Cancer-related pneumonia is a severe disease with a poor outcome. The course of the disease is usually severe and characterized by high mortality, reaching 35.0% (ICNARC 2017). Many risk factors associated with cancer predispose patients to develop pneumonia. The most common is the disruption of anatomical surfaces, swallowing function, obstruction caused by tumor progression, neutropenia, lymphopenia, and impaired cellular and humoral immunity (Rolston 2001; 2017). In addition, frequent hospitalizations, typical for oncologic patients, increase the exposure of patients to nosocomial pathogens (Wong and Ewans 2017). Diagnosis of pneumonia based on clinical symptoms, which may be poorly expressed in oncologic patients, may be difficult (Wong and Ewans 2017). Therefore, well-documented risk factors for mortality in pneumonia, such as an unidentified etiology (Azoulay et al. 2017), a delay in antibiotic administration in patients with severe disease (Nauclér et al. 2021), and the inappropriate antimicrobial empiric therapy (Fernandez-Cruz et al. 2020) are of particular importance among these patients. Pneumonia can be caused by many pathogens, making empiric antibiotic therapy difficult (Leoni et al. 2016; Conway 2018). The type of pathogen depends on the type and stage of the cancer and the type of immunological deficits (Rolston 2001). Typical bacteria, such as Gram-negative bacilli, are a frequent pathogen of these infections (Ren et al. 2021). Infections caused by Gram-positive bacteria (except
The aim of this study was to compare the results of the conventional quantitative culture method (CM) in determining the bacterial etiology of pneumonia and detecting the underlying mechanisms of microbial resistance with the results of the Pneumonia
Seventy-nine specimens of bronchoalveolar lavage (BAL) collected from patients in whom pneumonia was suspected: outpatients (n = 25) and patients hospitalized in the intensive care unit (n = 25), and other hospital wards (n = 29) of the Prof. F. Lukaszczyk Oncology Centre in Bydgoszcz were collected in the period from June 2020 to February 2023. The samples were tested in parallel using the CM method and the PNP test. The results of the CM method were analyzed independently of the results obtained from the panels. Diagnostic thresholds for the CM method were adopted following the applicable recommendations, 104–105 CFU/ml for BAL (Hryniewicz et al. 2016). In addition, at the request of the clinician, 69 out of the 79 samples constituted material for testing for yeast-like fungi and mold fungi.
The principle of the method used is a quantitative PCR with a melting curve analysis based on the intercalation properties of the LCGreen®Plus (BioFire Defense, USA) dye into double-stranded and heteroduplex DNA. The PNP test simultaneously detects 26 microorganisms (18 bacteria and eight viruses) and seven antimicrobial resistance genes. The presence of the 15 typical bacteria is reported semi-quantitatively with intervals of 10/4, 10/5, 10/6, or ≥ 10/7 genomic copies per ml (copies/ ml) of the sample. The presence of atypical bacteria, viruses, and antibiotic-resistance genes was determined qualitatively. The microorganisms and resistance genes detected by the PNP test are shown in the Table SI.
The test material after prior processing was inoculated quantitatively using calibrated loops with a volume of 0.01 ml (10–2) and 0.001 ml (10–3) on Columbia Agar with Sheep Blood PLUS (Thermo Scientific™, Oxoid™ Deutschland GmbH, Germany), and Haemophilus Selective Agar (Thermo Scientific™, Oxoid Deutschland GmbH, Germany), and reductively on MacConkey Agar No. 3 (Thermo Scientific™, Oxoid™ Deutschland GmbH, Germany), and Cetrimide Agar (Thermo Scientific™, Oxoid™ Deutschland GmbH, Germany). In addition, a direct Gram stain was made. The plates were incubated, respectively, in an atmosphere of 5% CO2 and aerobic conditions, maintaining the temperature at 35 ± 2°C.
After an incubation period of 16–24 hours (or incubation extended to 48 hours, in the case of a negative result after the first day of incubation), identification was carried out using mass spectrometry (matrix-assisted laser desorption/ionization-time of flight mass spectrometry) (IVD MALDI Biotyper Smart System, microflex® LT/SH smart, Bruker Daltonics GmbH & Co., Germany). At the request of the clinician, quantitative inoculation for fungi was also performed on Sabouraud Glucose Selective Agar with Gentamicin and Chloramphenicol (Thermo Scientific™, Oxoid™ Deutschland GmbH, Germany), and on Brilliance™ Candida Agar (Thermo Scientific™, Oxoid™ Deutschland GmbH, Germany). Incubation was carried out, respectively: up to 7 days, on the first day at a temperature of 35 ± 2°C, on days 2–7 at a temperature of 25 ± 2°C, and up to 3 days at a temperature of 30 ± 2°C. Colonies that grew were identified by mass spectrometry. The results obtained were presented as a number of CFU/ml (CFU – colony-forming units).
Antibiotic susceptibility was assessed using the Kirby-Bauer disc diffusion method on Mueller-Hinton E agar and Mueller-Hinton agar with 5% horse blood + 20 mg/l β-NAD (bioMérieux, France), as well as the automatic nephelometric method (BD Phoenix (BD, Sparks, MD), or additionally using antibiotic strips in a concentration gradient – ETEST® (bioMerieux, France), following the recommendations of European Committee on Antimicrobial Susceptibility Testing (EUCAST 2023a; 2023b).
Consistency in the interpretation of the results of the detection of bacterial etiology of pneumonia by the PNP test (reference method) and the CM method was assessed. If only viral pathogens were detected by the PNP assay, the samples were categorized as “bacteriologically negative” for the analysis, and failure to culture bacterial pathogens by the culture method was considered consistent with the PNP result. Results were considered positive in the PNP assay when at least one target microorganism was detected (at ≥ 104 copies/ml for semi-quantitative bacteria). In the comparative analysis of the results of the PNP test with the results of the CM in determining the etiology of pneumonia and detecting the basic mechanisms of microbial resistance, use was made of the parameters assessed in the verification and validation processes of the research method and laboratory tests: accuracy (%), sensitivity (%), specificity (%), positive predictive value (PPV) (%) and negative predictive value (NPV) (%) (PDA 2000; Clark et al. 2009; Murphy et al. 2015). The numbers of True Positive and True Negative results (TP; TN), and the numbers of False Positive and False Negative results (FP; FN) were used to determine the indicated parameters of the PNP test.
Statistical analysis was performed using Statistica software version 13.3 (TIBCO® Software Inc., USA). Specificity, Sensitivity, Accuracy, PPV, and NPV were evaluated using the supplement “Zestaw Plus” ver. 5.0.80 (StatSoft Polska Sp. z o.o., Poland). Relationships between diagnostic methods were assessed using R-Spearman correlation analysis. In all statistical analyses, the cut-off value for the probability ratio was set at
In the case of 32 out of 79 samples (40.5%) of the BAL, the result of the PNP test was “positive”, including three samples with both bacterial and viral pathogens detected. In five out of the 32 samples (15.6%), only viral pathogens were detected by the PNP test, meaning that these materials were negative for bacterial agents. Thus, 27 bacteriological positive and 51 negative results were considered in the reference method. Bacteriological results specified by the pathogen, including 27 positive BAL according to the criteria, are presented in Table I. The PNP test detected 42 bacteria vs. 36 bacteria isolated in the CM method, although only 21 out of 36 bacteria (58.3%) were cultured equal to or above the diagnostic threshold. The most frequently detected bacteria in both methods, if the quantitative criterion for individual bacteria in the CM method was excluded from the analysis, was
Comparison of bacteria identified by the BioFire® FilmArray® Pneumonia
Bacterium | Conventional quantitative culture methods (CM) | BioFire® FilmArray® Pneumonia |
---|---|---|
Total number of bacteria detected* | Total number of bacteria detected | |
1 (0) | 2 | |
7 (4) | 8 | |
4 (3) | 4 | |
1 (1) | 1 | |
3 (2) | 3 | |
0 (0) | 1 | |
3 (1) | 3 | |
5 (3) | 5 | |
0 (0) | 1 | |
9 (4) | 10 | |
1 (1) | 1 | |
2 (2) | 3 | |
Total number | 36 (21) | 42 |
Resistance genes | ||
CTX-M | 1# | 1 |
IMP | 0 | 0 |
KPC | 0 | 0 |
NDM | 0 | 0 |
VIM | 0 | 0 |
OXA-48-like | 0 | 0 |
0 | 0 | |
Total number | 1 | 1 |
Seventy-nine samples of BAL (bronchoalveolar lavage) were analyzed by CM and PNP. Twenty-seven samples were bacteriologically positive in the PNP test (reference method) at the detection level at > 104 genomic copies/ml.
bacteria identified in CM, regardless of the result of the diagnostic titer; in the brackets number of cultures equal to or above the diagnostic threshold (> 104–105 CFU/ml (CFU – colony-forming units))
microorganisms not detected in the culture method
microorganism not detected in the PNP test
ESBLs positive (extended-spectrum p-lactamases)
A false positive culture result was obtained in one sample containing
In the phenotypic method, resistance to carbapenems was detected in one of the four
No atypical bacteria were detected in the tested samples in the analyzed period. Nine samples were positive for viruses, including five samples in monoculture. Among the detected viruses were Respiratory Syncytial Virus (RSV) (n = 3), influenza A virus (n = 2), parainfluenza virus (n = 1), adenovirus (n = 1), and human rhinovirus/ enterovirus (n = 4).
The overall agreement of the interpretation of the result between methods (positive or negative) was 84.8% (n = 67/79). The culture method showed a concordance of 16/27 positive results (TP) according to the criteria and 51/51 negative results (TN), which accounted for 59.3% and 100.0%, respectively, in the given result groups. Of the TN, 19 materials (n = 19/51; 37.3%) grew one to three potentially pathogenic bacterial species, but the growth titer (CFU/ml) was lower than the established diagnostic threshold. In the remaining clinical samples (n = 12/79; 15.2%), the CM method yielded incorrect results: 1/12 FP (8.3%) and 11/12 FN (91.7%). Among the materials for which a false-negative result was obtained, 6/11 (54.5%) of the materials were negative. In comparison, in 5/11 (45.5%) samples the pathogen was cultured at a titer below the cut-off value. Summarizing the individual categories of results concerning the total number of BAL tested (n = 79), the data are as follows: TP – 16/79 (20.3%); TN – 51/79 (64.6%); FP – 1/79 (1.3%) and FN – 11/79 (13.9%). Based on the obtained results, the sensitivity, specificity, PPV and NPV were calculated. The cumulative results are presented in Table II. Moreover, in the case of 28/69 (40,6%) samples of clinical materials sent by clinicians for mycological tests, fungi growth was observed. However, only in 1/28 (3.6%) samples the growth titer was higher than the established diagnostic threshold (Table III). The Spearman correlation coefficient, determining the correlation between diagnostic methods, was R = 0.7466, which confirms a strong relationship between the results of the compared diagnostic methods. This relationship was statistically significant (
Comparison of conventional quantitative culture method (CM) and the BioFire® FilmArray® Pneumonia
Methods | Samples | TP | FP | FN | TN | SENS | SP | ACC | PPV | NPV | FPR | FNR | AUC | SE | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
PNP vs. comb | BAL | 27 | 0 | 1 | 51 | 0.964 | 1.000 | 0.987 | 1.000 | 0.981 | 0 | 0.036 | 0.982 | 0.02 | |
CM vs. comb | BAL | 17 | 0 | 11 | 51 | 0.607 | 1.000 | 0.861 | 1.000 | 0.823 | 0.000 | 0.393 | 0.804 | 0.06 | |
CM vs. PNP | BAL | 16 | 1 | 11 | 51 | 0.593 | 0.981 | 0.848 | 0.941 | 0.823 | 0.019 | 0.407 | 0.787 | 0.062 |
BAL – bronchoalveolar lavage, Comb = CM + PNP, TP – true positives, FP – false positives, FN – false negatives, TN – true negatives,
SENS – sensitivity, SP – specificity, ACC – accuracy, PPV – positive predictive value, NPV – negative predictive value, FPR – false positive ratio,
FNR – false negative ratio, AUC – area under the curve, SE – standard error of AUC,
The Spearman correlation coefficient R = 0.7466,
Performance of quantitative growth of fungi in conventional quantitative culture method (CM).
Name of fungi | Quantity (CFU/ml) | |||
---|---|---|---|---|
102-103 | 103-104 | 104-105 | > 105 | |
4 | 3 | 0 | 0 | |
5 | 8 | 0 | 0 | |
6 | 5 | 1 | 0 | |
2 | 2 | 0 | 0 | |
3 | 0 | 0 | 0 | |
Kluyveromyces marxianus (formerly |
1 | 0 | 0 | 0 |
Pichia kudriavzevii (formerly |
0 | 1 | 0 | 0 |
1 | 0 | 0 | 0 | |
1 | 0 | 0 | 0 | |
Total number | 23 | 19 | 1 | 0 |
Sixty-nine samples were analyzed for fungi by CM.
The results are presented as CFU/ml (CFU – colony-forming units).
The diagnostic threshold was ≥ 104 – 105 CFU/ml. Fungi growth was obtained in 28 out of 69 samples (40.6%), but only in one sample was the growth of
In the current work, we analyzed the results of studies of bacterial pathogens in samples of BAL among patients with solid tumors using the rapid PNP test and CM. Only 20.3% of samples were TP for bacterial etiology in CM compared to 34.2% TP in the PNP test. More than 60.0% of the analyzed samples were TN in both methods. The most prevalent pathogens in both methods were
These results correlate with the results described in other works, in which the authors emphasize the low sensitivity of the CM method compared to the high sensitivity of the PNP test (Kamel et al. 2022; Jitmuang et al. 2022). Ginochio et al. (2021) showed that the sensitivity of the PNP test in detecting the etiology of hospital-acquired and ventilator-associated pneumonia is over 98.0%. The advantage of the PNP test over the CM method is the detection of the presence of microbial genetic material. Since the method is not limited by the pathogen activity, detecting the DNA of damaged or non-viable pathogens can be a disadvantage if it generates FP results. It is also an advantage if the material is collected during antibiotic therapy, which is common among hospitalized patients (Buchan et al. 2020). Some bacteria are more difficult to detect using routine cultures (Ginocchio et al. 2021). In our study, differences in non-growth bacteria were found in
The simultaneous use of the CM method and the PNP test can increase the sensitivity of the diagnostics. Edin et al. (2020), combining the PNP test with CM, determined the etiology in 73.0% of the examined patients, which helped to make quick therapeutic decisions. With the PNP test, we also identified many more positive samples. Using the PNP test allowed us to determine the bacterial etiology in 11 (13.9%) out of the 79 samples tested.
The ability of the PNP test to identify genes responsible for important resistance mechanisms is helpful, especially in the case of the high prevalence of resistance detected by the PNP test in a given area. In the samples tested in the current study, the gene responsible for the production of CTX-M-type ESBLs cephalosporinases in
Viruses are the second most common group of pathogens in lower respiratory tract infections. With respiratory viral infections, mainly of community origin, caused by influenza, parainfluenza, adenoviruses, rhinoviruses/enteroviruses, and RSV, their detections by the PNP test were very useful in the choice of oncologic patient treatment, especially during the infectious season. Using the PNP test allowed the detection of bacterial and viral co-infection in 3.8% of the tested samples. Also, the ability to detect infections caused by atypical bacteria, including
The PNP test does not detect infections caused by
Non-infectious pneumonia caused by neoplastic changes, post-radiation complications, pulmonary edema, and drug-induced pneumonia may have a similar course to infectious pneumonia (Black 2016). Since the PNP test is characterized by a high NPV, reaching 90–100%, which is also confirmed by our research, a negative PNP test result may likely indicate a possible cause of pneumonia other than an infectious one. It should be noted, however, that a negative result in the PNP test does not confirm the existence of a non-infectious cause of pneumonia. However, it can be a good tool in the hands of a clinician and complement the diagnostic process and thus influence the direction of the treatment (Azoulay et al. 2020).
The PNP test allows for quick identification of pneumonia pathogens among oncologic patients. It can be used to assess bacterial etiology and to diagnose especially community-acquired viral infections. Due to the various levels of immunosuppression and different stages of cancer in oncologic patients, which are associated with the presence of a specific profile of microorganisms responsible for infections, the benefits of molecular diagnostics should be considered after validations in specific groups of patients with cancer. Similarly, in other works, the authors emphasize that rapid molecular tests should be an additional tool, which, at the present stage of clinical effectiveness, assessment should be supplemented with the standard microbiological tests (Alnimr 2023). Due to the difficulty in predicting etiology, diagnosis of opportunistic pneumonia requires extensive diagnostics, going beyond the microorganisms covered by the panel, appropriate for the type of immunodeficiency, clinical course, and additional tests performed. Therefore, until studies are conducted in groups of oncologic patients who may receive particular benefits from performing molecular tests, tests should be performed in parallel.
This study presents some limitations. The limitation is the lack of clinical characteristics of the patients. Due to the large diversity in the group of patients who were tested (outpatients and hospitalized patients), and the relatively small number of tests performed and the possible impact of local epidemiology, the results cannot be generalized.
In conclusion, the PNP test is a good tool for determining the etiology of bacterial pneumonia in oncologic patients with solid tumors. It provides important diagnostic information, which may support the care of an oncologic patient. However, further large-sample studies are needed to research in strictly defined groups of oncologic patients.