Assessment of the Roche Linear Array HPV Genotyping Test within the VALGENT framework

Abstract

Background: Cervical cancer screening programs are switching from cytology-based screening to high-risk (hr) HPV testing. Only clinically validated tests should be used in clinical practice.

Objectives: To assess the clinical performance of the Roche Linear Array HPV genotyping test (Linear Array) within the VALGENT-3 framework.

Study design: The VALGENT framework is designed for comprehensive comparison and clinical validation of HPV tests that have limited to extended genotyping capacity. The Linear Array enables type-specific detection of 37 HPV types. For the purpose of this study, Linear Array results were designated as positive only if one of the 13 hrHPV types also included in the Hybrid Capture 2 (HC2) was detected. The VALGENT-3 framework comprised 1600 samples obtained from Slovenian women (1300 sequential cases from routine cervical cancer screening enriched with 300 cytological abnormal samples). Sensitivity for cervical intraepithelial neoplasia of grade 2 or worse (CIN2+) (n=127) and specificity for <CIN2 (n=1216) were assessed for Linear Array and for HC2 and non-inferiority of Linear Array relative to HC2 was checked. In addition, the prevalence of separate hrHPV types in the screening population, as well as the concordance for presence of HPV16, HPV18 and other hrHPV types between Linear Array and the Abbott RealTime High Risk HPV test (RealTime) were assessed.

Results: The clinical sensitivity and specificity for CIN2+ of the Linear Array in the total study population was 97.6% (95% CI, 93.3-99.5%) and 91.7% (95% CI, 90.0-93.2%), respectively. The relative sensitivity and specificity of Linear Array vs HC2 was 1.02 [95% CI, 0.98-1.05, (p<0.001)] and 1.02 [95% CI, 1.01-1.03, (p<0.001)], respectively The overall prevalence of hrHPV using the Linear Array in the screening population was 10.5% (95% CI, 8.9-12.3%) with HPV16 and HPV18 detected in 2.3% and 0.9% of the samples, respectively. Excellent agreement for presence or absence of HPV16, HPV18 and other hrHPV between Linear Array and RealTime was observed.

Conclusions: Linear Array showed similar sensitivity with higher specificity to detect CIN2+ compared to HC2. Detection of 13 hrHPV types with Linear Array fulfils the clinical accuracy requirements for primary cervical cancer screening.

Keywords: Cervical cancer; HPV genotyping; Human papillomavirus; Linear array; Test validation; Valgent.