Clinical performance of sFlt ‐1/ PlGF ratio triage in suspected pre‐eclampsia: A prospective, observational multicenter diagnostic‐triage study in South Africa ( TsHEPPO study)

OBJECTIVE: To evaluate the diagnostic and short-term predictive performance of the soluble fms-like tyrosine kinase-1/placental growth factor (sFlt-1/PlGF ratio) in South African women with suspected pre-eclampsia (PE), comparing thresholds of ≥ 33 and ≥ 38 in a predominantly African population. METHODS: This was a prospective, observational diagnostic-triage multicenter study across four tertiary hospitals in South Africa. A total of 555 women with suspected PE with complete outcome data were enrolled from four study sites. The sFlt-1/PlGF ratio (Elecsys, Roche Diagnostics) was measured at presentation and comprehensive clinical and delivery outcome data were collected. Diagnostic and short-term predictive performance for sFlt-1/PlGF thresholds (≥ 33 and ≥ 38) were assessed using sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), receiver operating characteristics (ROC) and associated area under the curve (AUC). RESULTS: PE was diagnosed in 349/555 women (62.9%). The sFlt-1/PlGF ratio showed moderate discrimination for PE (area under the receiver operating characteristic curve [AUROC] 0.78, 95% confidence interval [CI]: 0.7-0.8) overall. Continuous-data ROC analysis in 418 women from Durban, Johannesburg and Pretoria yielded an AUROC of 0.8465 (95% CI: 0.8090-0.8841). In the all-site analysis, sensitivity, specificity, PPV, NPV and accuracy were 67.05%, 89.32%, 91.41%, 61.54% and 75.32% at ≥ 33, and 62.46%, 90.29%, 91.60%, 58.68% and 72.79% at ≥ 38, respectively. In short-term prediction, both thresholds showed high sensitivity (86.5%) for PE within 7 days while specificity/PPV were 77.3%/81.8% at ≥ 33 and 81.8%/84.9% at ≥ 38, with NPVs of 82.9% and 83.7%, respectively. By 28 days, NPVs declined to 62% and 64%. CONCLUSION: The sFlt-1/PlGF ratio is a clinically useful biomarker primarily as a rule-in triage tool for suspected PE in South Africa. The ratio showed high specificity and PPV in this predominantly African cohort. The exploratory ≥ 33 threshold increased sensitivity relative to ≥ 38, with a small reduction in specificity in the all-site estimates. The findings support its evaluation as an adjunct to clinical assessment in women with suspected PE. Both thresholds showed limited rule-out performance beyond short follow-up supporting the need for reassessment and repeat testing where clinically indicated. These findings support consideration of region-adapted thresholds and highlight the value of biomarker-guided triage in African settings.