Response to Goodness et al.

Goodness et al. ¹ from the Michigan Department of Community Health (MDCH) report their experience of a change of the resin lot of the BioRad Variant^TMnbs reagents causing a dramatic drop in FAST-peak values of the high-performance liquid chromatography (HPLC) analysis. Further to their findings, we present the analytical consequences of a gradient change in the Dutch screening programme.

In 2011 BioRad introduced a new gradient (method 1.013 ‘new’, replacing method 1.08 ‘old’) on the Vnbs System to produce a sharper and higher F1 peak (containing acetylated HbF). In our hands, using the Vnbs system, the percentage HbF (‘total F’) is calculated by adding the F1 peak percentage with the F peak percentage. To determine the effect of the new gradient, we re-analysed the samples from our previous paper ² with both method 1.08 (old) and 1.013 (new). We also analysed 186 random fresh samples of the routine screening programme, using both methods.

Method 1.013 (new) produced clearly higher F1 peak percentages and slightly lower F peak percentages. There was a drastic decline in the FAST peak percentage, which contains the HbBart's peak used in screening for HbH disease. The HbA peak percentages were unaffected.

Re-analysis of the samples from our previous study ² with method 1.013 (new) showed a good correlation (r = 0.98) between methods, with FAST peak percentages that were, on average, 37% lower than those obtained with method 1.08. With fresh samples from the routine screening programme, average FAST peak percentages achieved with method 1.013 (new) were 44% lower compared with the analysis with method 1.08 (old).

A change of aspects of the HPLC procedure by a manufacturer, whether a change in resin, gradient or some other aspect, will lead to changes in the Hb peak percentages. Inevitably, a re-evaluation of the cut-off values for the various haemoglobinopathies is needed, involving a validation consisting of a comparison of results obtained with the old and new methods. Because of the scarcity of cases, establishing cut-off values will ultimately be a combination of this analytical validation and the original cut-off values, rendering the establishment of new cut-off values, in part, arbitrary.

To illustrate this, the Dutch newborn population has a low prevalence of HbH disease, so for alpha-thalassaemia screening, method 1.013 (new) was introduced concomitantly with a lower cut-off value for the FAST peak (from 22.5% to 16%). This cut-off value would be expected to result in a similar number of referrals (N = 11, with no known missed patients) to the old cut-off value of 22.5% (N = 13). The new cut-off value was introduced on 1 October 2011. Initial results indicate that the number of referrals decreased slightly, from 41 to 24 annually, with no known false-negative results. In the Netherlands we will continue to monitor the referrals and false-negatives, and plan to publish our results.