Development and Clinical Evaluation of a CRISPR-Based Diagnostic for Rapid Group B Streptococcus Screening

Lingxiao Jiang; Weiqi Zeng; Wanting Wu; Yingying Deng; Fusheng He; Wenli Liang; Mingyao Huang; Hong Huang; Yongjun Li; Xiaorui Wang; Hang Su; Shilei Pan; Teng Xu


Emerging Infectious Diseases. 2021;27(9):2379-2388. 

In This Article


Development of CRISPR-GBS

To address the challenges in clinical GBS screening, we aimed to develop a rapid, highly sensitive, and simple-to-use GBS assay by combining an RPA reaction with a CRISPR/Cas13 step for target detection.[17] We established a rapid extraction method for high efficiency GBS DNA extraction by combining chemical, heat, and bead beating-based cell wall disruption, which eliminated the need for any column and organic solvents (Figure 1; Appendix Figure 1, This strategy takes advantage of both the polymerase-mediated DNA amplification and the CRISPR/Cas-mediated enzymatic signal amplification for greater sensitivity. Moreover, the rapid extraction and isothermal nature of such an assay eliminated the demand for sophisticated instruments such as thermal cyclers.

Figure 1.

Schematic diagram of CRISPR-based diagnostic for rapid GBS screening. Swab samples are first eluted and followed by a rapid DNA extraction step where the bacterial cell walls are disrupted by a combination of chemical, physical, and heating effects. The extracted DNA is then subjected to the CRISPR/Cas reaction. The collateral nuclease activity of Cas proteins are activated upon specific binding of gRNA to the atoB gene. Fluorescent signal produced from cleaved probes is captured and indicates the presence of GBS. GBS, group B Streptococcus. gRNA, guide RNA; ssRNA, single-stranded RNA.

We chose the thiolase (atoB) gene as the target region in this assay because it is highly conserved and specific for the GBS genome.[20] We screened multiple sets of RPA primers and CRISPR gRNAs targeting different regions within atoB (Appendix Table 2, Figure 2). The set that showed the best overall performance of sensitivity and specificity was then used in this study for assay optimization and clinical diagnostic evaluation.

Figure 2.

Analytical assessment of the sensitivity and specificity of CRISPR-based diagnostic for rapid GBS screening. Evaluation was performed by testing contrived negative swab samples with indicated CFUs of GBS (A), different copy numbers of GBS genomic DNA (B), and various microbes as interfering materials (C). GBS, group B Streptococcus. A. baumannii, Acinetobacter baumannii; E. coli, Escherichia coli; E. faecalis, Enterococcus faecalis; hDNA, human DNA; P.aeruginosa, Pseudomonas aeruginosa; S. aureus, Staphylococcus aureus; S. mitis, Streptococcus mitis; S. pneumoniae, Streptococcus pneumoniae; S. pyogenes, Streptococcus pyogenes.

We then sought to determine the analytical sensitivity by serial dilutions of GBS with negative swabs at various counts of CFU per mL. CRISPR-GBS managed to detect samples at 30 CFU/mL in 6 of 10 runs and at 60 CFU/mL in all 10 replicates (Figure 2, panel A). We further assessed the limit of detection of CRISPR-GBS by titrations of copies per reaction. The CRISPR assay consistently detected 5 copies of GBS in 10 of 10 runs and 2 copies in 4 of 10 replicates (Figure 2, panel B). These data indicate that CRISPR-GBS could detect a low number of genome copies or ≈50 CFU/mL and is more sensitive than most of the commercially available US Food and Drug Administration–approved GBS assays, such as GeneXpert GBS (300 CFU/mL) (Cepheid,, BD Max GBS (1,000 CFU/mL) (BD,, Quidel Solana GBS (2.6 × 105 CFU/mL) (Quidel,, and AmpliVue GBS (1.4 × 106 CFU/mL) (Quidel).[20–22]

With such a high sensitivity of CRISPR-GBS, we set out to confirm its specificity. For this purpose, we assayed DNA from humans and a panel of bacteria, including bacteria in the same genus (e.g., S. pneumonia, S. pyogenes, and S. mitis), microbes commonly found in vaginal swabs (e.g., E. coli, Staphylococcus aureus, and Enterococcus faecalis), and bacteria commonly found in nosocomial infections (e.g., Acinetobacter baumannii and Pseudomonas aeruginosa).[23] Of note, none of these interference samples triggered a false-positive reaction (Figure 2, panel C). Altogether, these analytical evaluations suggest that CRISPR-GBS, with its great sensitivity and specificity, is a promising molecular assay for GBS detection.

Clinical Diagnostic Evaluation of CRISPR-GBS

After the analytical study, we further assessed the diagnostic potential of CRISPR-GBS in settings of clinical screening. A total of 426 pregnant women with a median age of 29 years (20–47 years) were enrolled in this cohort study. Sample collection was performed at 34–38 weeks of gestation. Among these patients, 14 were excluded because of invalid test results, an insufficient specimen, or both. The remaining 412 patients were tested for GBS by culture, PCR, and CRISPR-GBS on their direct swab samples. We found no significant differences between patients who were negative or positive for GBS on the basis of patient age or weeks of gestation (Appendix Table 1).

When we conducted the CRISPR-GBS assay, we included a positive control of GBS DNA and a no-template control in parallel for each run. We used a fluorescent signal from no-template control normalize the signal of other samples in the same run to calculate the corresponding fold changes. We noticed clear distinctions in signal patterns of the reactions. Specifically, the fluorescent signal curve either remained flat (e.g., the no-template control runs) or had a distinguishable takeoff from the baseline (e.g., the positive control runs) (Figure 3, panel A). To determine the cutoff value as fold-changes for the CRISPR-GBS results, we first separated all the runs into a tentatively positive group and a tentatively negative group according to these distinct patterns. We then analyzed the cutoff values. The tentatively positives had fold changes ranging from 3.9 to 90.3 (median 26.3), whereas the tentatively negatives ranged from 0.5 to 2.9 (median 1.5) (Figure 3, panel B; Appendix Figure 3). Therefore, we were able to set the cutoff value at 3.5 for complete separation of the 2 groups. Consistently, this cutoff was further confirmed by the receiver operating characteristic analysis for optimal sensitivity and specificity (data not shown).

Figure 3.

Determination of assay cutoff for CRISPR-based diagnostic for rapid GBS screening. A) Representative signal curves produced by CRISPR-GBS. A positive control (red), a negative control (black), and 85 clinical samples (blue) are shown with distinct curve patterns (take-off vs. flat). B) Fold-change values by CRISPR-GBS obtained from our prospective cohort: positive (with take-off signal curves in red) and negative (flat curves in blue). A cutoff of 3.5 was set and is indicated in black dashed line. GBS, group B Streptococcus. Lines from the bottom to the top of box-and-whisker plots refer to minimum, first quartile (Q1), median, third quartile (Q3), and maximum number of the dataset.

To evaluate the diagnostic performances of different methodologies for GBS detection, we began by comparing direct culture and PCR. We found a concordance of 97.1% between these 2 traditional methods. Specifically, only 5 (1.2%) of 412 culture-positive and 7 (1.7%) of 412 PCR-positive cases were missed by the other test. When culture was used as the reference standard, PCR demonstrated a sensitivity of 90.9% (50/55 results) and specificity of 98.0% (350/357 cases).

We further assessed the CRISPR-GBS test in comparison with direct culture and the PCR-based assay (Table; Figure 4). When the comparison was made separately, CRISPR-GBS was able to detect most of the positive samples by either reference method, with a sensitivity of 94.5% (52/55 cases) compared with culture and 94.7% (54/57 cases) compared with PCR. When we included only the 400 cases with concordant culture and PCR results in the analysis, CRISPR identified 94.0% (47/50) of the positive results and offered a negative predictive value of 99.1% (320/323 cases).

Figure 4.

Overview and summary of the prospective cohort study assessing CRISPR-GBS. A) Study enrollment and result summary as categorized by agreements between different tests. B) Venn diagram demonstrating the overall concordance and discordance among direct culture, regular PCR, and CRISPR-GBS in the cohort. CRISPR-GBS, CRISPR-based diagnostic for rapid group B Streptococcus screening.

Among the cases reported negative by culture, PCR, or both, we also found ≈10% of them to be positive by CRISPR, which included 37 of 357 culture-negative cases, 35 of 355 PCR-negative cases, and 30 of 350 dual-negative cases (i.e., by culture and PCR). These data indicate a greater sensitivity or a lower specificity of CRISPR-GBS.

We designed and conducted additional validation studies in an attempt to validate the improved sensitivity of CRISPR-GBS. We developed a nested PCR–Sanger assay targeting the atoB gene, in which we performed 2 successive rounds of PCR in a nested manner to achieve greater amplification sensitivity compared with regular single-round PCR reactions. We then subjected the amplicons to Sanger sequencing for further validation. With this nested PCR assay, we tested the 30 specimens that were only positive by CRISPR-GBS but negative by both direct culture and regular PCR in our cohort. We were able to confirm 15 of 30 discordant cases (Figure 4, panel A). These data supported the previous findings and again indicate higher sensitivity of CRISPR-GBS compared with direct culture or PCR.

To further rule out the possibility of false-positive results, we set up a retrospective validation study and compared the sensitivity of CRISPR-GBS with enrichment culture, which had been shown to be more sensitive than direct culture.[5,24] The validation cohort of 31 patients consisted of 13 CRISPR-positive and direct culture–positive, 10 CRISPR-positive and direct culture–negative, and 8 CRISPR-negative and direct culture–negative samples. We tested each sample by direct culture, enriched culture, and CRISPR-GBS both before and after broth enrichment. We performed enriched culture by overnight culture in selective broth, followed by inoculation onto blood agar. We found that the samples that were negative by both direct culture and CRISPR originally would remain negative even after broth enrichment. However, of the 10 cases that were positive by CRISPR but negative by direct culture, adding the broth enrichment step yielded positive results in 90% of those cases (Figure 5). These results validated the greater sensitivity of CRISPR and suggested that the testing direct swabs by CRISPR-GBS conferred comparable sensitivity as enrichment culture. In our antepartum cohort of 412 pregnant women, the prevalence of GBS carriage was the highest by CRISPR at 21.6% (89/412) and was similar by culture (13.3% [55/412]), and PCR (13.8% [57/412]).

Figure 5.

Overview of the validation study with enrichment culture for CRISPR-based diagnostic for rapid group B Streptococcus screening. Testing results by culture and CRISPR before (left) and after (right) broth enrichment are shown.

When we compared turnaround time, we found that the CRISPR-GBS test required an average of <1.5 hours, which includes 30 minutes of rapid DNA extraction, 30 minutes for DNA amplification by RPA, and 20 minutes for Cas13 detection. This turnaround time is a considerable advantage over those for conventional culture-based (24–60 hours) and PCR-based (≈2.5 hours for a regular PCR assay and much longer for nested PCR–Sanger) methods.