High prevalence of colistin heteroresistance in Pseudomonas aeruginosa bloodstream isolates and associated clinical factors Authors: Ekin Kırbaş, Muhammed Cihan Işık, Rashad Ismayilov, Serhat Ünal, Banu Sancak Published: 2026 ABSTRACT AbstractPurposeCarbapenem-resistant Pseudomonas aeruginosa (CRPA) has been classified by the World Health Organization as one of the most alarming antimicrobial-resistant microorganisms. Despite the availability of newer therapeutic options, polymyxins remain important last-resort agents for the treatment of CRPA infections in many settings. With the extended use of polymyxins in the treatment of extensively drug-resistant bacterial infections, both colistin resistance rates, colistin heteroresistance (CHR) rates and antibiotic treatment failures have risen. In this study, we aimed to: (i) determine the prevalence of CHR in P. aeruginosa bloodstream isolates, (ii) compare the accuracy of the antibiotic gradient test (GT) with the population analysis profiling (PAP) method, and (iii) identify clinical risk factors for CHR.MethodsCHR was investigated among 70 colistin-susceptible P. aeruginosa strains isolated from blood cultures using PAP and GT methods. Clinical and demographical data were retrospectively reviewed, and risk factors for CHR were analyzed using logistic regression.ResultsWe detected a remarkably high prevalence of CHR (61.4%) among colistin-susceptible P. aeruginosa bloodstream isolates. The gradient test failed to detect any of the heteroresistant strains. Our analysis revealed that age ≥ 60 years, lack of recent hospitalization, and a colistin MIC of 1 mg/L, compared with 2 mg/L, were independently associated with an increased risk of CHR. CHR was not significantly associated with 30-day or 90-day mortality.ConclusionDespite the microbiological importance of heteroresistance, the clinical impacts and underlying mechanisms of this phenotype remain to be elucidated.Keywords: Pseudomonas aeruginosa, Bloodstream infections, Colistin heteroresistance, Clinical impact Introduction Pseudomonas aeruginosa stands out as a formidable pathogen, causing serious nosocomial infections associated with substantial morbidity and mortality. The World Health Organization (WHO) has classified carbapenem-resistant P. aeruginosa (CRPA) as a critical priority pathogen because of the limited therapeutic options available for its treatment [1]. Managing P. aeruginosa infections poses significant challenges due to its diverse intrinsic resistance mechanisms and remarkable capacity to acquire additional resistance determinants [2]. Importantly, bloodstream infections represent one of the most severe manifestations of P. aeruginosa infection and are associated with high mortality rates. Although current guidelines generally favor newer β-lactam/β-lactamase inhibitor combinations over polymyxins for the treatment of CRPA infections [3], polymyxins remain important last-resort agents in many resource-limited settings where access to these newer therapies is restricted [4]. Under these conditions, with the increased rates of multi-drugresistant P. aeruginosa and CRPA infections, polymyxins often remain among the few available therapeutic options despite their well-recognized toxicity [5]. Heteroresistance refers to the presence of resistant subpopulations within an phenotypically susceptible bacterial isolate [6, 7]. Conventional susceptibility testing methods, including broth microdilution (BMD), may fail to detect heteroresistance because resistant subpopulations are often present at low frequencies within the susceptible main population [6, 7]. This phenomenon may facilitate bacterial survival during antimicrobial exposure and contribute to treatment failure despite susceptibility results indicating activity of the administered agent. Furthermore, resistant subpopulations may rapidly become dominant during therapy, potentially resulting in microbiological failure despite initial susceptibility results [8]. Despite the microbiological importance of heteroresistance, the clinical impacts and underlying mechanisms of this phenotype remain to be elucidated. Although CHR has been described in P. aeruginosa, reported prevalence rates vary substantially across studies, likely due to differences in study populations and heteroresistance definitions. In this study, we aimed to: (i) determine the prevalence of CHR in P. aeruginosa bloodstream isolates, (ii) compare the performance of the antibiotic gradient test with PAP for detecting CHR, and (iii) identify clinical risk factors associated with CHR. Materials and methods Bacterial strains This single-center retrospective study was conducted at the 1200-bed Hacettepe University, Adult and Oncology Hospital in Ankara, Türkiye. Seventy non-duplicate colistin-susceptible P. aeruginosa strains were recovered from blood cultures sent to the Central Microbiology Laboratory of HUH from January 2014 to July 2018. From 2014 to 2017, the isolates were identified to the species level by using VITEK® MS (bioMérieux, Marcy-l’Étoile, France) and as of 2018 by MALDI Biotyper® (Bruker Diagnostics, Bremen, Germany). Strains with colistin MIC values of 1 and 2 mg/L were included in the study. Only one bacterial strain per patient was included. Demographic and clinical data of the patients were obtained from the medical records and laboratory databases in the hospital server system. Antibiotic susceptibility testing Colistin and meropenem minimum inhibitory concentrations (MICs) were determined using the broth microdilution method. Colistin sulfate (Product No. C4461, Sigma–Aldrich, Germany) and meropenem (Product No. PHR1772, Sigma–Aldrich, Germany) powders were dissolved in sterile water according to the manufacturers’ instructions. Two-fold serial dilutions were prepared to obtain concentration ranges of 0.06–128 mg/L for colistin and 0.03–64 mg/L for meropenem. MIC results were interpreted according to the European Committee on Antimicrobial Susceptibility Testing clinical breakpoints (Version 15.0, 2025; breakpoints for Pseudomonas spp.: susceptible: MIC ≤ 2 mg/L; resistant: MIC > 2 mg/L) [9]. CHR in P. aeruginosa strains with colistin MICs of 1 mg/L and 2 mg/L was assessed using the PAP method. Population analysis profiling (PAP) method For this purpose, Mueller-Hinton agar (MHA, Product No.: CM0337, Oxoid, United Kingdom) plates with 1, 2, 4, 8, 16, 32 and 64 mg/L colistin sulphate (Product No. PHR1772, Sigma–Aldrich) concentrations were prepared. For each isolate, serial 10-fold saline dilutions, ranging from 108 to 102 CFU/mL, were prepared from overnight cultures. 100 µL of each dilution was inoculated on MHA plates with two-fold serial concentrations of colistin. Plates were incubated at 35 ± 2 °C for 48 h, and colonies were counted at 24 and 48 h. HR was considered, if the ratio of the lowest colistin concentration exhibiting maximum inhibition to the highest non-inhibitory colistin concentration was equal or more than eight-fold, with the frequency of resistant subpopulation in a bacterial colony at least 1 × 10− 7 CFU/mL [5, 6]. The frequency of resistant subpopulations was determined by dividing the number of P. aeruginosa colonies grown on the MHA plate with the highest colistin concentration to the number of the colonies on the antibiotic-free plate. Escherichia coli ATCC 25,922 and E. coli NCTC 13,846 were included as quality-control strains during PAP experiments. Antibiotic gradient test method Bacterial suspensions with a 0.5 McFarland standard were spread on MHA plates using a sterile cotton swab. Colistin gradient test strips (bioMérieux, Marcy-l’Étoile, France) were placed on the inoculated plates. After incubation at 35 ± 2 °C for 48 h, the presence of bacterial growth within the inhibition zone of the colistin GT strip was examined. Defining the stability of the heteroresistance To evaluate the stability of the HR phenotype colonies from PAP plates with the highest drug concentration were subcultured daily in an antibiotic-free medium for seven consecutive days, following approaches commonly used in previous heteroresistance studies. Colistin BMD was performed on days 1 and 7 to assess CHR stability. Isolates remaining resistant after 7 days were classified as ‘stable CHR,’ while those becoming susceptible were ‘unstable CHR.’ Statistical analysis The statistical analysis of our study was performed using the Statistical Package for the Social Sciences software (IBM SPSS Statistics, Version 26). The median (interquartile range [IQR]) was used for non-normally distributed data. Categorical variables were expressed as percentages. Continuous variables were compared using the Mann–Whitney U test. Fisher’s exact test and the chi-squared test were used for categorical comparisons. To identify risk factors associated with CHR, a univariate logistic regression analysis was first performed. Variables with a p-value < 0.25 in the univariate analysis or considered clinically relevant were included in the multivariable model. Multicollinearity among independent variables was assessed using the Phi coefficient, with a correlation coefficient (r) > 0.60 considered indicative of significant multicollinearity. Additionally, the Variance Inflation Factor (VIF) was utilized to evaluate multicollinearity in the final multivariable model and all variables in the final model had a VIF value of less than 2, indicating no significant multicollinearity. Potential interactions between predictors were examined using the Breslow-Day test, while confounding effects were evaluated with the Mantel-Haenszel test. The final multivariable risk model was constructed using the backward stepwise likelihood ratio method. The overall significance of the model was assessed with the Omnibus Test of Model Coefficients, and a p-value < 0.05 indicated that the model was statistically significant. Model fit was evaluated using the Hosmer–Lemeshow test, with a p-value > 0.05 suggesting good fit to the data. Nagelkerke R² was used to assess the explanatory power of the model, and − 2 log likelihood (–2LL) was reported as a measure of model fit. Results were presented as odds ratios (ORs) with 95% confidence intervals (CIs), with multivariable results expressed as Adjusted Odds Ratios (aORs). Survival analysis for 30-day and 90-day mortality was performed using the Kaplan-Meier method, and survival distributions between the groups were compared using the log-rank test. A p-value of less than 0.05 was considered as statistically significant. Results Population analysis profiling results CHR was detected in 43 of 70 (61.4%) of the colistin-susceptible P. aeruginosa strains. The prevalence of CHR was 68.6% (n = 35) among the isolates with a colistin MIC of 1 mg/L and 42.1% (n = 8) among the isolates with a colistin MIC of 2 mg/L. The frequency of the colistin-resistant subpopulations in the main colistin-susceptible populations ranged from 1.1 × 10− 7 to 6.0 × 10− 6 CFU/mL. The PAP curves of CHR P. aeruginosa strains demonstrated a slight shift to the right at higher colistin concentrations (Fig. 1a). In one of the strains (Isolate-P132), we detected subpopulations with different morphologies (small and large colonies) on PAP plates containing high colistin concentration (Fig. 1b). No growth occurred within the inhibition zone of the colistin GT strips for any of the isolates. Table 1 shows the PAP results. ‘Skipped wells’ phenomenon, which is described as the absence of bacterial growth at lower antibiotic concentrations of BMD microplate, but the presence of bacterial growth at higher antibiotic concentrations, and linked to HR phenotype, was not detected with any of the isolates. (A) PAP curves of two selected colistin heteroresistant strains (P2, P241) and E. coli NCTC 13846, (B) Colistin heteroresistant P. aeruginosa isolates showing large and small colonies on PAP plate (red arrow: large colonies, green arrows: small colonies). (CFU/mL, Colony forming units per milliliter; log10, logarithm with a base equal to 10; mg/L, milligrams per liter Table 1The colistin MIC values, and PAP results of colistin-heteroresistant P. aeruginosa strainsIsolate No.Colistin MIC of the main pop.The max. antibiotic con. of the PAP plate growth observedGrowth within the inhibiton zone of GTFrequency of the subpop. (CFU/mL)Colistin MICs after 7 days of cultivationColistin MIC: 1 mg/LP218-4.2 × 10− 72P1818-1.3 × 10− 74P21116-2.1 × 10− 72P3618-2.5 × 10− 74P4614-1.5 × 10− 72P59132-6.5 × 10− 716P6818-2.5 × 10− 72P9618-6.7 × 10− 72P139132-3.7 × 10− 74P141132-1.3 × 10− 71P143116-7.5 × 10− 72P149132-1 × 10− 62P154116-8.7 × 10− 74P15718-4 × 10− 78P162116-1.2 × 10− 64P165116-5.3 × 10− 74P16614-2 × 10− 72P16718-6.3 × 10− 72P17218-1.7 × 10− 78P182164-1.7 × 10− 72P18718-1.5 × 10− 71P188164-1.2 × 10− 7128P18918-4.4 × 10− 78P19018-1.5 × 10− 71P19114-3.8 × 10− 70.5P193116-2 × 10− 716P194164-1.5 × 10− 616P210116-6 × 10− 61P21214-1.1 × 10− 71P21618-1.8 × 10− 70.5P21918-1.3 × 10− 71P221116-1.3 × 10− 71P235116-2.0 × 10− 716P236116-1.1 × 10− 74P241164-1.3 × 10− 72Colistin MIC: 2 mg/LP11524-1 × 10− 71P11928-1.5 × 10− 74P132232-6.6 × 10− 716*0.5**P14628-3 × 10− 71P18028-1.5 × 10− 72P185232-1.1 × 10− 78P19728-1.5 × 10− 72P237216-2.5 × 10− 74CFU/mL, colony-forming unit per milliliter; con., concentration; GT, gradient test; mg/L, milligrams per liter; MIC, minimum inhibitory concentration; No., numero; pop., population; *, small colony morphology; subpop., subpopulation; **; large colony morphology; -, absent The colistin MIC values, and PAP results of colistin-heteroresistant P. aeruginosa strains CFU/mL, colony-forming unit per milliliter; con., concentration; GT, gradient test; mg/L, milligrams per liter; MIC, minimum inhibitory concentration; No., numero; pop., population; *, small colony morphology; subpop., subpopulation; **; large colony morphology; -, absent Stability of heteroresistance After seven consecutive subculturing of the colistin-resistant subpopulations on 5% sheep blood agar plates, the colistin BMD test showed stable CHR in 54.2% of isolates and unstable CHR in 45.8%. Following a week-long cultivation of small and large colonies of Isolate-P132 on an antibiotic-free medium, it was observed that colonies exhibiting diverse morphologies possessed varying minimum inhibitory concentrations. Small colonies had a colistin MIC of 16 mg/L (colistin-resistant), while large colonies had a MIC of 0.5 mg/L (COL-susceptible). Demographic and clinical features of CHR-positive and CHR-negative patients Among patients, 31 (44%) were female and the median age was 60 years (IQR: 50–71). Before the detection of P. aeruginosa bacteremia, the CHR-positive group had a significantly higher median age (65 years; IQR 55–74) than the CHR-negative group (p = 0.003). The median Charlson comorbidity index was three with no significant difference between the two groups (Table 2). No significant difference was observed between the CHR-positive and CHR-negative groups in the presence of severe immunosuppressive status, history of invasive interventional procedures, and history of ICU admission. However, hospitalization within the past three months was lower in CHR-positive cases (p = 0.05). Table 2Clinical characteristics of patients with colistin susceptible P. aeruginosa in blood isolates grouped by colistin heteroresistance statusCharacteristics n (%)Total,n = 70CHR-Negative, n = 27CHR-Positive,n = 43P valueFemale sex31 (44.3)11 (40.7)20 (46.5)0.636Age, median (IQR)60 (50–71)54 (43–61)65 (55–74)0.003aGrouped Age0.004b< 60≥ 6034 (48.6)36 (51.4)19 (70.4)8 (29.6)15 (34.9)28 (65.1)CCI, median (IQR)3 (0–10)4 (0–10)3 (0–9)0.478Severe Immunocompromised status19 (27.1)6 (22.2)13 (30.2)0.463Prolonged Neutropenia10 (52.6)1 (16.7)9 (69.2)HSCT6 (31.6)4 (66.7)2 (15.4)SOT3 (15.8)1 (16.7)2 (15.4)Invasive intervention history (past 3 months)39 (55.7)15 (55.6)24 (55.8)0.983Hospitalisation history (past 3 months)39 (55.7)19 (70.4)20 (46.5)0.050ICU Admission (past 3 months)14 (20.0)7 (25.9)7 (16.3)0.326Pseudomonal colonisation (past 6 months)Non-MDRMDRXDR9 (14.8)7 (77.8)1 (11.1)1 (11.1)4 (18.2)3 (75)0 (0)1 (25)5 (12.8)4 (80)1 (20)0 (0)0.710Antibiotic history (past 3 months)45 (72.6)19 (82.6)26 (66.7)0.174Anti-pseudomonal antibiotic history (past 3 months)38 (61.3)16 (69.6)22 (56.4)0.304Colistin history (past 3 months)13 (21)5 (21.7)8 (20.5)1.000Source of infection0.895PneumoniaPrimer bacteraemiaIntrabdominalUrinarySoft tissueOther23 (32.9)18 (25.7)13 (18.6)6 (8.6)5 (7.1)5 (7.1)8 (29.6)8 (29.6)4 (14.8)2 (7.4)3 (11.1)2 (7.4)15 (34.9)10 (23.3)9 (20.9)4 (9.3)2 (4.7)3 (7)Presence of sepsis37 (52.9)16 (59.3)21 (48.8)0.395Meropenem Resistance30 (42.9)13 (48.1)17 (39.5)0.478Colistin MIC (mg/L)0.043b1251 (72.9)19 (27.1)16 (59.3)11 (40.7)35 (81.4)8 (18.6)Anti-pseudomonal antibiotics administered64 (91.4)25 (92.6)39 (90.7)0.783In-vitro active antibiotics administered55 (78.6)21 (77.8)34 (79.1)0.898Colistin-based therapy13 (18.6)4 (14.8)9 (20.9)0.522Colistin monotherapy3 (23.1)1 (25)2 (22.2)0.913Colistin duration, median (IQR)8 (2–13)7.5 (4–9.5)12 (1–15.5)0.938Clinical response (n = 62)*32 (51.6)8 (34.8)24 (61.5)0.042bMicrobiological outcome0.153PersistentEradicatedUndetermined3 (4.3)27 (38.6)40 (57.1)2 (7.4)7 (25.9)18 (66.7)1 (2.3)20 (46.5)22 (51.2)Mortality, 30-day39 (55.7)18 (66.7)21 (48.8)0.144Mortality, 90-day44 (62.9)21 (77.8)23 (53.5)0.041bCHR, colistin heteroresistance; IQR, interquartile range; CCI, Charlson comorbidity index; HSCT, hematopoietic stem cell transplantation; SOT, solid organ transplantation; ICU, intensive care unit; MDR, multiple drug-resistance; XDR, extensively drug-resistant; MIC, minimal inhibitory concentrationaMann-Whitney U-testbχ2 test or Fisher’s exact test*Clinical response was assessed in a total of 62 patients who received anti-pseudomonal treatment and who could be followed up (two patients were transferred to another hospital) Clinical characteristics of patients with colistin susceptible P. aeruginosa in blood isolates grouped by colistin heteroresistance status < 60 ≥ 60 34 (48.6) 36 (51.4) 19 (70.4) 8 (29.6) 15 (34.9) 28 (65.1) Pseudomonal colonisation (past 6 months) Non-MDR MDR XDR 9 (14.8) 7 (77.8) 1 (11.1) 1 (11.1) 4 (18.2) 3 (75) 0 (0) 1 (25) 5 (12.8) 4 (80) 1 (20) 0 (0) Pneumonia Primer bacteraemia Intrabdominal Urinary Soft tissue Other 23 (32.9) 18 (25.7) 13 (18.6) 6 (8.6) 5 (7.1) 5 (7.1) 8 (29.6) 8 (29.6) 4 (14.8) 2 (7.4) 3 (11.1) 2 (7.4) 15 (34.9) 10 (23.3) 9 (20.9) 4 (9.3) 2 (4.7) 3 (7) 1 2 51 (72.9) 19 (27.1) 16 (59.3) 11 (40.7) 35 (81.4) 8 (18.6) Persistent Eradicated Undetermined 3 (4.3) 27 (38.6) 40 (57.1) 2 (7.4) 7 (25.9) 18 (66.7) 1 (2.3) 20 (46.5) 22 (51.2) CHR, colistin heteroresistance; IQR, interquartile range; CCI, Charlson comorbidity index; HSCT, hematopoietic stem cell transplantation; SOT, solid organ transplantation; ICU, intensive care unit; MDR, multiple drug-resistance; XDR, extensively drug-resistant; MIC, minimal inhibitory concentration aMann-Whitney U-test bχ2 test or Fisher’s exact test *Clinical response was assessed in a total of 62 patients who received anti-pseudomonal treatment and who could be followed up (two patients were transferred to another hospital) Nine patients (14.8%) had a history of P. aeruginosa colonization within six months. Additionally, 38 patients (61.3%) had received anti-pseudomonal antibiotics within the last three months, and 13 patients (21%) had a history of colistin use. There was no significant difference between the two groups regarding prior exposure to anti-pseudomonal antibiotics or colistin. At the time of P. aeruginosa bacteremia detection, 52.9% of the patients were in sepsis. Pneumonia (n = 23, 32.9%) and primary bacteremia (n = 18, 25.7%) were found as the source of bacteremia in more than half of the patients. Anti-pseudomonal treatment was initiated in 64 patients (91.4%). Two of these patients were transferred to another facility; therefore, only mortality data were available for them. Overall, 78.6% (n = 55) of the patients received in vitro active treatment, with comparable rates between the CHR-positive and CHR-negative groups. Colistin-based regimens were initiated in 13 patients, with a similar distribution between the two groups. The overall clinical response rate was 51.6% (n = 32), which was significantly higher in the CHR-positive group (p = 0.042). While 30-day mortality rates were comparable between the groups, chi-square analysis revealed a lower 90-day mortality rate in the CHR-positive group (p = 0.041). However, log-rank analysis demonstrated no significant difference in 30 and 90-day survival distributions between patients with and without CHR, respectively (p = 0.693 and P = 0.448, respectively). Microbiological response could not be evaluated in 40 (57.1%) patients due to a lack of follow-up cultures; among the remaining patients, microbiological eradication was confirmed in 27. Among the 30 patients with meropenem-resistant isolates, 17 (56.7%) were CHR-positive, and 10 (33.3%) had received a colistin-containing regimen. In this subgroup, the clinical response rate was 57.1% in CHR-positive cases compared to 33.3% in CHR-negative cases (p = 0.225). The 30-day and 90-day mortality rates were both 58.8% in the CHR-positive group, compared to 69.2% and 76.9% in the CHR-negative group, respectively. However, no statistically significant difference was observed in the 30-day and 90-day mortality rates between these two groups (p = 0.708 and p = 0.440, respectively). In the CHR-positive group, 81.4% of isolates had a colistin MIC of 1 mg/L, whereas this rate was 59.3% in the CHR-negative group. This difference was statistically significant (p = 0.043). Carbapenem resistance rates were similar between CHR-positive (39.5%) and CHR-negative (48.1%) groups (p = 0.478). Risk factors for CHR Table 3 summarizes the results of univariable and multivariable logistic regression analyses assessing clinical factors associated with CHR. In the final multivariable model, age ≥ 60 years (aOR: 9.222, 95% CI: 2.229–38.154, p = 0.002), lack of hospitalization in the past three months (aOR: 6.285, 95% CI: 1.480–26.690, p = 0.013), and a colistin MIC of 1 mg/L compared to 2 mg/L (aOR: 5.490, 95% CI: 1.165–25.877, p = 0.031) were independently associated with CHR. Colistin history in the past three months was also included in the multivariable model due to its confounding effect, although it did not show a statistically significant independent association with CHR (aOR: 1.929, 95% CI: 0.389–9.578, p = 0.422). Other evaluated variables, including prior antibiotic use, and anti-pseudomonal antibiotic history were not significantly associated with CHR. Table 3Clinical risk factors for CHR in a logistic regression modelVariableUnivariableMultivariableOR (95% CI)P valueaOR (95% CI)P valueAge (years)1.036 (1.006–1.067) 0.020 - - Age ≥ 604.433 (1.571–12.507) 0.005 9.222 (2.229–38.154) 0.002 No hospitalization history (past 3 months)2.731 (0.984–7.578)0.0546.285 (1.480–26.690) 0.013 Antibiotic history (past 3 months)0.421 (0.119–1.495)0.181--Anti-pseudomonal antibiotic history (past 3 months)0.566 (0.190–1.685)0.307--Colistin history (past 3 months)0.929 (0.264–3.273)0.9091.929 (0.389–9.578)0.422Colistin MIC (1, compared to 2 mg/L)3.008 (1.015–8.910) 0.047 5.490 (1.165–25.877) 0.031 CI, confidence interval; MIC, minimum inhibitory concentration (mg/L); OR, odds ratio; aOR, adjusted OR; bold texts indicate the statistically significant p-valuesOmnibus Test of Model Coefficients chi-square = 23.390, df = 4, P < 0.001; Hosmer–Lemeshow test P = 0.970; Nagelkerke R² = 0.429; all VIF values ranged between 1.010 and 1.022 Clinical risk factors for CHR in a logistic regression model CI, confidence interval; MIC, minimum inhibitory concentration (mg/L); OR, odds ratio; aOR, adjusted OR; bold texts indicate the statistically significant p-values Omnibus Test of Model Coefficients chi-square = 23.390, df = 4, P < 0.001; Hosmer–Lemeshow test P = 0.970; Nagelkerke R² = 0.429; all VIF values ranged between 1.010 and 1.022 Discussion Within this conceptual framework, our findings may also be interpreted in light of the emerging paradigm of antimicrobial heteroresistance. In our study, we detected a remarkably high prevalence of CHR (61.4%) among COL-susceptible P. aeruginosa bloodstream isolates using the PAP method, considerably higher than the results of previous studies conducted. Notably, the GT failed to identify any heteroresistant strains, highlighting the limitations of conventional susceptibility testing methods in detecting minority resistant subpopulations. Furthermore, since there are limited studies investigating the effect of HR on clinical outcomes, in the present study, we found that age ≥ 60, lack of hospitalization history, and a colistin MIC of 1 mg/L were risk factors for CHR. However, no significant association was observed between CHR and mortality. The remarkably high prevalence of CHR observed in our study warrants careful interpretation. Several methodological and epidemiological factors may have contributed to this finding, including differences in heteroresistance definitions and PAP interpretation criteria. Previous studies have applied different resistant-subpopulation frequency thresholds (e.g., 10⁻⁶ versus 10⁻⁷), different concentration cut-offs, and different PAP interpretation approaches. Consequently, direct comparison of CHR prevalence across studies should be made with caution, and the prevalence observed in our study may not be directly comparable to previously published reports. Although the prevalence observed in our study was higher than most previously reported rates, relatively high CHR frequencies have also been reported in studies using PAP-based methodologies [10]. This observation may suggest that at least part of the variability across studies reflects true epidemiological differences in addition to methodological heterogeneity. Nevertheless, the lack of an internationally standardized definition of colistin heteroresistance remains a major challenge for cross-study comparisons and highlights the need for consensus criteria for CHR detection. The available literature on colistin heteroresistance in P. aeruginosa is summarized in Table 4. Table 4Summaries of studies on colistin heteroresistance in P. aeruginosa isolatesCountryClinical sampleNumber of samplesMethod for determination of HRPrev. of CHR (%)Ref.BrasilVCS24PAP (colonies growing on polymyxin B > 2 mg/L) and with an increase of 8 x MIC of the main pop.4.2 [11]PAP (colonies growing on polymyxin B ≤ 2 mg/L) and with >2xMIC increase compared with the main pop.37.5HungaryBlood76Agar screen (colonies growing on colistin concent. of 4 mg/L)27 [12]ChinaVCS231PAP (colonies growing on colistin ≥ 2 mg/L)3.9 [13]SwedenVCS143PAP (colonies growing on 4 and 8 mg/L of colistin), at a frequency of ≥1 × 10− 66 [14]PAP (colonies growing on colistin concent. ≥8xMIC of the main pop.), at a frequency of ≥1 × 10− 726IranVCS100PAP (colonies growing on colistin concent. ≥8xMIC of the main pop.), at a frequency of ≥1 × 10− 78 [15]TurkeyBlood70PAP (colonies growing on colistin concent. ≥8xMIC of the main pop.), at a frequency of ≥1 × 10− 761.4The present studyGT (bacterial growth within the inhibition zone)0 Summaries of studies on colistin heteroresistance in P. aeruginosa isolates In addition to methodological variability, epidemiological characteristics of the study population and local antimicrobial selection pressures may also have influenced the observed CHR prevalence. Only 21% of patients had documented prior colistin exposure, suggesting that previous colistin therapy alone cannot explain the high prevalence of CHR observed in our study. The high prevalence of CHR may be influenced by factors beyond individual antibiotic exposure. Clinically achievable plasma colistin concentrations are limited by toxicity concerns and may not completely suppress resistant-subpopulations, potentially facilitating their persistence [5]. Furthermore, sustained colistin use in our hospital setting may have contributed to the maintenance and circulation of heteroresistant strains within the healthcare environment. These factors may partly explain the high frequency of CHR detected among bloodstream isolates in our study. Additionally, CHR has been reported more frequently among multidrug-resistant and carbapenem-resistant strains. Since our collection largely consisted of isolates recovered from a tertiary-care hospital with a high burden of resistant Gram-negative bacteria, the high CHR rate observed in our study may be associated with the predominance of resistant hospital lineages rather than prior colistin exposure alone [16]. The markedly lower CHR prevalence reported in the recent Iranian study may further illustrate the substantial variability observed across studies [15]. In addition to differences in PAP interpretation criteria, the two studies differed considerably in isolate selection. While the Iranian study included clinical isolates from various infection sites, our study was restricted to bloodstream isolates and predominantly included strains with colistin MICs of 1–2 mg/L, a factor independently associated with CHR in our cohort. Therefore, differences in study populations and isolate characteristics may have contributed to the discrepant prevalence estimates. Another possible explanation for the high prevalence observed in our study is the composition of the isolate collection. Only bloodstream isolates with colistin MICs of 1–2 mg/L were included, and isolates with a MIC of 1 mg/L constituted the majority of the study population. Notably, a colistin MIC of 1 mg/L was independently associated with CHR, suggesting that the MIC distribution within the study sample may have contributed to the observed prevalance estimate. Since isolates with a MIC of 1 mg/L were independently associated with CHR in our cohort, the distribution of MIC values within the study population may have contributed to the observed prevalence estimate. This finding is noteworthy because it challenges the conventional assumption that higher MIC values necessarily indicate a greater likelihood of resistance-related phenomena. Instead, it suggests that isolates categorized as fully susceptible may still contain resistant subpopulations detectable by PAP. While the proportion of isolates with a colistin MIC of 1 mg/L in the study by Howard-Anderson et al. (66%, n = 95) was similar to that observed in our cohort, those authors did not identify a significant association between MIC category and CHR, in contrast to our findings [14]. The biological basis of stable and unstable heteroresistance remains incompletely understood. Stable heteroresistance may suggest the presence of heritable alterations that allow resistant subpopulations to maintain their phenotype after serial passages in antibiotic-free media. In contrast, unstable heteroresistance may reflect reversible adaptive responses to antibiotic exposure, including transient phenotypic switching or persistence-like states [6, 7]. Such mechanisms could permit survival during antibiotic treatment while not being permanently maintained in the bacterial population. Because the present study did not investigate the molecular mechanisms underlying heteroresistance stability, these interpretations should be considered speculative and require confirmation by future genomic and transcriptomic studies. In the current study, the clinical response rate was statistically higher in the CHR-positive group than in the CHR-negative group (p = 0.042). It is crucial to highlight that the number of patients receiving colistin-based regimens in our cohort was highly limited (n = 13); therefore, this difference in clinical outcomes cannot be attributed to the choices or efficacy of colistin therapy. Importantly, this counterintuitive finding was not driven by baseline clinical disparities between the two groups. Major clinical determinants and confounding factors that directly influence prognosis—including the CCI, severe immunocompromised status, presence of sepsis at onset, meropenem resistance rates, the administration of in vitro active antibiotics, and colistin-based therapy—were highly comparable and showed no statistically significant differences between the CHR-positive and CHR-negative groups (Table 2). Given that these key clinical variables were well-balanced between the groups, the biological basis of the higher clinical response rate observed among CHR-positive patients remains unclear. One possible explanation may be the fitness cost associated with heteroresistance. However, this hypothesis was not directly evaluated in the present study and requires confirmation in future experimental investigations. Additionally, consistent with the findings of a previous study [14], no significant association was observed between CHR and overall mortality in our cohort. As with the clinical response rates, this lack of association cannot be attributed directly to the definitive clinical effect, or lack thereof, of the CHR phenotype, due to the small size of the subpopulation treated with colistin. Therefore, future prospective studies with larger cohorts of colistin-treated patients, complemented by molecular and genomic analyses, are warranted to fully elucidate the definitive clinical trajectory and underlying genetic mechanisms of colistin heteroresistance in bloodstream infections. Another notable finding was the association between the absence of recent hospitalization and the presence of CHR. While hospitalization is traditionally considered a risk factor for antimicrobial resistance, this observation may indicate that heteroresistance is not solely driven by nosocomial selective pressures. This observation should be interpreted cautiously and validated in future studies. Furthermore, increasing age, which emerged as another independent predictor of CHR in our multivariable model, is a well-known risk factor for resistant P. aeruginosa infections [17]. Even though the PAP method is regarded as the most reliable method for detecting HR, its high cost and labor-intensive nature make it difficult implement in clinical settings. Therefore, diffusion-based tests are widely utilized in clinical microbiology laboratories, particularly in countries with a high prevalence of HR. In the present study, no bacterial growth was observed within the inhibition zone of colistin GT strips for any of the isolates identified as heteroresistant by the PAP method. Standard susceptibility testing methods may overlook clinically significant minority variants, potentially leading to underestimation of resistance and suboptimal treatment decisions. This limitation underscores the need for diagnostic approaches capable of capturing population-level complexity. This suggests that GT testing is insufficient for detecting CHR, a finding consistent with previous studies [18, 19]. The limited diffusibility of colistin on agar surfaces may account for the failure to detect CHR, similar with the limitations of diffusion-based techniques in determining colistin susceptibility. The observation of distinct colony morphologies within heteroresistant subpopulations in one isolate further supports the concept of phenotypic diversification within bacterial populations. Although heteroresistance and small colony variant phenotypes are considered as distinct, our findings suggest that these two phenotypes may not be entirely separate concepts and could potentially overlap [20]. To our knowledge, no study in the literature presents data similar with our observation. Our study has several limitations. First, it was conducted retrospectively and included a relatively small number of isolates. The small sample size in our multivariable analysis resulted in wide confidence intervals for certain predictors; therefore, these identified risk factors for CHR should be interpreted with caution and validated in larger cohorts. Second, our analysis was based on culture-dependent methods and did not incorporate molecular or metagenomic approaches that could better characterize microbial population diversity. Therefore, the underlying mechanisms driving heteroresistance and its relationship with the broader bloodstream microbial environment remain to be elucidated. Third, PAP experiments were performed once for each isolate and formal reproducibility or inter-run variability analyses were not performed. Finally, the classification of stable and unstable heteroresistance was based on a seven-day serial passage protocol and should be interpreted in the absence of a standardized definition for heteroresistance stability. Conclusions In conclusion, we identified a remarkably high prevalence of colistin heteroresistance among colistin-susceptible P. aeruginosa bloodstream isolates. GT failed to detect any heteroresistant isolate, highlighting the limitations of conventional susceptibility testing methods for identifying resistant subpopulations. Although CHR was not associated with mortality in our cohort, its high prevalence and potential implications for antimicrobial therapy warrant further attention. Future studies incorporating larger patient cohorts and molecular investigations are needed to better define the clinical significance, epidemiology, and underlying mechanisms of CHR in P. aeruginosa bloodstream infections. Acknowledgements This study was presented at the ESCMID Global Congress 2023, Copenhagen, Denmark, 15-18 April 2023 as an oral poster (Abstract No.E0885). Author contributions Banu Sancak and Serhat Ünal designed the study. Ekin Kirbas performed the PAP and disk diffusion assays. M. Cihan Isik analysed the demographic and clinical features of the patients. Rashad Ismayilov performed the statistical analysis. Ekin Kirbas and M. Cihan Işık prepared the manuscript. Ekin Kirbas prepared Fig. 1, and Rashad Ismayilov prepared Fig. 2. All authors reviewed the manuscript. Funding Open access funding provided by the Scientific and Technological Research Council of Türkiye (TÜBİTAK). This study was carried out using the authors’ own resources, with no external financial support. The authors declare that no funds, grants, or other financial support were received during the preparation of this manuscript. Data availability Data available from the corresponding author on reasonable request. Declarations Ethics approval This study was conducted in compliance with the ethical principles outlined in the Declaration of Helsinki and received approval from the Health Sciences Research Ethics Committee of Hacettepe University (Date: 06 September, 2022, Decision No: 2022/13–19, GO 22/761). The requirement for informed consent was waived because of the retrospective methodology. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement Data available from the corresponding author on reasonable request.