Rapid Identification of Pathogens in Positive Blood Culture of Patients with Sepsis: Review and Meta-Analysis of the Performance of the Sepsityper Kit
Table 1
The 21 published reports on the Sepsityper sample preparation kit. BC: blood culture, ID: identification, cfu: colony forming units, LOD: lower limit of detection, and MALDI-TOF MS: Matrix-Assisted Laser Desorption Ionization Time-of-Flight Mass Spectrometry.
ID up to 48 hours earlier compared to conventional methods In 18 of 27 mixed cultures at least one isolate was identified correctly Misidentified were 5 Streptococcus mitis as Streptococcus pneumonia
All genus results would be reported as species results by present day software Misidentified were 4 Streptococcus mitis as Streptococcus pneumonia, corrected in later databank updates
ID 24–130 h faster as compared to conventional methods Log(score) >1.7 was sufficient for ID on species level, since overall concordance to routine ID was 96.6% to genus and 94.1% to species level Five discrepant results: three Streptococcus oralis had Streptococcus pneumonia ID by MALDI; two were misidentified by conventional methods (16S sequencing confirmed MALDI)
Bactec 9240 Bactec Plus Aerobic/F, Anaerobic/F, and Peds Plus/F
52 99% species
71 64% species
—
123 82.9% species
>1.7 species (if matching organism stated several times in list)
Sepsityper performed better than alternative in-house method with gel separator tubes ID for Gram negative better than for Gram positive Misidentification of S. hominis instead of S. epidermidis and S. hominis instead of S. haemolyticus
Sepsityper performed better and was faster than two alternative sample preparation methods. Anaerobe BacT/Alert SN bottles lead to unreliable results, charcoal containing bottles not suitable
>1.4 genus >1.6 species (if 0.3 difference to next in list)
Relatively low number of BC tested. Gram positive bacteria resulted in better ID than Gram negative (unusual report, not confirmed by other studies). In direct comparison an in-house method performed better than Sepsityper. Log(score) cutoffs could be lowered to >1.6 for correct species ID, which increased the correct ID for both methods
Correct species ID accepted at any log(score), if first three database matches were identical No difference between Sepsityper and in-house method using saponin lysis ID of Salmonella paratyphi B only to genus level Direct ID faster than conventional method (1–24 h)
Bactec 9240 Bactec Plus Aerobic/F Anaerobic/F Peds Plus/F
35 94% genus 83% species
64 80% genus 70% species
—
99 85% genus 75% species
>1.5 genus >1.7 species
Sepsityper and two in-house methods compared One Streptococcus viridans was identified as Streptococcus pneumoniae Log(score)s <1.5 in combination with Gram stain results would lead to another 9 correct ID
Bactec Plus Aerobic, Bactec F Lytic Anaerobic, and Bactec Myco Lytic
106 99% genus 88.7% species
75 96% genus 72% species
—
181 97.8% genus 81.8% species
>1.6 genus >2.0 species
Sepsityper used with Biotyper generated significantly more accurate identifications than with Vitek MS Only genus level ID in both systems for Salmonella spec. In 21 mixed cultures, correct double ID by Biotyper in 5
BacT/Alert 3D FA, BacT/Alert 3D PF, and BacT/Alert SN
123 90% species
168 64% species
6 0%
297 74% species
>1.7 species
Most ID failures resulted from BC bottles with charcoal; therefore BacT/Alert bottles with charcoal are not the optimal culture medium in combination with Sepsityper
No data on genus identification shown. Definition of positive ID often independent of log(score). The term “misidentified” instead of “unidentified” was used for ID scores <1.5, 10 species not identified by routine method but MALDI
The fungi experiments contained mainly spiked BC samples A lower cutoff (>1.5 for species ID) was also applied. Combined with Gram staining, a log(score) >1.3 increased Gram positive results to 85% without misidentifications
Some Gram positive specimens were excluded from species analysis Total number of samples contained also 7 cerebrospinal fluids spiked into BC bottles and 10 polymicrobial samples
100–200 cfu of yeast was spiked into BC bottles and after positive growth extracted by Sepsityper A lower log(score) (>1.5) for correct species ID and repeated measurements increased ID on species level to 84% Gram staining not sufficient for fungal ID!
Bactec Plus Aerobic/F, Lytic/10, Anaerobic/F, and Peds Plus/F
64 98.4% species
—
—
64 98.4% species
>2.0 species
Main focus of the study was the use of the Sepsityper kit for direct antibiotic resistance testing in the BD Phoenix system with 97.9% concordance in susceptibility testing
Additional washing step added in protocol Anaerobic BC cultures excluded by design Also at least one correct ID in 77% of polymicrobial cultures Misidentification of 4 Streptococcus viridans specimens as S. pneumoniae
Bactec 9240, Plus Aerobic/F, Plus Anaerobic/F, Mycosis-IC/F, and Peds Plus/F
—
—
24 62.5% species
24 62.5% species
>1.5 species If first two results were identical
Sepsityper protocol was followed without prior washing steps (compare to Yan et al. [41]). Manual shooting and double protein concentration for MALDI were tried out to increase log(score)s >1.7, Mean yeast for positive ID was 4.5 × 107. Results were 23.5 h earlier than standard method Sepsityper pellets generated antibiotic susceptibility profiles in 72.7% of samples 15 h earlier