This is the first study describing a high-dose hook effect in the Liaison Borrelia IgG immunoassay. The hook effect was primarily observed in undiluted samples with a reactivity of > 98.7 AU/mL. Despite this, a hook effect could still be observed in a few samples with a reactivity of < 98.7 AU/mL. The lowest level at which a hook effect was observed was 77.8 AU/mL, which, following dilution, increased to 1579 AU/mL. The breakpoint of 48 AU/mL, as estimated by the segmented linear regression, represents the lowest theoretical value at which the hook effect is predicted to occur. However, many of the observed overdilutions (48.3%) occurred above the breakpoint making it an unfeasible dilution threshold. The lowest observed hook effect was within the confidence interval of the linear segmented regression (48–108 AU/mL), corroborating that the hook effect can occur within this interval. Thus, caution must be exercised when interpreting IgG antibody results in patients strongly suspected of late manifestations of LB even at these lower levels. Of note, a double hook effect was observed when a sample was reanalysed. This is not surprising since every diluted sample result is multiplied by the dilution factor and is therefore just as vulnerable to the hook effect as undiluted samples. Diluted samples with results in the span between 98.7 AU/mL and 240 AU/mL (values before multiplying with the dilution factor) should therefore be considered for further dilution since the IgG reactivity of the diluted sample can still be high enough to cause a hook effect.

Prior studies on the hook effect within clinical microbiology have focused on flocculation tests or lateral flow assays, which typically lead to false negative results [3, 4, 5, 6]. The hook effect has also been described in assays detecting albumin and a wide range of tumor markers [10]. In general, the vast majority of assays with hook effect interferences are direct immunoassays and can lead to both false negative and false low results. Indirect immunoassays, however, are in general less sensitive to the hook effect but can still lead to false low results [10]. The Liaison Borrelia IgG assay is an indirect immunoassay displaying hook effect, which is uncommon.

The hook effect entails several clinical implications for patients under investigation for LB. When present in serum, it can lead to late manifestations of LB, characterised by high IgG levels, being missed [1]. However, even though this study investigated serum samples, it is likely that the hook effect also applies to cerebrospinal fluid (CSF), as the hook effect is a limitation of the immunoassay, not the type of sample. A similar dilution threshold of 98.7 AU/mL is therefore warranted for CSF until the hook effect has been further investigated. In this context, a hook effect could impact the calculation of the intrathecal antibody index, as a false-low result in CSF can lead to a false-negative index, which may result in a missed diagnosis of neuroborreliosis and potential undertreatment [1, 11]. Conversely, a false-positive index can also occur if the serum sample alone displays a hook effect, resulting in possible misdiagnosis and unnecessary antibiotic treatment. No studies have reported a hook effect in Bbsl IgM assays, but such a finding would warrant similar precautions when calculating the intrathecal antibody index.

The inclusion criteria (≥ 75 AU/mL and < 75 AU/mL with concomitant IgM) were based on our experience that the hook effect is mainly observed in the upper range. As overdilutions are overrepresented in the lower range of the immunoassay, we used the presence of IgM reactivity ≥ 75 AU/mL to select samples for dilution, thereby limiting unnecessary overdilutions. These criteria were a pragmatic approach to reduce the time and costs of the study but may have excluded some patients with late manifestations in the lower range without detectable IgM reactivity. No hook effect was observed in the 45 samples displaying a reactivity below 77.8 AU/mL in the undiluted samples and it is therefore unlikely that the inclusion criteria have introduced a substantial selection bias. A single dilution group, which included both 1:10 and 1:50 dilutions, was used because the 1:50 dilution group was too small for any subgroup analysis but was necessary for the descriptive statistics due to truncated values (> 240 AU/mL) in the 1:10 dilution group.

The lowest change in reactivity categorised as the hook effect cluster was an 129% increase and the GMM can therefore be considered relatively conservative since the methodological variation is significantly lower [12]. The segmented linear regression may underestimate the real breakpoint since not all samples display a hook effect.

Whether a hook effect also affects IgM antibodies or is also present in other immunoassays detecting Bbsl IgG antibodies is currently unknown and should be subject to further investigation. In addition, studies investigating the hook effect in CSF are warranted.

In conclusion we detected a hook effect in a substantial number of samples, which could lead to late manifestations of LB, often presenting with high IgG levels, being missed. Furthermore, the hook effect can lead to both false-positive and false-negative intrathecal index calculations in patients suspected of neuroborreliosis. Thus, a dilution threshold of > 240 AU/mL, as recommended by the manufacturer, is inadequate. Instead, a dilution threshold of 98.7 AU/mL for the analysis of Bbsl IgG antibodies using the Liaison Borrelia IgG immunoassay is reasonable. Despite this, caution is warranted, even at IgG values below this threshold, when testing patients suspected of late manifestations of LB.