LAUSR

The high genetic variability of hepatitis B virus (HBV) can impair DNA quantification. Here, we investigate a major underquantification of HBV by the cobas TaqMan HBV assay (CTM; Roche). In France, between 2005 and 2017, HBV DNA was detected in 3,102 blood donations by use of the CTM (95% limit of detection [LOD95], 4.8 IU/ml). HBV strains were sequenced in the S region (LOD95, ∼30 IU/ml). Concordant (n = 120) and discordant (n = 45) samples were identified according to the agreement between the plasma viral load (pVL) determined by the CTM and sequencing; all samples were also quantified using the RealTime HBV assay (RTH; Abbott). ABSTRACT The high genetic variability of hepatitis B virus (HBV) can impair DNA quantification. Here, we investigate a major underquantification of HBV by the cobas TaqMan HBV assay (CTM; Roche). In France, between 2005 and 2017, HBV DNA was detected in 3,102 blood donations by use of the CTM (95% limit of detection [LOD95], 4.8 IU/ml). HBV strains were sequenced in the S region (LOD95, ∼30 IU/ml). Concordant (n = 120) and discordant (n = 45) samples were identified according to the agreement between the plasma viral load (pVL) determined by the CTM and sequencing; all samples were also quantified using the RealTime HBV assay (RTH; Abbott). The viral signature, cloning, and mutagenesis were used to characterize the polymorphism responsible for CTM misquantification. A CTM-RTH discordance (>1 log IU/ml) was found in 14/45 samples that had low pVLs and were successfully genotyped (pVLlow genoS+). PreC/C clones of concordant (C1, C2) and discordant (D1, D2) strains were used to challenge the CTM. Strains D1 and D2 were highly underquantified (42- and 368-fold). In clones, mutating the region corresponding to the CTM reverse primer from a discordant sequence to a concordant sequence restored the levels of quantification to 24% (D1→C1) and 59% (D2→C1) of theoretical levels, while mutating the sequence of a concordant strain to that of a discordant strain led to 78-fold (C1→D1) and 146-fold (C1→D2) decreases in quantification. Moreover, mutating positions 1961 and 1962 was enough to induce a 5-fold underquantification. We conclude that the CTM underestimates pVLs for HBV strains with mutations in the reverse primer target. Specifically, the polymorphism at nucleotides 1961 and 1962 is naturally present in 4.79 and 4.22% of genotype A and D strains, which are highly frequent in Europe, leading to a 5-fold decrease in quantification. Quantification using the new-generation Roche C4800 assay is not affected by this polymorphism.