DeAndre Derrick

Mr. DeAndre Derrick


Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
21 November 2025 in Diagnostics
Verified
Journal Article
Open access
PDF file
Moustafa KardjadjMoustafa Kardjadj
Moustafa Kardjadj
IPItoe P Priestly
Itoe P Priestly
RCRoel Chavez
Roel Chavez
DDDeAndre Derrick
DeAndre Derrick
THThomas K. Huard
Thomas K. Huard
Background: Rapid molecular detection of antimicrobial resistance (AMR) can shorten time to effective therapy in complicated urinary tract infections (cUTI), but the ability of gene presence and quantitative PCR signal (Ct, and ΔCt = Ct_marker − IC_Ct) to predict phenotypic non-susceptibility and clinical outcomes requires rigorous evaluation. We analyzed marker-level concordance, Ct→MIC relationships, and the clinical impact pathway in the randomized NCT06996301 trial. Methods: Marker–phenotype concordance metrics (sensitivity, specificity, PPV, NPV, LR+, LR−, κ) were computed for selected marker × species strata with stable sample sizes. Mixed-effects models (log2[MIC] ~ ΔCt_marker + IC_Ct + collection_method + prior_abx + (1|site)) assessed quantitative Ct→MIC associations. ROC analyses evaluated ΔCt discrimination of phenotypic non-susceptibility. A pre-specified sensitivity analysis included smaller strata (n ≤ 20) with bootstrap 95% confidence intervals for ΔCt slopes and AUCs. Clinical analyses compared PCR-guided (n = 193) versus culture-guided (n = 169) arms for time-to-antibiotic and treatment success using adjusted logistic regression and causal mediation (time-to-antibiotic as mediator; bootstrap inference). Results: High genotype–phenotype concordance was observed for canonical markers (e.g., blaCTX-M in E. coli: sensitivity 0.94 [95% CI 0.88–0.97], specificity 0.995 [95% CI 0.990–0.998], κ ≈ 0.93). Mixed models showed modest but significant Ct→MIC associations for select markers (e.g., blaCTX-M in E. coli: ΔCt slope −0.15 [95% CI −0.27 to −0.02], p = 0.015). The sensitivity analysis (n ≤ 20 strata) confirmed consistent negative directions, with robust bootstrap CIs excluding zero for qnrS (E. coli), tetM (E. coli), blaNDM (Klebsiella), and qnrS (Proteus). ROC AUCs for ΔCt prediction of non-susceptibility ranged from 0.62 to 0.81 (95% CIs ≈ 0.47–0.97). Clinically, PCR guidance shortened median time to antibiotic initiation (20 h vs. 52 h) and increased treatment success (88.1% vs. 78.1%; adjusted OR 1.95 [95% CI 1.12–3.40], p = 0.018). Mediation analysis estimated that 63% (ACME 0.112 [95% CI 0.045–0.178], p = 0.002) of the treatment success benefit was mediated through earlier antibiotic initiation. Conclusions: Binary detection of high-impact AMR genes by multiplex PCR reliably predicts phenotypic non-susceptibility and accelerates effective therapy when integrated with stewardship workflows. Quantitative PCR (ΔCt) provides modest but reproducible information about MIC magnitude and may flag heteroresistant subpopulations. A pragmatic implementation model combining rapid PCR with conventional culture is recommended to optimize clinical benefit while retaining isolate recovery for definitive AST.

21 November 2025 in Diagnostics
Verified
Journal Article
Open access
PDF file
Moustafa KardjadjMoustafa Kardjadj
Moustafa Kardjadj
IPItoe P. Priestly
Itoe P. Priestly
RCRoel Chavez
Roel Chavez
DDDeAndre Derrick
DeAndre Derrick
THThomas K. Huard
Thomas K. Huard
Background: Quantitative urine culture (CFU/mL) remains the reference standard for diagnosing urinary tract infections (UTIs) but is limited by delayed turnaround times and sensitivity to pre-analytic factors. Multiplex PCR panels offer rapid detection; however, standardized mappings between molecular signals and viable bacterial burdens are insufficiently defined. We used the multicenter NCT06996301 paired dataset to evaluate the analytical validity (AV), clinical validity (CV), and pre-analytic robustness of ΔCt (Ct_target − IC_Ct) as a semi-quantitative indicator of bacterial load. Methods: We analyzed 1027 paired PCR and quantitative urine culture specimens from six sites. The primary molecular predictor was ΔCt (Ct_target − IC_Ct). Species-level Spearman and Pearson correlations, species-specific linear mixed-effects calibration models (log10CFU ~ ΔCt + (1|site)), and ROC analyses were performed for the taxa meeting pre-specified sample thresholds. A pooled multilevel model assessed the collection method and time-to-processing (hours) effects (log10CFU ~ ΔCt × collection_method + ΔCt × time_to_processing_h + (1|site) + (1|run)). AV was assessed via reproducibility, internal control normalization, and site run variance. CV was assessed by ΔCt calibration and discrimination. Clinical utility (CU) was contextualized using outcomes from the parent randomized trial. Results: PCR positivity exceeded culture positivity across all sites (PCR ~82–88% vs. culture ~66–70%); this excess likely reflects a combination of molecular detection of non-viable DNA, detection of fastidious taxa less readily recovered by culture, and pre-analytic effects. For six common uropathogens (n = 90 pairs/species), ΔCt correlated strongly with log10CFU (Spearman ρ = −0.64 to −0.75; Pearson r = −0.75 to −0.83). Species-specific mixed models yielded slopes of −0.746 to −0.922 log10CFU per ΔCt unit (all p < 0.001), indicating that each 1 unit ΔCt change corresponds to a ~5.6–8.4-fold CFU difference. ROC AUCs for ΔCt discrimination were 0.78–0.84 (interpreted as good discrimination, i.e., ΔCt meaningfully improves the clinician’s probability estimate of a high CFU but does not perfectly classify every specimen). The collection method (catheter vs. clean-catch) did not materially modify the ΔCt→CFU relationship, whereas the processing delay was associated with reduced recovered CFU (~0.048 log10CFU lost per hour) and a significant ΔCt × time interaction, consistent with time-dependent viability loss driving the PCR+/culture− discordance. Conclusions: ΔCt from the DOC Lab UTM 2.0 panel shows a reproducible, analytically valid semi-quantitative measure of urinary bacterial load. Laboratories can derive assay- and workflow-specific ΔCt cut points for semi-quantitative reporting, but thresholds must be validated prospectively and paired with operational controls for specimen handling.