To the Editor:
In a recent issue of Blood, Uckun et al1 reported a high frequency of the MLL/AF4 expression not only among infant patients with acute lymphoblastic leukemia (ALL) but also among children older than 1 year (12 out of 127) at the diagnosis of the disease. This group with reverse transcription-polymerase chain reaction (RT-PCR) positivity only in the second round comprises both patients with or without cytogenetically detectable t(4;11) and with or without MLL gene rearrangement detectable on the DNA level by the Southern blot analysis. Moreover, investigators found low levels of MLL/AF4 expression also in a significant proportion of fetal tissues and in bone marrow (BM) of 1 healthy infant out of 6 tested. Based on these results, Uckun et al conclude that MLL/AF4 fusion transcripts can be present in normal hematopoietic cells and that their presence per se is not sufficient for leukemic transformation of normal lymphocyte precursors. Furthermore, the authors question the reliability of RT-PCR detection of minimal residual disease in MLL/AF4–positive patients as well as its use for the screening of cord bloods before transplant. In contrast, our results show that in pediatric ALL patients, nested RT-PCR may achieve detection limit 0.01% without compromising specificity. Moreover, our findings suggest that MLL/AF4 fusion is not expressed in the cord blood samples of healthy newborns.
We conducted a large prospective study on pediatric ALL on behalf of the Czech Pediatric Hematology Working Group, and analyzed the frequency and clinical impact of various translocations, including t(4;11). We used the nested RT-PCR approach for the detection of the fusion gene MLL/AF4. RNA was extracted using a method modified after the single-step technique described by Chomczynski and Sacchi.2 RNA integrity and quantity was evaluated in nondenaturing agarose gel electrophoresis. The single-round amplification of ABL exons A2 to A3 was then used for cDNA quality control. As we know from our previous work (Divoky et al, manuscript submitted), exon 6 of the MLL gene is frequently absent in the alternatively spliced MLL/AF4 transcripts. Therefore, in contrast to the method used by Uckun et al, our RT-PCR used primers in MLL exon 5 for both the first and second round of our PCR. The sensitivity of this approach, evaluated using limiting dilution of the MLL/AF4–positive cells into normal cells, is 10−4(0.01%), the sensitivity identical to the one described by Uckun et al.
We started our prospective study in June 1994. Since then, we found 7 infant patients positive for MLL/AF4 out of 11 tested (64%), contrasting with only 1 out of 206 patients aged 1 to 18 years (0.5%). Out of these 8 children, 7 were positive in the first round and 1, tested only after the induction therapy block and at subsequent relapse, in the second round. Cytogenetic analysis was available in 5 of these patients, and in 4 cases t(4;11) was detected. In 1 remaining patient, cytogenetic analysis revealed deletion in 11q23 locus. To test the hypothesis that MLL/AF4–expressing cells could be present among hematopoietic cells of healthy newborns, we applied the identical two-round RT-PCR approach to the cord blood samples from 103 full-term healthy newborns born in our hospital from February to March 1998 and found none of them to be MLL/AF4–positive.
The individuals positively tested for MLL/AF4 expression in the study of Uckun et al could be divided (according to the results of cytogenetic analysis, Southern blot for MLL rearrangement, and RT-PCR) into 3 subgroups. The first subgroup (A) comprises patients with t(4;11)+, MLL rearrangement present, and RT-PCR positivity in the first round (standard PCR), thus representing expected results. Subgroup B then includes t(4;11)+ or −, MLL rearrangement +, and RT-PCR second round positives; and the subgroup C includes t(4;11)−, MLL rearrangement −, and RT-PCR second round positives. Patients encompassed in the two latter subgroups (B and C) must represent two rather distinctive MLL/AF4–positive populations: one relatively large (at least 5% of all mononuclear cells, taking into account that MLL-rearranged cells are detectable by Southern blot analysis, but Southern blots presented in the report seem to indicate much higher proportion of MLL-rearranged cells) with low expression of MLL/AF4, and the other relatively minute population with high expression level (subgroup C). Remarkably, without exception all examined normal samples (comprising fetal livers, fetal BM samples, and BM of presumably healthy infants) fall into the last subgroup, whereas majority of ALL patients (noninfant patients in particular) fall under the characteristics of the previous one. Although it is not clear from the report of Uckun et al how many of the “normal” MLL/AF4–positive samples were actually analyzed by Southern blot, this distribution seems to be notable. However, the authors did not attempt to interpret this distinction.
In the report of Uckun et al, we find another intriguing phenomenon. We know from our previous experience (Divoky et al, manuscript submitted) that leukemic cells of MLL/AF4–positive patients express usually more than one (and up to six) different alternatively spliced transcripts, representing very specific pattern of PCR products in the post-PCR gel. Surprisingly, gel photographs included in the report of Uckun et al show only a very limited amount of alternatively spliced variants (3 in total), with no more than one PCR product visible in the gel for one individual patient. DNA-based analysis, documenting distinctive patient-specific junction of MLL and AF4 genetic material, would ultimately exclude the possibility of PCR artifact or carry-over contamination.3 Southern blot analysis shown by Uckun et al is used as an independent method to support some of the RT-PCR-based findings. However, to cite an instance, it is technically impossible to detect the same rearrangement using the PCR with primers in exon 6 and by the Southern blot with 98.40 probe after BamH1 restriction (Table 4, patients INF-6, PEDS-5). The probe and the restriction sites lie upstream from the sixth exon.4,5
One of the conclusions made by Uckun et al was that nested RT-PCR may not be suitable for MLL/AF4 screening of cord blood because they have found positive samples among presumably healthy fetal or infant tissues (other than cord blood). We consider analysis of healthy cord blood more appropriate for such a statement.
We tried to summarize some intriguing details in the study of Uckun et al that in our opinion need further elucidation. Our data do not support any of the hypotheses implied by Uckun et al. We found neither patients with low expression of MLL/AF4 fusion gene nor the MLL/AF4–positive cells in any of 103 cord blood samples tested. We believe that nested RT-PCR method can be used for the MRD follow-up in t(4;11)+ patients as well as for potential screening of cord blood before transplant. Uckun’s data, although showing unanticipated biological observation, should not be interpreted as an argument for considering an MLL/AF4–positive specimen free of leukemic blasts.
This work was supported in part by Grants No. 3425-3 and 3920-3 from Ministry of Health, Czech Republic.