Response: challenges and pitfalls in the mapping of the natural history of t(12;21)–positive childhood ALL

We thank Zuna et al and Greaves et al for their interest in our recent report in Blood¹  and for addressing the conflicting observations regarding levels of ETV6/RUNX1 (previously TEL/AML1)–positive cells in healthy neonates.²  Clarification of this issue is highly important both for the understanding and continued mapping of the natural history of ETV6/RUNX1-positive acute lymphoblastic leukemia (ALL) as well as for potential future disease preventive measures. Unfortunately, the differences are not easily reconciled from the available data and we agree with Greaves et al that further studies are needed.

All 3 groups acknowledge that the ETV6/RUNX1-translocation is a frequent prenatal hit in children who later develop this ALL subtype, and all groups also found ∼ 1% ETV6-RUNX1–positive cord blood samples from healthy neonates in their initial screening rounds. Zuna et al report that 5 of 253 cord blood samples and 1 of 1 spleen sample were positive in the qualitative endpoint ETV6/RUNX1 polymerase chain reaction (PCR) screening, but the published data do not allow estimation of the level of positive cells. However, there is a remarkable ∼ 100-fold difference in the estimated frequencies of positive cells between our data (< 10⁻⁵) and the data of Mori et al²  (10⁻³-10⁻⁴). Since all studied populations were of European descent, variation in the levels of ETV6/RUNX1-positivity of such a magnitude is likely to have technical rather than biologic explanations.

Our screening was based on isolation of mRNA from fresh cord blood samples within 24 hours from birth and on linear standard curves down to a level of 10⁻⁵. We therefore find it unlikely that we would have failed to identify ETV6/RUNX1-positive mononuclear cells at levels of 10⁻³ to 10⁻⁴. Zuna et al detected the ETV6/RUNX1 transcript only in the second-run PCR in their initial screening,³  which would be compatible with low levels. Meanwhile, Mori et al²  only classified sample as positive if levels were above 10⁻⁵ by quantitative PCR, and although no information was provided regarding samples positive at lower levels, cDNA contamination at levels of 10⁻³ to 10⁻⁴ seems highly unlikely.

All 3 groups attempted to confirm their initial PCR-based screening results by alternative methods. Because our low-level findings in the first screening round precluded confirmatory fluorescence in situ hybridization (FISH) analyses, we used cell sorting of thawed fresh-frozen cord blood samples. A relative increase in the detected level of positivity by quantitative PCR would confirm the true positivity of the first screening. We screened ∼ 10⁶ CD19-positive cells, but without positive results. Greaves et al suggest instability of our frozen samples may have played a role. However, our previous comparison between ETV6/RUNX1 and ABL householding gene mRNA stability⁴  and the low ABL cycle threshold values in the present study do not indicate such instability.

Both Zuna et al and Greaves et al used FISH to confirm the PCR results of their initial screening, and the illustrations provided by Zuna et al and Greaves et al are convincing and support their findings. Irrespectively, manual screening by FISH of almost 25 000 mononuclear cells (as done by Greaves et al) is difficult, and without appropriate adjustment, which should be based on blinded screening of a similar number of cells classified as ETV6/RUNX1-negative in their initial screening, the number of FISH-positive cells might be overestimated. Thus, in the original report by Mori et al,²  one such “false-positive” cell by FISH was found among 3199 cells, and the difference (1/3198 PCR-negative vs 35/20 901 PCR-positive) is not significant, although it strongly indicates the presence of ETV6/RUNX1-positive cells in healthy neonates.

A reliable accurate characterization of the occurrence of preleukemic cells in healthy neonates is highly relevant to both the design and execution of future investigations of the natural history of ETV6/RUNX1-positive ALL. We therefore fully support Greaves et al's call for further studies of the issue.