Matrix-Assisted Laser Desorption Ionization Time-of-Flight Mass Spectrometry for Identification of Microorganisms in Clinical Urine Specimens after Two Pretreatments

Urinary tract infections (UTIs) are one of the most common infections (Klein and Hultgren 2020). Although mild UTIs do not cause severe organ damage, pathogenic bacteria can lead to adverse consequences such as pyelonephritis, kidney abscess formation, and acute kidney injury through the urethra, bladder, ureter, and other ways, causing septicemia and even death (Korbel et al. 2017; Hsu and Melzer 2018). Fast and reliable microbial identification is essential for the diagnosis and treatment of UTIs. The current diagnosis of UTIs relies on routine urine culture identification, a process that requires 48 hours or longer (de Cueto et al. 2017). Although 16S rRNA gene sequencing, multiplex PCR, and fluorescence in situ hybridization can quickly detect pathogens in urine cultures, only 20% to 30% of clinical urine samples show significant bacterial growth (Manickam et al. 2013). These expensive and cumbersome methods are not suitable for clinical practice (Akoachere et al. 2012).

Matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) is a new method for rapid microbial identification without the need for target amplification (Nomura 2015). It can replace methods such as 16S rRNA gene sequencing (Vincent et al. 2013). It has a great potential in identifying bacteria directly from urine samples and culture-positive blood samples (Sauget et al. 2017; Nomura et al. 2020). Nevertheless, this novel method has not been widely used in clinical practice. Therefore, in this study, we compared a centrifugation-based MALDI-TOF MS, a short-term culture with MALDI-TOF MS, and the conventional diagnostics of UTI in a clinical practice setting.

Among the 965 patients with suspected UTIs, 425 were male (44.0%), and 540 were female (56.0%). Because our institution is a teaching hospital and a children’s hospital, 374 patients (38.8%) were under 13 years of age, and 202 (20.9%) were under 1 year of age. The mean age was 38.5 (patients under 1 year were counted as 1 year old), the median age was 45 years (IQR 3–68 years). The sources of the samples were as follows: 814 inpatients (84.4%); 149 outpatients (15.5%); one unknown (0.1%).

Conventional urine culture and two MALDI-TOF MS methods. Among 965 urine specimens, 211 positive specimens were identified by conventional urine culture, including Gram-negative bacteria 170/211 (80.6%) and Gram-positive bacteria 41/211 (19.4%). The most common microorganisms were E. coli in 100 cases (44.2%), followed by Klebsiella pneumoniae in 28 cases (12.4%) and Enterococcus faecalis in 15 cases (6.6%) (Table I).

159/211 (75.4%) positive specimens were detected by the centrifugation-based MALDI-TOF MS. The corresponding rates with routine culture were Gram-negative bacteria: 140/170 (82.4%), Gram-positive bacteria: 19/41 (46.3%) (p = 0.053). The top three detected bacteria were E. coli (n = 84, detection rate: 84.0%), K. pneumoniae (n = 23, 82.1%), and Proteus mirabilis (n = 10, 100%). Short-term culture combined with MALDI-TOF MS detected 182/211 (86.3%) positive specimens. The corresponding rates with routine culture were Gram-negative bacteria: 155/170 (91.2%), and Gram-positive bacteria: 27/41 (65.9%) (p = 0.230). The top three detected bacteria were E. coli (n = 95, detection rate: 95.0%), K. pneumoniae (n = 26, 92.9%), Enterococcus faecium (n = 12, 85.7%) (Table I). There was no significant difference between the three methods (p = 0.135) (Table II).

Comparison between the two MALDI-TOF MS methods. The centrifugation-based MALDI-TOF MS identified 159 cases with a score ≥ 1.70, and 135/159 cases (84.9%) were identified with scores ≥ 2.00, including Gram-negative bacteria 121/140 (86.4%) and Gram-positive bacteria 14/19 (73.7%). The mean score was 2.17 ± 0.24. Short-term culture combined with MALDI-TOF MS identified 182 cases with scores ≥ 1.70, and 153/182 cases (84.1%) were identified by scores ≥ 2.00, with Gram-negative bacteria 137/155 (88.4%) and Gram-positive bacteria 16/27 (59.3%). The mean score was 2.09 ± 0.20, slightly lower than centrifugation-based MALDI-TOF MS. Statistical analysis showed that, although the mean MALDI score of the centrifugation-based MALDI-TOF MS was higher than that of short-term culture combined with MALDI-TOF MS in the identification of Gram-negative and Gram-positive bacteria, the difference was insignificant. Nevertheless, in terms of overall identification, the difference in identification score was statistically significant (p = 0.003, Table III).

To determine correlations between colony counts and the detection rates of MALDI-TOF MS, we compared the identification results in various colony ranges using two MALDI-TOF MS methods. The statistical results showed that there was no difference between the two MALDI-TOF MS methods for the identification of Gram-positive bacteria (p = 0.075). In contrast, when identifying Gram-negative bacteria, the difference was significant (p = 0.016). Further analysis showed that, when the colony count was more than 1 × 10⁵ CFU/ml, the centrifugation-based MALDI-TOF MS and short-term culture combined with MALDI-TOF MS detected 133/147 (90.5%) and 137/147 (93.2%) of Gram-negative bacteria, respectively; there were no significant differences (p = 0.394). When counts were between 1 × 10⁴ and 1 × 10⁵ CFU/ml, however, the detection rate of Gram-negative bacteria in the centrifugation-based MALDI-TOF MS was significantly lower than that in short-term culture combined with MALDI-TOF MS, which were 5/15 (33.3%) and 12/15 (80.0%), respectively; the difference was significant (p = 0.025). Moreover, when the count was less than 1 × 10⁴ CFU/ml, Gram-negative bacteria were detected in 2/8 (25.0%) and 6/8 (75.0%) by the two methods, respectively. The detection rate decreased significantly with decreased colony count, especially for the centrifugation-based MALDI-TOF MS (Table IV).

We found that, although the identification score of short-term culture combined with MALDI-TOF MS was slightly lower than that of the centrifugation-based MALDI-TOF MS, the detection ability of the short-term culture combined with MALDI-TOF MS was significantly higher than that of the centrifugation-based MALDI-TOF MS, especially when the colony count was low.

Rapid identification of urinary microorganisms and timely application of antibiotics can significantly reduce the length of hospital stay and costs (Sood et al. 2015). MALDI-TOF MS is a clinical bacterial identification method that can provide microbial identification results within 15 minutes; it is simple to operate and moderately priced (< 1 USD/sample). It is suitable for the microbiological identification of urine specimens from patients with UTIs (Dierig et al. 2015; Sauget et al. 2017).

The primary pathogens of UTIs are Gram-negative bacteria, followed by Gram-positive bacteria, and fungi are in the minority (Flores-Mireles et al. 2015). The detection rates of Gram-negative and Gram-positive bacteria by the centrifugation-based MALDI-TOF MS were 82.3% and 46.3%, respectively. In contrast, the detection rates of Gram-negative and Gram-positive bacteria by short-term culture combined with MALDI-TOF MS were higher, reaching 91.2% and 65.9%. Although there was no statistically significant difference in the overall identification rate between the two MALDI-TOF MS methods (p = 0.437), there was a significant difference in the detection results of Gram-negative bacteria when the colony count was between 1 × 10⁴ and 1 × 10⁵ CFU/ml (p = 0.027). The reason may be that the decrease in urine colony counts had a significantly more significant impact on the centrifugation-based MALDI-TOF MS than on short-term culture combined with MALDI-TOF MS. When the colony count was higher than 1 × 10⁵ CFU/ml, the bacterial density could meet the requirements for direct identification by MALDI-TOF MS after centrifugation and washing. However, when the bacterial colony number decreased to 1×10⁴ and 1×10⁵ CFU/ml, the detection rate of Gram-negative bacteria by the centrifugation-based MALDI-TOF MS dropped significantly from 90.5% to 33.3%, compared with that by short-term culture combined with MALDI-TOF MS, which only decreased from 93.2% to 80.0%, the difference was statistically significant (p = 0.027). When the colony number was lower than 1 × 10⁴ CFU/ml, the number of positive samples was too low, and the statistical results were not reliable. Among all Gram-negative bacteria, E. coli was the most common: 81/87 cases (93.1%) were detected by the centrifugation-based MALDI-TOF MS when the colony count was ≥ 1 × 10⁵ CFU/ml, but when the colony count was between 1 × 10⁴ and 1 × 10⁵ CFU/ml, the detection rate decreased sharply to 2/11 (18.2%). Although the detection rate of E. coli by short-term culture combined with MALDI-TOF MS was also affected by the decrease in the bacterial count, it was much less affected than the centrifugation-based MALDI-TOF MS, with the detection rate reduced from 96.6% to 81.8%. Therefore, it is necessary to pay attention to the possible false-negative specimens in clinical practice.

Although the detection rate of bacteria by short-term culture combined with MALDI-TOF MS was higher than that by centrifugation-based MALDI-TOF MS, its MALDI score was slightly lower. The reason may be that even after 5 hours of short-term culture, the colony number was still low, and it was easy to scrape the medium agar when applied to the target plate, which leads to too few bacteria scraped or failure to scrape bacteria, ultimately affecting the identification results. In the experiment, we could find that the MALDI spectrogram of the centrifugation-based MALDI-TOF MS was cleaner than that of short-term culture combined with MALDI-TOF MS. Nevertheless, in the experiment, we found another interesting phenomenon: when too many colonies were collected, the MALDI score could not be improved, and it is directly affected the identification results. It is probably because the bacteria were so clustered that it was not easy to spread evenly on a coated target board. There was a specimen with apparent bacterial growth, but MALDI-TOF MS did not detect it. Under the microscope, we found the bacteria gathered together. After a second test, Proteus mirabilis was identified. A similar phenomenon had been observed in other Gram-negative bacteria, such as K. pneumoniae. Therefore, in the clinical work, the appropriate bacterial density and the uniform coating on the target plate could guarantee their effective identification by MALDI-TOF MS.

Short-term culture combined with MALDI-TOF MS can eliminate the interference of proteins in MALDI-TOF MS. After short-term culture, it had higher detection ability for specimens with fewer colonies. The centrifugation-based MALDI-TOF MS could directly identify pathogens in clinical urine specimens within 1 hour and showed good identification ability on urine samples when colony counts was more than 1×10⁵ CFU/ml. However, it was affected by colony counts. To reduce the influence of small bacterial numbers on the detection results and improve the identification efficiency, we will introduce urine flow cytometry in subsequent experiments. As a new urine tangible component analysis technology, it can accurately provide the bacterial counts in urine samples (Wang et al. 2013; Sun et al. 2020). We plan first to count the number of urine bacteria using the urine flow cytometer. Urine samples with colony count higher than 1 × 10⁵ CFU/ml will be identified using the centrifugation-based MALDI-TOF MS, and if less than 1 × 10⁵ CFU/ml by short-term culture combined with MALDI-TOF MS.

Although the two MALDI-TOF MS methods have many advantages in clinical urine bacteria identification, there are some limitations. First, during the identification of Gram-positive bacteria by short-term culture combined with MALDI-TOF MS, the bacterial growth was not significant even after 5 hours of culture. Second, due to their thick and highly anionic cell walls, even the formic acid was added in the process, the detection of Gram-positive bacteria by MALDI-TOF MS was not optimal. It is necessary to improve the identification of Gram-positive bacteria. We tried to mix the specimens with formic acid and centrifuged at high speed; however, this did not work. Third, the study did not include antibiotic sensitivity tests and could not distinguish among antibiotic-resistant strains. Oviaño successfully used MALDI-TOF MS to rapidly identify the carbapenemase-producing Enterobacteriaceae in urine samples and detect their drug resistance (Oviaño et al. 2017). We will try to co-culture urine sediments and drugs on MALDI steel plates for short periods and rapidly detect the antimicrobial resistance of bacteria based on the mass spectrum peak change.

MALDI-TOF MS can accelerate the bacteriological identification of organisms causing UTIs. The centrifugation-based MALDI-TOF MS warrants fast identification. However, it is greatly affected by the number of bacterial colonies, while short-term culture combined with MALDI-TOF MS has a higher detection rate but a relatively slow identification speed. The screening the urine colony count first, then assigning to the two MALDI-TOF MS methods for identification, may be an effective and reliable alternative to the traditional urine culture, and it has great potential in measuring antibiotic susceptibility (Bizzini et al. 2011; Croxatto et al. 2012; Oviaño et al. 2017).