Acute lymphoblastic leukemia (ALL) is the most common type of pediatric cancer while acute myeloid leukemia (AML) occurs approximately 5 times less frequently in children. There is evidence that some childhood acute leukemias originate in utero. Indirect proof of the pre-natal origin has been given by the cases of monozygotic twins with genotype-identical leukemia [1]. Direct evidence of the prenatal origin of ALL was demonstrated in 1997. Gale et al. backtracked MLL/AF4 fusion gene, a leukemo-specific clonal marker, to neonatal dried blood spots on filter-paper (Guthrie cards) of 3 children who developed ALL at the ages of 5 months to 2 years [2]. Guthrie cards positive for the TEL/AML1 (ETV6/RUNX1) were found in more than 10 ALL patients with this fusion gene [3–6].

Clonal rearrangements of immunoglobulin (Ig) and T-cell receptor (TCR) genes are found in >90% of childhood ALL and can be used for retrospective analysis of cases without detectable chromosomal translocations. Identical rearrangements of Ig/TCR genes in twins and triplets with concordant leukemia [7–9] and their presence in Guthrie cards of patients who developed leukemia [10–12] provide satisfactory evidence of prenatal origin of ALL.

The availability of clonal markers in AML is lower. Approximately 12 % of childhood AML cases bear the translocation between chromosomes 8 and 21 resulting in AML1/ETO (RUNX1/CBFA2T1) fusion gene. Translocation t(15;17) with PML/RARalpha fusion occurs in ~6% and t(16;16) or inv(16) with chimera CBFbeta/MYH11 is found with similar frequency among children with AML [13]. Internal tandem duplications of FLT3 gene (FLT3/ITD) are always in frame but vary in length. FLT3/ITD is associated with poor prognosis and occurs in ~17% of children with AML [14].

There are only few studies concerning in utero origin of AML. Twins studies with AML patients indicate that both infant [15, 16] and non-infant [17] acute myeloid leukemia can be initiated in utero. The AML1/ETO fusion gene was detected in neonatal blood spots in 5/10 t(8;21) positive patients [18]. Apart from this study, only 3 single cases of AML with positive Guthrie card were reported: for PML/RARalpha, CBFbeta/MYH11 [19] and MLL/AF10 fusion genes [20], respectively.

Our study aimed to screen Guthrie cards in larger cohort of non-infant childhood AML and ALL patients for the presence of their respective leukemic markers. Clone-specific markers used for the screening included PML/RARalpha, AML1/ETO and CBFbeta/MYH11 fusion genes, chimeric products with the MLL gene and for the first time also the FLT3/ITD in patients with AML. In ALL cases we used TEL/AML1 fusion gene and Ig/TCR gene rearrangements. These molecular markers represent suitable targets for backtracking of leukemic cells in newborn material, because they are clonotypic and their incidence is ~40% and >90% in AML and ALL, respectively. To our knowledge, this is the largest study analyzing prenatal origin of childhood leukemias presented so far.

We examined neonatal blood spots for the presence of clonotypic markers of (pre-)leukemic cells in the cohort of 25 patients with acute leukemia using a sensitive PCR technique.

We confirmed the results of previous studies on prenatal origin of ALL based on the same approach. In our cohort, the presence of clonotypic marker of the leukemic cells was backtracked in 3/12 ALL patients (25%). The data on the prenatal origin of AML are less comprehensive. A convincing study on AML backtracking in AML1/ETO patients was published by Wiemels et al. [18] with 5/10 patients positive. Other subtypes of non-infant childhood AML were examined rarely. There are studies with single positive Guthrie cards for MLL/AF10, PML/RARalpha and CBFbeta/MYH11 fusion genes, respectively [19, 20], and to our knowledge no other cases of non-infant childhood AML were ever confirmed to be of pre-natal origin using this approach. It is more demanding to perform a large Guthrie card study on AML patients compared to ALL – the diagnosis is about five times less frequent and there is no universal marker for the genetic screening in AML cases (unlike the Ig/TCR rearrangements in ALL). In our study we were able to collect Guthrie cards from 13 AML children with a suitable molecular marker – we found intronic breakpoints in 12 patients with 6 different fusion genes and, moreover, we used for the first time FLT3/ITD for the analysis in two patients. Three different fusion genes involving MLL locus were used in our study (MLL/AF6, MLL/AF9 and MLL/AF10). MLL translocations are closely linked to congenital and infant leukemias. However, to our knowledge only three non-infant patients with MLL rearrangement detected in Guthrie card has been presented so far. The case report of MLL/AF10 positive AML mentioned above [20] and two MLL/AF4 positive patients with ALL diagnosed at 2 and 6 years, respectively [2, 4]. In our study, the youngest AML patient (case AML9 with MLL/AF6 fusion gene) was 18 months old at diagnosis.

The limitation of the Guthrie card approach is the sensitivity of the assay. A segment (usually 1/8) of a blood spot analyzed in one PCR reaction contains only ~30 000 nucleated cells and the DNA on the spot can be of a poor quality (depending on the storage conditions and time). To estimate a yield of our DNA isolation from Guthrie cards we performed beta-globin housekeeping gene quantitative PCR analysis in some samples. Providing the DNA content of one cell is ~6 pg and there is approximately 200 000 – 300 000 nucleated cells in one spot our results show that we were able to extract ~20–60% of the DNA theoretically originally present on the Guthrie card. To our knowledge this type of analysis has never been shown and taking into account the storage time and conditions we believe the result is very satisfactory.

We aimed to optimize the PCR to reach the sensitivity of at least 10^-4 in all patients and, moreover, we designed the PCR so that the 1^st round amplicon is shorter than 150–200 base pairs enabling us to successfully analyze also potentially highly fragmented DNA on the Guthrie card. To determine the sensitivity of each target we used dilutions of DNA from the diagnostic sample into DNA of healthy volunteer in our dilution series. Although the optimalization of such experiments is more laborious compared to reactions using dilution into water (there is higher possibility of non-specific amplifications from the volunteer DNA) we believe this approach better reflects the real situation, offers more reliable results and should be used in all backtracking studies. We fulfilled the assessed sensitivity criteria in all targets except for two patients (AML8 and AML13) where only some of the replicates of the 10^-4 dilution were positive and the sensitivity was determined as 10^-3-10^-4. However, both these two patients had low blast count in the diagnostic sample (that serves as a 10⁰ dilution) and thus the reached sensitivity in fact is 3 positive cells/10^-4-10^-5 cells and 5 positive cells/10^-5-10^-6 cells in patients AML8 and AML13, respectively. Therefore, our approach allowed us to detect the pre-leukemic clone in all samples providing 1–3 positive cells were present in the analyzed sample.

On one hand, the highest possible specificity of the analysis is crucial, however, on the other hand, the approach used in the study requires an extreme caution in terms of possible contamination. Once a clonal marker from the diagnostic sample is amplified by PCR and used in Guthrie card experiments as a positive control, threat of contamination can not be excluded. Moreover, it is impossible to distinguish a sporadic contamination from real positivity. We applied very strict anti-contamination precautions. The Guthrie card experiments were performed in a separate room and only highly diluted diagnostic samples (10^-4-10^-5) were used as positive controls. During the study we experienced no contamination in any of the negative-control samples.

The results from the ALL cohort confirm that our approach is appropriate both for the Ig/TCR targets (3/15 positive) and for the fusion gene analysis (1/3 positive).

Previously, Wiemels et al reported 5 out of 10 t(8;21) childhood AML cases with fusion gene positive Guthrie cards [18]. Only single examples of t(15;17), inv(16) and MLL-AF10 with fusion gene positive Guthrie cards have been reported [19, 20]. As the present study included only two cases of AML1-ETO positive leukemia, it is not possible to say that this occurs prenatally at a higher frequency.

Young age at diagnosis is probably not a crucial prerequisite for the pre-natal origin of childhood acute leukemia as cases with a long latency to leukemia onset were described [1, 4, 7, 18, 19, 29]. However, the fact that the ALL patients in our study were significantly younger at diagnosis compared to AML patients (median 3.5 and 6 years, respectively) should be mentioned here.

The negative result in Guthrie card analysis can be caused at least by three different reasons (providing the sensitivity of detection is adequate) – the leukemia does not originate prenatally, the pre-leukemic clone at birth is too small to be present at Guthrie card spot or the marker used for the backtracking is not the initiating or early event in leukemogenesis. The latter is probably the case with the FLT3/ITD. This abnormality was never tested before as a marker of leukemic cells in Guthrie cards. The positivity of this abnormality in neonatal blood is rather unlikely as the latest data indicate that this aberration has probably a post-natal origin [30, 31]. And indeed, we did not detect the FLT3/ITD in any of the Guthrie cards tested.

The sensitive patient-specific PCR targeted either to the FLT3/ITD breakpoint or to the intronic fusion of the various chimeric genes provide potentially a very useful tool not only for backtracking of leukemia but also for the minimal residual disease monitoring in almost 50% of children with AML, where common molecular marker is unavailable (unlike Ig/TCR rearrangements in ALL).

In our study on childhood non-infant acute leukemias we confirmed that the prenatal origin is common in ALL patients (25% in our cohort) while all neonatal blood spots in our AML cohort were negative. Although we found all of the Guthrie cards from AML patients negative, we do not think that our results completely contradict the previous reports with positive neonatal blood spots of AML patients [18–20]. However, we feel that with the exception of AML1/ETO fusion gene, where the prenatal origin has been proven reliably, reports of prenatal origin in non-infant childhood AML are isolated single cases. These should be confirmed by other studies or reports as a false-positivity can not be ruled out completely with the approach used. For example the E2A/PBX1 fusion gene was found to be "weakly" positive in Guthrie cards of 2 out of 15 patients tested, and, nevertheless, the data on E2A/PBX1 are interpreted as being suggestive of a post-natal origin of this fusion gene [32].

The results of our study using the same approach for both ALL and AML clearly indicate that either the prenatal origin of AML is less frequent or the load of pre-leukemic cells is significantly lower at birth in AML compared to ALL cases.