Deriving the optimal limit of detection for an HCV point-of-care test for viraemic infection: Analysis of a global dataset

Abstract

Background & aims: Affordable point-of-care tests for hepatitis C (HCV) viraemia are needed to improve access to treatment in low- and middle-income countries. Our aims were to determine the target limit of detection (LOD) necessary to diagnose the majority of people with HCV eligible for treatment, and identify characteristics associated with low-level viraemia (LLV) (defined as the lowest 3% of the distribution of HCV RNA) to understand those at risk of being misdiagnosed.

Methods: We established a multi-country cross-sectional dataset of first available quantitative HCV RNA measurements linked to demographic and clinical data. We excluded individuals on HCV treatment. We analysed the distribution of HCV RNA and determined critical thresholds for detection of HCV viraemia. We then performed logistic regression to evaluate factors associated with LLV, and derived relative sensitivities for significant covariates.

Results: The dataset included 66,640 individuals with HCV viraemia from across the world. The LOD for the 95th and 99th percentiles were 3,311 IU/ml and 214 IU/ml. The LOD for the 97th percentile was 1,318 IU/ml (95% CI 1,298.4-1,322.3). Factors associated with LLV, defined as HCV RNA <1,318 IU/ml, were younger age 18-30 vs. 51-64 years (odds ratios [OR] 2.56; 95% CI 2.19-2.99), female vs. male sex (OR 1.32; 95% CI 1.18-1.49), and advanced fibrosis stage F4 vs. F0-1 (OR 1.44; 95% CI 1.21-1.69). Only the younger age group had a decreased relative sensitivity below 95%, at 93.3%.

Conclusions: In this global dataset, a test with an LOD of 1,318 IU/ml would identify 97% of viraemic HCV infections among almost all populations. This LOD will help guide manufacturers in the development of affordable point-of-care diagnostics to expand HCV testing and linkage to care in low- and middle-income countries.

Lay summary: We created and analysed a dataset from 12 countries with 66,640 participants with chronic hepatitis C virus infection. We determined that about 97% of those with viraemic infection had 1,300 IU/ml or more of circulating virus at the time of diagnosis. While current diagnostic tests can detect as little as 12 IU/ml of virus, our findings suggest that increasing the level of detection closer to 1,300 IU/ml would maintain good test accuracy and will likely enable development of more affordable portable tests for use in low- and middle-income countries.

Keywords: Diagnosis; Hepatitis C virus; Limit of detection; Point-of-care; Viraemia, Affordable.

Figure 1.

Frequency distribution of HCV RNA (log₁₀ IU/mL) among participants in a combined cross-sectional dataset from Cambodia, Canada, Cameroon, Egypt, Georgia, India, Indonesia, Malaysia, Mozambique, Pakistan, Thailand, and Vietnam with chronic hepatitis C virus (HCV). The analytic level of detection for: a) centralized nucleic acid tests (NAT) Xpert® HCV Viral Load (40 IU/mL) and Abbott RealTime HCV Viral load (12 IU/mL), b) point of care NAT Genedrive® HCV ID Kit (2362 IU/mL) and c) Abbott ARCHITECT HCV core antigen (HCVcAg) test (in IU/ml; 3000IU/ml approximately equivalent to 3 fmol/L) are marked in comparison to the limits of detection (LOD) derived for the 99^th, 97^th, and 95^th percentiles in the dataset. *Marks the lower LOD for tests performed in India that were detected but not quantified.

Figure 2.

Violin plot of the HCV RNA distribution (log₁₀ IU/mL) for each covariate in the total population cohort. In each violin plot, a circle denotes the median and the interquartile range is shown by box in solid black. Overlaid on this box plot is a kernel density plot indicating more data where the plot is thicker and less where it narrows. The dashed horizontal marker indicates the 1318 IU/mL level of detection derived from the HCV RNA frequency distribution.

Figure 3.

Graph of relative percentiles of HCV viraemia corresponding with the limit of detection derived from the total population dataset (1318 IU/mL) for covariates significantly associated with low-level viraemia in the (A) total population dataset and (B) imputed dataset. The reference (ref) groups are included for each category. The dashed vertical line marks the 97^th percentile from the total population dataset.

Figure 3.

Graph of relative percentiles of HCV viraemia corresponding with the limit of detection derived from the total population dataset (1318 IU/mL) for covariates significantly associated with low-level viraemia in the (A) total population dataset and (B) imputed dataset. The reference (ref) groups are included for each category. The dashed vertical line marks the 97^th percentile from the total population dataset.