms; obtaining standardized results is imposed by standard definitions for various categories of drug-resistant bacteria—such as multiple-drug resistant (MDR), extensive drug-resistant (XDR), and pan drug-resistant (PDR). This retrospective study conducted in three university teaching hospitals in Romania has analyzed urine probes from 15,231 patients, of which 698 (4.58%) presented multidrug-resistant strains. *Escherichia coli* was the leading uropathogen 283 (40.54%), presenting the highest resistance to quinolones (R = 72.08%) and penicillin (R = 66.78%) with the most important patterns of resistance for penicillin, sulfonamides, and quinolones (12.01%) and aminoglycosides, aztreonam, cephalosporins, and quinolones (9.89%). *Klebsiella* spp. followed—260 (37.24%) with the highest resistance to amoxicillin-clavulanate (R = 94.61%) and cephalosporins (R = 94.23%); the leading patterns were observed for aminoglycosides, aminopenicillins + β-lactams inhibitor, sulfonamides, and cephalosporins (12.69%) and aminoglycosides, aztreonam, cephalosporins, quinolones (9.23%). The insufficient research of MDR strains on the Romanian population is promoting these findings as an important tool for any clinician treating MDR-UTIs.

**Keywords:** urinary tract infections; UTIs; MDR; *Escherichia coli*; *Klebsiella*; uropathogens; AMR; antibiotic resistance

#### **1. Introduction**

Urinary tract infections (UTIs) represent a common disorder treated by urologists and general medical practitioners, accounting for an important percentage of the yearly healthcare costs [1]. Most UTIs are treated on ambulatory patients [2]. However, the increasing resistance to the first-line antibiotic treatment [3–5] and the rising quota of this condition [6] have urged the research of new lines of therapy. In practice, we must combine updated data on uropathogens' resistance profiles and sensibility rates of antimicrobial agents used in the treatment of UTIs.

**Citation:** Petca, R.-C.; Negoit,˘a, S.; Mares, , C.; Petca, A.; Popescu, R.-I.; Chibelean, C.B. Heterogeneity of Antibiotics Multidrug-Resistance Profile of Uropathogens in Romanian Population. *Antibiotics* **2021**, *10*, 523. https://doi.org/10.3390/ antibiotics10050523

Academic Editors: Pavel Bostik and Milan Kolar

Received: 8 April 2021 Accepted: 1 May 2021 Published: 2 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Several factors are linked to promoting the increasing spread of bacterial resistance to antibiotics in community settings. The most important vector of increasing resistance is represented by the overuse of antimicrobials in general medicine, veterinary, or agriculture, which enables the selection and spread of drug-resistant strains [7]. Other risk factors are host-related; an extensive review of the literature [8] aiming to detect the risk factors associated with multidrug-resistance (MDR) UTIs has highlighted 12 possible factors:


International assemblies of heads of departments from international specialized forums such as the European Center for Disease Prevention and Control (Stockholm, Sweden), Office of Infectious Diseases, Department of Health and Human Services from Center for Disease Prevention and Control (Atlanta, GA, USA), and Division of Epidemiology, Tel Aviv Sourasky Medical Center (Tel Aviv, Israel) [9] proposed standard definitions for various categories of drug-resistant bacteria. These are classified as [9]:


They admitted that a better understanding of highly resistant bacterial strains and obtaining comparability data would be facilitated if these definitions were applied worldwide.

The European Association of Urology (EAU) via EAU Guidelines on Urological Infections from 2020 [10] recommends empirical treatment of uncomplicated urological infections using sulfonamides (TMP-SMX), phosphonic acids (fosfomycin), or nitrofurantoin. Fluoroquinolones (ciprofloxacin and levofloxacin) may be used only as an alternative therapy while also considering locoregional resistance rates. Carbapenems (imipenem and meropenem) should be used only as reserved therapy or in special conditions such as urosepsis.

This study aimed to determine the resistance profiles of the most frequent multidrugresistant uropathogen strains involved in UTIs on a Romanian male and female cohort. The preliminary data [11] have shown *Escherichia coli* as the most frequent bacteria (42.9%) implicated in UTIs, followed by *Klebsiella* spp. (21.17%), *Enterococcus* spp. (18.66%), *Proteus* spp. (7.75%), *Staphylococcus* spp. (4.91%), and *Pseudomonas aeruginosa* (4.58%). The limited number of MDR strains studied in previous research, the necessity of determining specific resistance patterns for each bacteria against common antibiotic classes in treating UTIs, and comparing the results with the international findings were the decisive factors in initiating the study.

#### **2. Results**

A total number of 698 cases of MDR-UTIs were registered in all three centers during research, as follows: 262 patients (37.53%) at "Prof. Dr. Th Burghele" Clinical Hospital (BCH), 278 cases (39.82%) at Elias University Hospital (EUH), and 158 samples (22.63%) at Mures County Hospital (MCH). A detailed report on MDR-UTIs uropathogen distributions for each center is presented in Table 1 and Figure 1.



*n*—number, %—percentage, BCH—Burghele Clinical Hospital, EUH—Elias University Hospital, and MCH—Mures County Hospital.

**Figure 1.** Distribution of the MDR uropathogens in the study centers.

Except for BCH center, where *Klebsiella* spp. (43.51%) surpassed *E. coli* (28.62%) in prevalence, the rest of the Gram-negative uropathogens respected the distribution in all subjects. Overall, *Enterococcus* spp. was the most incriminated Gram-positive bacterial strain, except for the BCH center, where *Staphylococcus* spp. surpassed it in prevalence; in the EUH center, a single strain of MDR-*Staphylococcus* spp. was detected; as for the MCH center, none of these pathogens were registered.

In terms of age-group distribution, both males and females showed an increased prevalence at the lower pole of the distribution axis with 6.45% females and 6.68% males in their 18–29 years. We noted a progressive increase in MDR-UTIs with every decade of life. This highlights the correlation between age and incidence of UTIs; a high quota was observed in seniors between 60–69, representing 24.73% in females and 30.07% in males, with a peak of incidence in patients over 70 years old—48.28% in the overall population. Detailed data on age-group distribution is displayed in Table 2.


**Table 2.** Female and male age group distribution of the MDR uropathogens.

*n*—number and %—percentage.

*E. coli*, the most frequent microbial strain, showed the highest resistance rate (R) to quinolones-levofloxacin—72.08%, followed by penicillin-ampicillin—66.78%, cephalosporinsceftazidime—60.07%, and aminopenicillins + β lactamase-amoxicillin-clavulanate—56.89%. A good sensitivity (S) was observed for fosfomycin—83.74%, followed by amikacin—66.78% and nitrofurantoin—39%.

Detailed information of each Gram-negative bacterial strains, including resistance and sensitivity profiles and overall statistics, are included in Figure 2 and Table 3.

*Klebsiella* ranked as the second-most common uropathogen in the study. It showed an outstanding resistance to all antibiotic classes, led by amoxicillin-clavulanate—94.61%, followed by ceftazidime—94.23%, levofloxacin—63.84%, and amikacin—53.07%. No good sensitivity was observed for either of the tested antibiotics. None of them showed resistance below 10%. The lowest resistance was obtained for fosfomycin—15.75%, nitrofurantoin— 21.92%, and carbapenems—imipenem and meropenem—21.15% and 23.46%, respectively.

**Figure 2.** Gram-negative uropathogen resistance profiles (AG—aminoglycosides, AM + C—amoxicillin + clavulanic ac., AMP—ampicillin, AZ—aztreonam, SF—sulfonamides, C—cephalosporins, FO—Fosfomycin, IMP—imipenem, Q quinolones, MER—meropenem, and NF—nitrofurantoin).

**Table 3.** Gram-negative uropathogen resistance profiles.


*n*—number, %—percentage; R—resistant, S—sensitive, and NA—not available.

The third-most frequent uropathogen, *P. aeruginosa*, shows almost complete resistance in MDR strains to quinolones-levofloxacin—96.66%; alarmingly, a high resistance is also observed for cephalosporins-ceftazidime—88.33%, followed by aminoglycosidesamikacin—88.33%. Surprisingly, the resistance to carbapenems in MDR *P. aeruginosa* is the highest in all pathogens for this antimicrobial class, accounting for imipenem—75% and meropenem—70%.

*Proteus* spp. is considered a nosocomial uropathogen, consistently discovered in patients presenting complicated UTIs. *Protea* (a group of pathogens including *Proteus*, *Providentia*, and *Morganella* spp.) are naturally resistant to colistin and nitrofurantoin and have raised resistance to carbapenems [12]. Our study discovered lower resistance rates than other Gram-negative bacteria, showing the highest resistance profile to ceftazidime—82.92%, followed by amoxicillin-clavulanate—80.48%, levofloxacin—58.53%, and ampicillin—56.09%. Relatively preserved sensitivity was observed for amikacin and meropenem—both 65.85%.

Both Gram-positive bacteria in this study make up for less than 10% of the total strains: *Staphylococcus* spp.—3.43% and *Enterococcus* spp.—4.29%; in both cases, the highest resistance was observed for quinolones (*Staphylococcus* spp.—91.66% and *Enterococcus* spp.—63.33%) and penicillin (*Staphylococcus* spp.—83.33% and *Enterococcus* spp.—70.0%).

The most frequent association of antimicrobial classes involved in common MDR strains was represented by amoxicillin-clavulanate, aztreonam, cephalosporins, and quinolones in 56 isolates (8.02%), followed by aminoglycosides, amoxicillin + clavulanate, sulfonamides, and cephalosporins in 5.01%; penicillin, sulfonamides, and quinolones in 4.87%; and aminoglycosides, amoxicillin + clavulanate, aztreonam, cephalosporins, carbapenems, and quinolones in 4.01%. Detailed outcomes of the 10th-most common MDR strain resistance patterns can be found in Table 4.

**Table 4.** Most common MDR profiles.


*n*—number and **%**—percentage.

The resistance profile for *E. coli* and *Klebsiella* presented similarities, as well as noticeable differences, in the results. For *E. coli*, a high resistance to various combinations can be observed, such as penicillin, sulfonamides, and quinolones (*n* = 34 strains); aminoglycosides, aztreonam, cephalosporins, and quinolones (*n* = 28 strains); and aminoglycosides, aminopenicillins + β-lactams inhibitor, aztreonam, cephalosporins, and quinolones (*n* = 9 strains)—Figure 3.

**Figure 3.** *Escherichia coli* resistance profiles of MDR strains (Ag—aminoglycosides, Am + C aminopenicillins + β-lactams inhibitor, Az—aztreonam, C—cephalosporins, P—penicillin, Q—quinolones, and Sf—sulfonamides).

*Klebsiella* spp. revealed resistance to aminoglycosides, aminopenicillins + β-lactams inhibitor, sulfonamides, and cephalosporins (*n* = 33 strains), followed by aminoglycosides, aztreonam, cephalosporins, and quinolones (*n* = 24 strains) and aminoglycosides, aminopenicillins + β-lactams inhibitor, aztreonam, cephalosporins, carbapenems, and quinolones (*n* = 18 strains)—Figure 4. *E. coli* proved resistant mostly to penicillin and quinolones, while *Klebsiella* spp. to aminoglycosides, the aminopenicillins+ β-lactams inhibitor, or even carbapenems.

**Figure 4.** *Klebsiella* spp. resistance profiles of MDR strains (Ag—aminoglycosides, Am + C aminopenicillins + β-lactams inhibitor, Az—aztreonam, C—cephalosporins, Cb—carbapenems, Nf—nitrofurantoin, Q—quinolones, and Sf—sulfonamides).

*E. coli* proved resistant mostly to penicillin and quinolones, while *Klebsiella* spp. to aminoglycosides, the aminopenicillins+ β -lactams inhibitor, or even carbapenems.

#### **3. Discussion**

Infections located in the urinary tract are a common cause of urological treatment among the general population. Moreover, acquiring a bacterial strain that shows resistance to multiple antimicrobial agents in use overlaps the primary morbidity of UTIs alone, leading to an excessively dangerous disease that in the absence of prompt and adequate treatment, can cause lots of problems. Due to overuse of antibiotics in various fields, such as general medicine, veterinary, or agriculture, an alarming increase of MDR strains among uropathogens was detected; taking into account the lack of locoregional data on MDR-UTIs incidence and resistance patterns, results from a six-month multicenter "cross-sectional" retrospective study are provided.

#### *3.1. MDR Uropathogens in Relation with Patients Age*

Throughout the entire cohort of patients, one could see a general trend of progressive growth in MDR-UTIs with every decade of life. This highlights the correlation between age and incidence of UTIs, just like Rowe et al. [13] summarized in a literature review. A high quota was observed in seniors between 60–69 years old, representing 24.73% in females and 30.07% in males, with a peak of incidence in patients over 70 years old—48.28% in the overall population. It has been shown that multiple risk factors are associated with the prevalence of UTIs in the elderly male population, such as prostate enlargement, urolithiasis, urinary tract neoplasia, renal failure, or urethral strictures [14,15].

In the elderly female population, various risk factors are associated with UTIs. The studies performed by Hu et al. [16], Brown et al. [17], and a literature review from Mody et al. [18] demonstrated that a history of UTIs during early lifetime, diabetes, functional disability, urinary retention, presence of urinary catheters, history of urogynecology surgery (all of them eventually combined with estrogen deficiency) are common risk factors-related with this age category. As an exception, a spike of incidence is underlined at the lower pole of the distribution axis; the increased number of positive MDR urine cultures in these patients is linked to the most active sexual period in both sexes, at the end of puberty, as Foxman B. [19] and Chu et al. [20] previously showed in their papers. It has been proven that high frequency of sexual intercourse, use of male condoms, contraceptive diaphragms, spermicides, and abusing some of the antimicrobials are key factors of acquiring UTIs at a younger age [21].

#### *3.2. Comparison of MDR Escherichia coli Patterns with Other Studies*

For the most common uropathogen in the tested cohort, high resistance rates to the most common antimicrobial agents were observed; similar results for MDR *E. coli* strains, with amoxicillin R = 55.6% and nitrofurantoin R = 7.9%, were obtained by Baral et al. [22] in 2012, while the results for ceftazidime and amikacin were R = 100% and R = 6.2%, respectively. Dehbanipour et al. [23] performed a study in Iran in 2016 that showed an alarming resistance in MDR *E. coli* to amikacin (R = 89.1%) higher than our findings and with an even higher difference compared to nitrofurantoin (R = 85.9%). The same paper admitted increasing resistance to carbapenems (meropenem), while we observed a still very low resistance in this group—R = 0.7%—both to imipenem and meropenem.

In terms of MDR patterns, *E. coli* showed the highest one to penicillin and quinolones; the last one appeared in almost all MDR *E. coli* patterns. Linhares et al. [24] developed a similar study in 2015 in Aveiro, Portugal analyzing 4376 MDR uropathogens strains, searching for resistance patterns in MDR microorganisms; the report highlighted the highest resistance to the combination of penicillin, sulfonamides, and quinolones with 12.6%, followed by cephalosporins, nitrofurans, and penicillin with 7.8%; cephalosporins, penicillin, and sulfonamides in 6.6%; and cephalosporins, penicillin, quinolones, and sulfonamides with 6.0%. Contrary to the Portuguese research, the current findings suggest an increasing resistance for aminopenicillins and aminoglycosides in the tested MDR strains. A recent study from Iasi, Romania, conducted in early 2020 [25], obtained similar locoregional results, highlighting an increasing resistance to carbapenems. Surprisingly, the tested strains showed good sensitivity to fosfomycin (S = 83.74%); similar findings were reported by Sultan et al. in 2014 [26]. His study analyzed the resistance rates among uropathogens, especially MDR strains, highlighting 100% sensitivity to this drug from ESBL and non-ESBL *Enterobacteriaceae*, *Staphylococcus aureus*, and *Enterococcus fecalis*.

#### *3.3. Comparison of MDR Klebsiella Patterns with Other Studies*

*Klebsiella* spp. is the Gram-negative bacteria ranked as the second-most frequent uropathogen in MDR-UTIs but the first in terms of resistance, as previously shown. Mishra et al. [27] published in 2013 extensive research on 996 MDR uropathogen strains sampled from hospitalized patients that were tested for common antibiotics in three different time phases. High rates of resistance were observed for *Klebsiella* spp. to multiple antibiotic

classes. The Indian study highlighted higher resistance in 2013 compared to our results for amikacin (R = 65%, 76%, and 79%); trimethoprim-sulfamethoxazole (R = 66%, 69%, and 78%); and nitrofurantoin (R = 69%, 71%, and 76%), while, for amoxicillin-clavulanate (R = 54%, 67%, and 76%); ceftazidime (R = 64%, 69%, and 73%); and levofloxacin (R = 47%, 49%, and 53%), better sensitivity in all three phases was observed, contrary to our results.

One can see that, for both *E. coli* and *Klebsiella* spp., the two most common uropathogens in MDR strains, an alarmingly increased resistance for beta-lactams and fluoroquinolones has been reported by many researchers. Various mechanisms of resistance—such as enzymatic inactivation, target modification, reduced permeability reduction, or active efflux, are cited for the beta-lactams [28]. For fluoroquinolones, chromosomal mutations altering target enzymes DNA gyrase and topoisomerase IV or activating the efflux systems and pumping drugs out of the cytoplasm, concomitant with the loss of porin channels for drug entry [29], are key factors in the resistance patterns. Numerous studies have highlighted the increasing resistance to beta-lactams (penicillin, aminopenicillins, and cephalosporins) and—even if numbers are still low—a continuous growth for carbapenems [30] for all uropathogens (especially for *E. coli* and *Klebsiella* spp. [31–33]). The same can be stated in regard to quinolones [34,35].

Thus, a careful administration for both antimicrobial classes is recommended, always considering the locoregional studies on antimicrobial resistance patterns. The European Guidelines on Urological Infections [10] also suggest similar cautiousness.

#### *3.4. How MDR Strains Modify Our Daily Practice in UTIs*

According to the European Association of Urology [10], the primal evaluation of a patient suspecting an UTI imposes an empiric antibiotic treatment, taking into account locoregional patterns of resistance, followed by the collection of urine specimens for uroculture; the antibiogram will later conduct the definitive treatment of the specific UTI. In suspected cases of MDR microorganisms, a careful decision on recommending an antibiotic should be made.

A recent study from Germany conducted by Bischoff et al. [36] that was following the empiric antibiotic therapy in UTIs in patients with risk factors for resistant uropathogens noticed that, in patients with two or more risk factors, the lowest susceptibility was represented by quinolones-ciprofloxacin (S = 51.8%). A second-generation cephalosporincefuroxime (S = 53.7%) followed, then a third-generation cephalosporin-cefpodoxime (S = 60.7%), a wide broad-spectrum penicillin with a beta lactamase inhibitor-piperacillin/ tazobactam (S = 75.0%), an aminoglycoside-gentamycin (S = 75.0%), another third-generation cephalosporin-ceftazidime (S = 76.8%), and the highest susceptibility was reported for carbapenems-imipenem (S = 91.1%). Another recent study from Italy published by Gasperini et al. [37] concerning MDR UTI bacteria in geriatric population also highlighted the increased uropathogen resistance for quinolones-levofloxacin (R = 80.0%) and ciprofloxacin (R = 55.3%); cephalosporins-cefepime, cefotaxime, and ceftazidime (R = 45.8%, 59.4%, and 51.6%, respectively); and aminopenicillin with beta lactamase inhibitor-amoxicillin/clavulanic acid (R = 50.0%) was noted. Favorable sensitivity patterns for aminoglycosides-amikacin (R = 9.4%) and carbapenems-meropenem (R = 8.5%) were also noted.

All this data endorses the presented results, as this study observed the highest resistance for quinolones, cephalosporins, and aminopenicillins; intermediate resistance was noted for amikacin; and confined sensitivity was observed for carbapenems, although not enough ubiquitous testing in all three centers was conducted for this drug class. Surprisingly, promising sensibility was noted for fosfomycin; this "old forgotten drug", as Bradley J Gardiner [38] referred to it in a study from early 2019 concerning the resistance patterns for nitrofurantoin and fosfomycin, could be an effective alternative for these patients, especially in uncomplicated UTIs.

Although fosfomycin is both a safe and effective drug in the management of the empiric treatment of a UTI, as MDR strains breed complicated features, its efficiency and usage must be limited to cystitis.

When a MDR uropathogen is suspected in complicated UTIs, the only viable treatment is represented by carbapenems.

#### *3.5. Limitations*

The impossibility of determining each bacterial species' genus in all involved centers is the major limitation of the study. It would have been helpful if the minimal inhibitory concentration of each drug was measured, providing a better understanding of the resistance dynamics in each of the tested strains. The clinical information of every patient was not available, especially in the ambulatory patients, to allow a correlation between resistance profiles and risk factors, plus medical history. One of the study's primary limitations was represented by the lack of assessment of the risk factors contributing to the emerging of multidrug resistance, resulting in the impossibility of suggesting specific recommendations. Considering that the study was performed only in tertiary-care hospitals where patients with more severe pathology are managed, this could represent a risk of selection bias.

All the aforementioned would have improved the accuracy of the recommendations that would eventually reduce acquiring MDR bacteria.

#### **4. Materials and Methods**

#### *4.1. Study Design and Sample Population*

This descriptive "cross-sectional" retrospective study was conducted at three different clinical hospitals in two University centers: "Prof Dr. Th Burghele" Clinical Hospital (BCH) and Elias University Hospital (EUH) from Bucharest, Romania and Mures County Hospital (MCH) from Targu Mures, Romania. The data collection was conducted for six months, between 1 September 2018 and 28 February 2019.

Urine probes from 15,231 patients were collected for bacterial analysis, out of which 3444 (22.61%) presented positive urine cultures with more than 10<sup>5</sup> CFU/mL and 698 (4.58%) showed criteria of MDR-UTIs. The representative diagram of patient dynamics is illustrated in Figure 5.

**Figure 5.** Diagram of the screened and enrolled patients in the study.

Information on age, sex, and social–demographic status were registered for each patient; both hospitalized and ambulatory treated patients were considered for this study; thus, an exhaustive medical history for the second group was not available.

#### *4.2. Inclusion and Exclusion Criteria*

The inclusion criteria:


4. Standards of MDR-UTIs (nonsusceptible uropathogen to one or more antibiotic agents in three or more antimicrobial categories).

The exclusion criteria:


#### *4.3. Sample Collection, Bacterial Culture, Identification of Uropathogens, and Antibiotic Susceptibility Test*

The European and Romanian Association of Urology guidelines [10,39] on urological infections are followed to treat UTIs among patients from all clinics. In each case, a minimum 7–10 days were considered between the last antibiotic treatment and urine sampling for proper microbiological testing.

International Safety Standards [40] were followed for urine collecting techniques. After inoculation and incubation, microorganisms were identified based on specific Gram reactions, morphology, and biochemical characteristics. In all cases, the Clinical Laboratory Standard Institute (CLSI) [41] guidelines were followed for a comprehensive determination of sensitivity and resistance rates for each of the antibiotics tested to obtain the antibiogram.

Bacterial culture, the identification of uropathogens, and the antibiotic susceptibility test used were previously described in more detail [3,4,11].

#### *4.4. Statistical Analysis*

Data analysis was conducted using Microsoft Excel software (version 2020, Microsoft Corporation, Redmond, WA, USA); simple descriptive statistics were calculated. The relations of the variables were analyzed using frequency and percentage.

#### **5. Conclusions**

The current study acknowledged *E. coli* as the most common urinary pathogen among the MDR bacteria involved in UTIs among hospitalized and ambulatory patients. The incidence of MDR-UTIs increases with age distribution, peaking among the over 70 years old group of patients. In both the Gram-negative and Gram-positive groups, the highest resistance was noted for quinolones and β-lactams; alarmingly, the resistance to carbapenems rose, peaking with *P. aeruginosa*. Overall, the most common MDR resistance profiles were associated with aminopenicillins, quinolones, and cephalosporins. Together, *E. coli* and *Klebsiella* represented more than three-quarters of the identified MDR strains.

As MDR in urological infections are evolving, we strongly recommend the surveillance of resistance profiles and assessing the risk factors. A proper antibiotic administration policy should be implemented considering new MDR resistance data entailing the locoregional results.

**Author Contributions:** Conceptualization, R.-C.P., A.P. and C.B.C.; methodology, R.-C.P., S.N. and C.M.; validation, C.M. and R.-I.P.; formal analysis, C.M. and R.-I.P.; investigation, S.N., CM, R.-I.P. and C.B.C.; resources, S.N. and A.P.; data curation, R.-C.P. and A.P.; writing—original draft preparation, S.N., C.M., R.-I.P., and C.B.C.; writing—review and editing, R.-C.P. and A.P.; visualization, S.N. and C.B.C.; and supervision, R.-C.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki. The ethics committee from every hospital approved the protocol: Burghele Clinical Hospital (no.2/2019), Mures County Hospital (no. 6522/2020), and Elias University Hospital (no. 2517/2020).

**Informed Consent Statement:** The data collected retrospectively did not contain any personal information. For each patient, written informed consent was obtained.

**Data Availability Statement:** Data supporting the reported results are available from the authors.

**Acknowledgments:** The results from BCH, regarding MDR, were partially presented in the paper "Antibiotic resistance profile of common uropathogens implicated in urinary tract infections in Romania" published by Petca R.C. et al. in Farmacia, 2019, 67, 994–1004, doi:10.31925/farmacia. 2019.6.9. The results from BCH, EUH, and MCH, not regarding MDR, were partially presented in the papers: "A clinical perspective on the antimicrobial resistance spectrum of uropathogens in a Romanian male population" published by Chibelean C.B. et al. in Microorganisms, 2020, 8, 848, doi:10.3390/microorganisms8060848 and "Spectrum and antibiotic resistance of uropathogens in Romanian females" published by Petca R.C. in Antibiotics, 2020, 9, 472, doi:10.3390/antibiotics90804 72.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Short-Course Versus Long-Course Colistin for Treatment of Carbapenem-Resistant** *A. baumannii* **in Cancer Patient**

**Wasan Katip 1,2,\* , Suriyon Uitrakul <sup>3</sup> and Peninnah Oberdorfer 2,4**


**Abstract:** Carbapenem-resistant *Acinetobacter baumannii* (CRAB) is one of the most commonly reported nosocomial infections in cancer patients and could be fatal because of suboptimal immune defenses in these patients. We aimed to compare clinical response, microbiological response, nephrotoxicity, and 30-day mortality between cancer patients who received short (<14 days) and long (≥14 days) courses of colistin for treatment of CRAB infection. A retrospective cohort study was conducted in cancer patients with CRAB infection who received short or long courses of colistin between 2015 to 2017 at Chiang Mai University Hospital (CMUH). A total of 128 patients met the inclusion criteria. The results of this study show that patients who received long course of colistin therapy had a higher rate of clinical response; adjusted odds ratio (OR) was 3.16 times in patients receiving long-course colistin therapy (95%CI, 1.37–7.28; *p* value = 0.007). Microbiological response in patients with long course was 4.65 times (adjusted OR) higher than short course therapy (95%CI, 1.72–12.54; *p* value = 0.002). Moreover, there was no significant difference in nephrotoxicity (adjusted OR, 0.91, 95%CI, 0.39–2.11; *p* value = 0.826) between the two durations of therapy. Thirty-day mortality in the long-course therapy group was 0.11 times (adjusted OR) compared to the short-course therapy group (95%CI, 0.03–0.38; *p* value = 0.001). Propensity score analyses also demonstrated similar results. In conclusion, cancer patients who received a long course of colistin therapy presented greater clinical and microbiological responses and lower 30-day mortality but similar nephrotoxicity as compared with those who a received short course. Therefore, a long course of colistin therapy should be considered for management of CRAB infection in cancer patients.

**Keywords:** cancer patients; duration of treatment; colistin; propensity score analysis; multidrugresistant *Acinetobacter baumannii*

#### **1. Introduction**

Patients with cancer are at-risk of infections caused by antibiotic resistant Gramnegative bacteria. *Acinetobacter baumannii* is one of the most commonly reported nosocomial infections in cancer patients [1]. *A. baumannii* has been identified in patients with solid tumors, hematological malignancies, neutropenia, and those in the Intensive Care Unit (ICU) [2–4]. Infections caused by *A. baumannii* can be fatal in patients with suboptimal immune defenses, especially cancer patients [1,2]. Moreover, the prevalence and mortality rate of carbapenem-resistant *A*. *baumannii* (CRAB) has increased. The mortality rate of patients with cancer and multidrug-resistant (MDR) *A*. *baumannii* infection has reached 55% [5].

Colistin treatment for *Acinetobacter baumannii* infection is one of the most debated regimens [6]. Therefore, several alternative treatments are suggested, such as tigecycline,

**Citation:** Katip, W.; Uitrakul, S.; Oberdorfer, P. Short-Course Versus Long-Course Colistin for Treatment of Carbapenem-Resistant *A. baumannii* in Cancer Patient. *Antibiotics* **2021**, *10*, 484. https://doi.org/10.3390/ antibiotics10050484

Academic Editor: Pavel Bostik

Received: 9 March 2021 Accepted: 20 April 2021 Published: 22 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

amikacin, and sulbactam. Tigecycline is one of the active antibiotics against CRAB and appears to be a potential alternative therapeutic option for the treatment of CRAB [6,7]. However, tigecycline provides low concentration in plasma, and this limits its use in blood stream infections [6]. Furthermore, tigecycline has been shown to be inferior to the comparator drugs and has shown higher mortality rate in VAP patients [7]. For amikacin and sulbactam, although they have shown anti-CRAB efficacy, their nephrotoxicity and high resistance rate among CRAB limits their use [6]. Therefore, as compared with the other drugs, colistin is still safe and effective for the treatment of CRAB infection [6].

Colistin is one of the most widely prescribed medicines for the treatment of carbapenemresistant *A. baumannii* (CRAB). It has long been acknowledged that higher consumption of colistin resulted in higher risk of MDR bacteria, as well as higher treatment cost [6]. Theoretically, short-course treatment decreases ecological pressure and eliminates adverse effects without affecting the outcome [8]. Nonetheless, a subgroup of ventilator-associated pneumonia (VAP) patients who were infected with non-maturing Gram-negative microorganisms had higher recurrence of pulmonary infection with the short-course (8 days) treatment regimen than with the long-course regimen [9].

The duration of colistin treatment CRAB is typically ≥7 to 14 days [10–12]. However, CRAB is known as a significant and difficult-to-treat pathogen with complex resistance. This characteristic is a real challenge to all clinicians and leads to the use of colistin for longer than 2 weeks for established infections in many patients [12].

There was a randomized, open-label, clinical trial that studied 210 patients with lifethreatening infections due to extensively drug-resistant (XDR) *A. baumannii*. The recruited patients were randomly assigned to either colistin alone or colistin plus rifampicin groups. The primary end point of the study was overall 30-day mortality. Treatment had to be administered for at least 10 days and up to a maximum of 21 days [12].

However, the optimal treatment duration for a specific group of CRAB-infected cancer patients still remains to be determined. Therefore, the primary objective of this study was to compare clinical outcome, microbiological response, and nephrotoxicity between cancer patients receiving a short course (<14 days) and long course (≥14 days) of colistin for treatment of CRAB. The secondary objective of this study was to compare 30-day mortality rates between patients who received short and long courses of antimicrobial therapy for CRAB pneumonia.

#### **2. Results**

One hundred and twenty-eight cancer patients with CRAB infection were recruited in the study; there were 84 patients in the short course and 44 patients in the long course of colistin therapy. For median duration of therapy, the short-course group had 7 days duration (interquartile range (IQR), 5–9 days) while the long-course group had 14 days duration (IQR, 14–15 days). Overall, the median age was 62 years, and 78 patients (61%) were female. The majority of infectious disease was pneumonia (72%). Patients in the short and long courses of colistin therapy were comparable in most baseline demographics and clinical characteristics, although the numbers of patient in both group were different (Table 1). The number of patients who received short and long courses of colistin therapy is shown in Figure 1.

When assessing the outcomes and toxicity of colistin therapy, Fisher's exact test showed that the clinical and microbiological response of CRAB infection was higher in the long-course therapy group than in the short-course. There was no significant difference in nephrotoxicity between both patient groups (54 cases (64.29%) in short-course group and 27 cases (61.36%) in long-course group, *p* value = 0.847). Moreover, the 30-day mortality of CRAB infection was higher in the short course than in the long course of colistin therapy group (32 (38.10%) and 5 (11.36%), respectively, *p* value = 0.002), as shown in Table 2.


**Table 1.** Demographic and clinical characteristics of patients in short and long courses of colistin therapy.



IV, intravenous; SCr, serum creatinine; GFR, glomerular filtration rate; SD, standard deviation; GI, gastrointestinal tract; UTI, urinary tract infection; IQR, interquartile range. \* Patients with > 1 disease; \*\* Patients prescribed > 1 drug; # Other included intercostal drainage, surgical site infection.

**Figure 1.** Number of patients who received short and long courses of colistin therapy.


**Table 2.** Overall outcomes and toxicity in short and long courses of colistin therapy.

#### *Univariate and Multiple Logistic Regression Analyses of the Outcomes*

The results of univariate analysis indicate significant differences in clinical response, microbiologic response, and 30-day mortality rates between the short- and long-course groups. Table 3 shows similar nephrotoxicity rates in both groups. However, a significantly higher clinical response rate (odds ratio (OR), 3.16; 95% confidence interval (CI), 1.37–7.28; *p* value = 0.007), higher microbiological response rate (OR, 4.65; 95% CI, 1.72 to 12.54; *p* value = 0.002), and lower 30-day mortality rate (OR, 0.11; 95% CI, 0.03 to 0.38; *p* value = 0.001) were observed in patients who received long-course therapy as compared with short-course therapy based on multiple logistic regression analysis. However, the nephrotoxicity rates were not different (OR, 0.91; 95% CI, 0.39 to 2.11; *p* value = 0.826) (Table 3). Other predictors of clinical response were septic shock and Charlson score ≥ 4, and the predictor of microbiological response was septic shock. Age of 60 years or more could predict higher nephrotoxicity. Other independent risk factors for 30-day mortality were septic shock, Charlson score ≥ 4, and baseline Scr ≥ 1 mg/dl (Table 3).

The results of propensity score analysis using inverse probability weighting with variables associated with long-course therapy showed a significant difference in clinical response, microbiological response, and 30-day mortality rate (*p* value = 0.001, *p* value = 0.001 and *p* value = 0.001, respectively, Table 3). Factors that were used in the analysis were age, sex, vancomycin, amphotericin B, septic shock, Charlson score, comorbidities, stay in ICU during infection, mechanical ventilation during infection, baseline serum creatinine, source of CRAB infection, and type of malignancy.

**Table 3.** The outcomes of cancer patients receiving short-course and long-course colistin for CRAB infection (*n* = 128).



**Table 3.** *Cont.*

CI, confidence interval; ICU, intensive care unit; OR, odds ratio; IPW, inverse probability weighting.

#### **3. Discussion**

The results from this study point out that greater clinical and microbiological responses were achieved in cancer patients who received a long-course colistin therapy compared with a short-course. Additionally, no significant difference in nephrotoxicity rate between the two groups was detected. Furthermore, cancer patients who received long-course colistin therapy had lower 30-day mortality rate than the short-course. These findings were supported by logistic regression analysis and propensity score analysis using inverse probability weighting for both primary outcome (i.e., clinical response, microbiologic response, and nephrotoxicity) and secondary outcome (30-day mortality).

Guidelines of antimicrobial therapy are important tools that can help clinicians to make decisions about the duration of treatment. The optimization of antimicrobial duration might also be an important factor for management of CRAB infection in cancer patients. However, prior studies examining the impact of duration of antimicrobial therapy in cancer patients with CRAB infection remain limited. Most randomized controlled trials on duration of antibiotic therapy do not include difficult-to-treat patients or pathogens, such as immunocompromised patients, critically ill patients, specific infection foci, *P*. *aeruginosa* infection, and *A*. *baumannii* infection [13]. Therefore, the optimal duration of antimicrobial therapy for cancer patients with CRAB infection remains uncertain and needs to be investigated.

In some recent studies, the duration of antibiotic therapy for *Acinetobacter baumannii* infection was not well defined [10–14]. In previous studies, the definition typically used for short-course treatment was less than 10 days, while long-course treatment was typically defined as more than or equal to 10 days [15–17]. However, some studies defined treatment of longer than 7 or 8 days as a long-course duration [9,18]. In the clinical practice guidelines by the Infectious Diseases Society of America and the American Thoracic Society (IDSA/ATS), a 7-day course of antimicrobial therapy was strongly recommended for hospital-acquired and ventilator-associated pneumonia rather than a longer duration, even in non-fermentative Gram-negative bacilli infection including *P. aeruginosa* and *A. baumannii* [19]. Remarkably, *A. baumannii* might be one of the most debated pathogens for antibiotic treatment duration; many clinicians usually consider continuation of antibiotic therapy for up to 2 weeks in many patients with established infection [10–14]. However, using antibiotics for 14 days was classified as a long treatment duration in many studies and guidelines [9–12,14–19]. Based on the mentioned information, especially in immunocompromised patients, a 14-day duration was defined as long course therapy; this cut-off was longer than most studies but was set based on the principle of immunocompromised status of patients [9–12,14–19]. Nevertheless, with regard to the patient distribution, where approximately 70% of patients in the long-course group received 14 days of treatment, the

effect on the 14-day patients could dominate the effect of other patients. This, therefore, should be considered because the results might be altered if another cut-off were used.

The difference in duration of treatment between 8 and 15 days was analyzed by subgroup in a randomized controlled trial that was conducted in 401 patients with VAP. There was no significant difference in 28-day mortality rate between the two groups with non-fermentative Gram-negative bacilli; the 8-day group had 23.4% and the 15-day group had 30.2% 28-day mortality rates (−6.8% difference; 90% CI: −17.5–4.1). Likewise, no significant difference was observed in clinical response. However, in the 8-day group, the rate of recurrent pulmonary infection was higher than in the other group: 40.6% and 25.4%, respectively (15.2% difference; 90% CI: 3.9–26.6) [9]. Therefore, longer antibiotic therapy might be necessary and should be considered in patients with VAP caused by non-fermentative Gram-negative bacilli (1.7% was *A. baumannii*) [9].

Multiple logistic regression and propensity score analysis showed that patients in this study who received a long course of colistin were more likely than those who received a short course to experience clinical response (61% and 26%, respectively; *p* value = 0.005) and microbiological response (61% and 34%, respectively; *p* value = 0.013). These data further support the notion that longer colistin therapy is highly useful in the treatment of MDR *A. baumannii* in cancer patients.

The results of this study were consistent with the retrospective study by Nelson et al. [20] that included 117 patients with a short course and 294 patients with a long course of antimicrobial therapy (median duration of 8.5 and 13.3 days, respectively) for uncomplicated Gram-negative bloodstream infection. The propensity score adjusted risk of treatment failure was higher in patients with a short course of antimicrobial agents compared with a long course (HR 2.60, 95% CI: 1.20–5.53, *p* value = 0.02). Moreover, the compromised immune status was found to be a risk factor for treatment failure (HR 4.30, 95% CI: 1.57–10.80, *p* value = 0.006). However, the abovementioned study was not conducted using a specific treatment for *A. baumannii* infection, so this result might not be applicable in this pathogen [20].

Another study by Hachem et al. [21] evaluated the efficacy of colistin in cancer patients, mostly patients with hematological malignancy treated for multidrug-resistant *Pseudomonas aeruginosa* infection; it was found that clinical cure rate was 61% and median treatment duration of colistin was 20 days (range from 5–58). However, the objective of this study was not to investigate duration of treatment, so the result found for duration in the Hachem et al. [21] study could not indicate a suitable duration of treatment.

Nazer et al. [22] described that microbiological clearance was observed in 51 patients (66.2%) on day 13 ± 9 (mean ± SD) after starting colistin therapy. They found that 35 out of 89 patients with cancer and CRAB infection (39.3%) met the RIFLE criteria for nephrotoxicity, with a mean of 9 ± 7 days from the initiation of colistin therapy. Duration of IV colistin was 15.8 ± 11 days (mean ± SD). However, the mentioned study was not designed to compare the courses of antibiotic therapy, so there was no control group for comparison [22].

Theoretically, in cases where a patient has good clinical response without clinical features of infection, the duration of antibiotic therapy should be shortened from the traditional 14–21 days (long course) to a period as short as 7 days (short course) because a short-course therapeutic approach can reduce ecological pressure and diminish side effects. However, cancer patients with CRAB in this recent study had lower clinical and microbiological responses when using a short-course treatment regimen. There were several reasons for the requirement of a long course (>14 days) of antibiotic therapy in CRAB-infected patients. First, one of the potential causes of delayed eradication of CRAB from the body might be attributed to chemotherapy administration, which diminishes neutrophil function [23]. Since neutrophils have several essential roles in host resistance to respiratory infection from many organisms as well as from *A. baumannii* [24], cancer patients infected with CRAB should considered for longer antibiotic therapy. Secondly, antibacterial therapy aims to reduce the number of bacteria, but in most cases, antibacterial

efficacy relies on the immune system to eliminate bacteria completely. However, patients with malignancy can sometimes have neutropenia and lack of the defense mechanisms that are present in patients with intact immune systems. Therefore, a higher pharmacokinetic (PK)–pharmacodynamic (PD) target and longer antibiotic therapy may be required in immunosuppressed patients such as cancer patients. This hypothesis is supported by the neutropenic mouse thigh model, where it was found that neutropenia increased the required magnitude of PK/PD index by 50% to 100% (i.e., 1.5–2.0-fold) [25,26].

The poor outcomes reported in this study might be caused by underlying malignant diseases in patient population, but there were also other potential contributing factors. After adjusting the potential confounders, other predictors of clinical response were detected, including septic shock (OR, 0.3; 95% CI, 0.13 to 0.68; *p* value = 0.004) and Charlson score ≥ 4 (OR, 0.34; 95% CI, 0.16 to 0.74; *p* value = 0.006), and the predictor of microbiological response was septic shock (OR, 0.37; 95% CI, 0.15 to 0.88; *p* value = 0.024). The other independent risk factors for 30-day mortality were septic shock (OR, 6.20; 95% CI, 1.67 to 23.10; *p* value = 0.007), Charlson score ≥ 4 (OR, 7.12; 95% CI, 2.40 to 21.10; *p* value = 0.001), and baseline Scr ≥ 1 mg/dl (OR, 2.85; 95% CI, 1.02 to 7.97; *p* value = 0.046).

In the present study, 54 (64.29%) and 27 (61.36%) patients developed nephrotoxicity during short course and long course therapy (*p* value = 0.847), respectively. The incidence of nephrotoxicity observed in this study was within the range that has been reported in previous studies, ranging from 20% to 69% [27–29]. The other factor that was found correlated with nephrotoxicity was age equal or greater than 60 years (OR, 2.31; 95%CI, 1.01–5.33).

Regarding antimicrobial stewardship, reducing the length of antibiotic course might be effective in reducing antibiotic resistance through the abovementioned mechanism. However, this study observed higher rates of clinical and microbiological response, similar nephrotoxicity, and a lower rate of 30-day mortality in the long-course therapy than in shortcourse therapy. Therefore, in the specific group of cancer patients with CRAB infection, consideration of longer treatment was necessary.

This study differed from other previously reported studies since all patients here had underlying cancer. Eighty-six percent of the patients had solid tumor, and the rest had hematologic malignancy. Additionally, all patients in this study had CRAB infection without any other infections.

There were some limitations in this study. Firstly, the methodology of this study was retrospective, which could possibly have allowed unknown variables to affect the results. However, sensitivity analysis was used to adjust for the suspected confounding factors, and the obtained results still led to the same conclusion. Secondly, in a previous prospective study, the incidence of acute kidney injury (AKI) during therapy was strongly correlated with baseline renal function and plasma colistin concentration. Patients with a higher creatinine clearance (CLCR) had higher rates of AKI than those with a lower CLCR [30]. As this study did not measure plasma colistin concentration, there was a lack of pharmacokinetic information, which was considered one of limitations of this study. However, therapeutic drug monitoring, especially for colistin, is not routinely performed in clinical practice in Thailand. RIFLE criteria were therefore used to assess nephrotoxicity outcome. The baseline renal functions were not significantly different between short- and long-course treatment groups. Moreover, logistic regression analysis and propensity score analysis using inverse probability weighting with variables associated with long-course therapy were performed to adjust any variables that were different between short- and longcourse treatment groups. Thirdly, this study did not describe the type of chemotherapy that patients might have been administered, so the results of this study should be carefully interpreted based on cancer patient status. Fourthly, this study did not explore patient blood counts, which might affect the outcomes.

#### **4. Materials and Methods**

This retrospective cohort study was conducted at Chiang Mai University Hospital (CMUH), Chiang Mai, Thailand, from January 2015 to August 2017. The methodology was approved by the ethics committee on human research of the Faculty of Medicine, Chiang Mai University, with a waiver of informed consent for retrospective data collection under the condition of anonymously stored data collected. Medical chart records and microbiology laboratory data of cancer patients with CRAB infection were reviewed. The criteria used to identify and classify infection were outlined by the Center for Disease Control and Prevention (CDC) [31]. The inclusion criteria were age equal or greater than 18 years, colistin treatment for more than 2 days for documented CRAB infection, and receipt of only one course of colistin treatment. The exclusion criteria were the presence of other types of Gram-negative infection, and treatment with hemodialysis or renal replacement therapy. The primary exposure was duration of antibiotic treatment, divided into short-course and long-course therapy. The first day of colistin therapy was defined as the day that CRAB cultures were obtained. Short-course antimicrobial therapy was defined as a total duration of colistin <14 days after the first day. Long-course therapy was defined as a total duration of colistin ≥14 days after the first day. The duration was the first day of intravenous colistin use until fully discontinuation.

The following information was collected: age, sex, type and site of malignancy, comorbidities, dates of admission and discharge, medical history, ICU admission, the need for mechanical ventilation, Charlson score, previous nosocomial infection, clinical course, length of hospital stay, and hospital discharge diagnosis. Data on positive bacterial cultures, antibiotic susceptibilities, colistin minimum inhibitory concentration (MIC), source of CRAB infection, subsequent culture results, and the duration of colistin therapy were also collected. Side effects related to colistin, baseline serum creatinine, baseline GFR, total colistin dose, and the development of septic shock during the evolution of CRAB infection were recorded, as well as the concomitant nephrotoxic medications that could potentially cause nephrotoxicity, hospital discharge diagnosis, outcome, nephrotoxicity (base on RIFLE criteria), and 30-day mortality.

CRAB was defined as *A. baumannii* that was resistant to carbapenems but sensitive to colistin. Dosage of antibiotic regimens was based on the respective hospital guidelines: LD colistin 300 mg of colistin base activity (CBA) once at the start of treatment course and then 150 mg of CBA every 12 h. Colistin administration was adjusted according to renal function. Dose and interval were adjusted according to Cockcroft and Gault creatinine clearance estimates if patients had moderate-to-severe renal impairment (creatinine clearance rate, <50 mL/min). For example, a loading dose of 300 mg followed by maintenance dose of 150! mg CBA every 24 h was administered for creatinine clearance rate of 20–50 mL/min, or 150 mg CBA every 48 h was administered for creatinine clearance rate of <20 mL/min.

#### *4.1. Outcome Assessment*

Three primary outcomes in this study were clinical response, microbiological response, and nephrotoxicity after treatment. Clinical response of treatment was assessed by resolution or partial resolution of the present symptoms and signs of CARB infection at the end of colistin treatment. Patients who failed to achieve all criteria for clinical response were defined as clinical failures. Microbiological response was defined as obtaining two consecutive negative CARB cultures from the site of infection after the initial positive culture, whereas microbiological failure was defined as persistence of CARB in the subsequent specimen cultures. Renal toxicity was defined as detection of any stage of acute kidney injury outlined in the RIFLE classification [32]. The secondary outcome of the study was 30-day mortality, which was defined as death within 30 days after initial colistin treatment for CARB infection.

#### *4.2. Antimicrobial Susceptibility Testing*

*A. baumannii* was discovered using traditional cultures and biochemical methods at CMUH's Clinical Microbiology Division. The Clinical and Laboratory Standards Institute (CLSI) protocol [33] was used to assess antimicrobial susceptibility. Antibiotic susceptibility to *A. baumannii* was determined using the VITEK 2 method, and colistin susceptibility was determined using broth microdilution, with resistance identified as a colistin MIC breakpoint >2 mg/L. The Vitek 2 system (bioMerieux, Marcy I 'Etoile, France) is a fully automated system that uses a fluorogenic approach to identify organisms and a turbidimetric process to assess susceptibility. Antimicrobial susceptibility testing with VITEK 2 demonstrated high compliance with standard methods for evaluating antimicrobial MICs, with a time benefit of hours to days and increased reproducibility [34,35].

#### *4.3. Statistical Analysis*

Stata 14 software was used to analyze all of the results (Stata-Corp, College Station, TX, USA). The duration of colistin therapy was mainly compared between the two treatment groups.

General characteristics and basic information of patients were analyzed with descriptive statistics, including percentage, frequency, average, and standard deviation. Using Fisher's exact test, the average comparison case of sample basic data and the average of other statistical methods was the independent *t* test when data were distributed normally, and the Mann–Whitney U test when data were not normally distributed. The significance level was set as 0.05. Fisher's exact test was used to compare differences in rates of clinical response, microbiologic response, nephrotoxicity, and 30-day mortality between short-course and long-course colistin therapy.

Furthermore, factors that might affect the four outcomes, i.e., clinical response, microbiologic response, nephrotoxicity, and 30-day mortality, were adjusted using logistic regression. The evaluated factors included age, comorbidities, stay in an intensive care unit (ICU) during infection, course of colistin therapy, septic shock, Charlson score, baseline serum creatinine, type of malignancy, mechanical ventilation during infection, type of nephrotoxic medication, and source of CRAB infection. Firstly, univariate analysis was performed to evaluate the predictive effect of each factor. Next, any factors with a *p* value of < 0.25 from univariate test were included in a full multiple logistic model. Lastly, factors were removed from the model one at a time until all factors remaining in the model had 5% significance level, except that course of colistin therapy remained in the model, regardless of its *p* value.

For sensitivity analysis, a propensity score for courses of colistin therapy and estimated ORs with inverse probability weighting methods was developed using variables likely to influence the outcomes of both primary outcome (i.e., clinical response, microbiologic response and nephrotoxicity) and secondary outcome (30-day mortality).

#### **5. Conclusions**

The results of this study suggest that a long course of colistin was preferred in the treatment of CRAB infection in a cancer population, based on higher rates of clinical and microbiological response. Furthermore, cancer patients who received a long course of colistin had a lower 30-day mortality rate but the same nephrotoxicity rate. A long course of colistin therapy, therefore, should be considered for the management of CRAB infection in cancer patients.

**Author Contributions:** Conceptualization, W.K.; data curation, W.K.; formal analysis, W.K.; investigation, W.K.; methodology, W.K.; project administration, W.K.; software, W.K.; supervision, P.O.; validation, P.O.; writing—original draft, W.K.; writing—review and editing, W.K. and S.U. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of the Faculty of Medicine, Chiang Mai University (NONE-2560-04839).

**Informed Consent Statement:** Patient consent was waived due to retrospective data collection under the condition of anonymously stored data collected.

**Data Availability Statement:** The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

**Acknowledgments:** This research work was partially supported by Chiang Mai University.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

