Response to Tang et al.

Tang et al. ¹ contribute valuable and new information on the relationship between the number of affected genes in α-thalassaemia and the percentage of HbBart's haemoglobin in the haemoglobin (Hb) profile. With a method different from the one described in our paper, ² they show a similar relationship between the percentage of HbBart's and the number of affected alpha genes, but with a profoundly higher degree of separation of the genotype associated with HbH disease and in α-thalassaemia, in comparison with that presented in our paper. While we applaud this achievement, especially in respect to the focus of neonatal screening, it raises some issues.

Firstly, the method of Hb detection is quite different from ours and may inherently produce a better separation of the HbBart's peak from the other Hb peaks and degradation products. In a previous paper we showed that the coefficient of variation for peak percentages was rather high compared with other systems with the same technique. ³ Secondly, Tang et al. measure in a different matrix and in a different postnatal time frame from our study. They measure in fresh cord blood, we measure in dried blood spots collected on filter paper for routine neonatal screening, with a collection time window of about 3–7 days postnatally and a transit time of 1–4 days. The postnatal collection time, the transit period and the matrix itself (dried blood versus fresh cord blood) may cause a variation in Hb degradation products (ending up in the FAST peak [first peak eluted from the column]), as well as variation in the relative distribution of the various Hb peaks, ultimately producing a less efficient separation of the HbBart's peak in dried blood spots compared with fresh cord blood.

Therefore, it remains to be seen whether the results from Tang et al. still stand when their method is applied in the context of routine newborn screening using dried blood spots sent by mail.

Tang et al. present the results of seven cases of HbH disease with a HbBart's peak of 17.5–22.5%. Our study was from few cases, and these additional data give us some confidence that our current cut-off value of 16% FAST peak is applicable, taking into account that we generally observe higher FAST-peak levels compared with those in Tang et al.