Previous literature reported that MLL-r positive ALL can develop among children at all ages, accounting for 2.5–5% of children with initial ALL [2, 3], and the detection rate of MLL-r positive ALL in infants younger than 1 year of age is even as high as 23.8–79% [9,10,11]. The detection rate of MLL-r positive ALL patients in this study was 4.9%, which was basically consistent with the literature. However, the detection rate in infants younger than 1 year of age was only 3.2%, which was much lower than the above mentioned literature, and the reason for this might be related to the sample size of MLL-r positive patients included in different studies.
Compared with the MLL-r negative group of patients, the clinical characteristics of MLL-r positive cases in this study included: a young age of onset, with a median age of onset of 3.1 years, and 12.9% of infant leukemia, which was much higher than that of the MLL-r negative group. In addition, a high WBC of onset, with a significantly higher proportion of initial WBC greater than 50 × 109/L than that of the negative group and a high percentage of 64.5% in the high-risk group, which were consistent with previous studies. The MLL-r, as one of the subgroups of ALL, may be associated with certain immunophenotypic characteristics. Moorman et al  showed that MLL-r positive ALL was mostly of B-cell lineage and common B or pro-B immunophenotypes were more common, accounting for 62% of the cases. Peterson et al  retrospectively found that in 806 children with T-lineage ALL, 27 (3.3%) MLL-r positive cases were detected. In contrast, our findings showed that the immunophenotype of MLL-r positive children was more prevalent in common-B, immature B and T cells compared to the MLL-r negative group, which was basically consistent with Moorman and Peterson’s report. In China, it has been reported that the immunophenotype pro-B is a risk prognostic factor in MLL-r positive cases . In contrast, the results of the multivariate analysis in this study showed that the T-lineage immunophenotype was an independent prognostic factor affecting MLL-r positive ALL, which was inconsistent with the above-mentioned studies and we implied that it might be related to the small number of reported cases in China (only 6 cases).
Some studies [10, 11, 14, 15] reported that MLL-r positive compared to MLL-r negative children had a higher proportion of patients with poor early treatment response in addition to more risk factors at initial diagnosis. In the current study, we also found that the proportion of M1 patients on Day 15 of induction chemotherapy in MLL-r positive children with ALL was significantly lower than that in the negative group, and the proportion of MRD positive patients on Day 33 was significantly higher than that in the negative group, suggesting that the poor early treatment response in MLL-r positive patients may be due to the dominant clone of MLL-r positive leukemia cells being insensitive or resistant to chemotherapy at the time of initial diagnosis, rather than chemotherapy “selected” for the inferior clone [16, 17]. However, after one course of standardized induction chemotherapy, the percentage of MLL-r positive patients with CR of BM could reach 96.8%, which is similar to the 97.6% CR rate reported in China , which might be related to the early strong chemotherapy.
In this present study, only 13 of 31 MLL-r positive ALL cases were successfully detected for BM karyotype, and 8 cases were normal karyotype, and the CCA compliance rate was only 38.5%, which may be due to the fact that most of the chromosomes in MLL-r positive children belong to normal. On the other hand, it may also be due to the small 11q23 broken fragment and the influence of the quality of the split phase and the cryptic chromosomal translocation, so there was a missed detection . FISH technique can detect basically all cases with MLL-r positive and had higher resolution, stronger sensitivity and shorter cycle time, but it cannot detect MLL partner genes and MLL-PTD due to the limitation of probes. Combined with PCR method, our department can detect MLL-AF4, MLL-AF6, MLL-AF9, MLL- AF10, MLL-ELL, MLL-ENL, MLL-AF1q, MLL-AF17, and MLL-PTD, which are common MLL-associated fusion genes, thus compensating for the inadequacy of FISH methods . Thus, we suggested the clinical use of a combination of the three above-mentioned assays, which will help to improve the detection rate of MLL-r positive patients.
Meyer et al.  analyzed 1420 MLL-r positive children with ALL, and the most common partner genes were MLL-AF4 (57%), MLL-ENL (18%), MLL-AF9 (13%), MLL-AF10 (4%), MLL-EPS15 (2%), and MLL-AF6 (2%). In China, the statistical results of Sun et al  on 57 cases of 11q/23 MLL-r pediatric patients with ALL showed that the MLL-AF4 accounted for 29.8% and MLL-PTD accounted for 26.3%, followed by MLL-AF9 (22.8%), MLL-ENL (12.3%) and MLL-AF10 (8.8%), respectively. In our study, MLL-PTD (29%), MLL-AF4 (25.8%), and MLL-ENL (16.1%) were predominant among the 31 MLL-r positive cases, which were consistent with that reported by Sun Yulan et al. and different from that reported by Meyer et al. We speculated that it might be due to the racial differences. Although the detection rate of MLL-r positive ALL children was inconsistent among different studies, the most common partner gene was still MLL-AF4. Some clinical reports  showed that MLL partner genes were the main determinants of leukemia phenotype, and MLL-AF4 was mainly associated with lymphoid malignancies, while MLL-AF9 was more likely to cause myeloid malignancies, which might may explain why MLL-AF4 was most common in ALL.
It has been reported in the literature [7, 20] that WBC and MLL-r positive at disease onset were important factors affecting the prognosis of children with ALL, and the results of the multivariate analysis of 124 children with ALL in our study also showed that WBC ≥50 × 109/L was an independent risk factor affecting the outcome of patients with ALL, which was consistent with the literature. However, our study also observed that MLL-r positive was not a prognostic factor for patients with ALL, but its partner gene MLL-AF4 was an independent risk factor for the prognosis of children with ALL, suggesting that the prognosis of our children with ALL may be closely related to the MLL-r partner gene type.
There was a consensus that the overall prognosis of MLL gene rearrangement leukemia is poor, and a large pediatric leukemia collaborative group [8, 20] has shown that the 5-year EFS and OS of MLL-r positive children with ALL were 60–65% and 68–74%, respectively, while our K-M survival analysis showed that the EFS of MLL-r positive children were 56.01 ± 16.89% and OS were 73.32 ± 16.6%, which were significantly lower than the MLL-r negative group and basically close to those reported by the collaborative group, and both were lower than the 5-year EFS (72–80%) and OS (83–85%) levels in domestic patients with ALL, which laterally confirmed that concomitant MLL gene rearrangement positive was a more malignant type of childhood ALL. Therefore, this study further analyzed the prognostic factors affecting MLL-r positive ALL, and the results of Cox model multivariate analysis showed that T-cell phenotype, WBC ≥50 × 109/L and D15 MRD positive were independent risk factors affecting MLL-r positive children, which were consistent with the findings of Tomizawa et al  With the increasing maturity of conventional chemotherapy, most children with MLL-r ALL were able to achieve CR with treatment, and the mainstream treatment regimen was still chemotherapy. Our survival analysis of patients with ALL who received only chemotherapy showed that the 10-year EFS and OS of MLL-r positive children who received only chemotherapy were lower than those of the negative group, indicating that although children could enter remission with conventional chemotherapy, some children still experienced relapse after remission, resulting in a significant decrease in EFS, and although the relapse rate of children with MLL-r associated leukemia could be reduced by increasing the intensity of chemotherapy, the treatment-related relapse rate was significantly lower. Although the relapse rate of children with MLL-r associated leukemia could be reduced by increasing the intensity of chemotherapy, the OS due to treatment-related mortality and infection-based complications decreased accordingly. This suggested that chemotherapy-only may be less effective in MLL-r positive cases.
The role of SCT in the treatment of MLL-r-positive leukemia has been controversial. The results showed that the 5-year EFS of 53 children treated with SCT and 47 children treated with chemotherapy alone were 48.8 and 48.7%, respectively, with no statistical difference (P = 0.6), indicating that SCT for MLL-r-positive ALL did not show any advantage. ALL did not show any advantage . In contrast, a recent study by the Japanese Pediatric Leukemia/Lymphoma Collaborative Group  of 43 MLL-r-positive high-risk children (age < 6 months and/or CNS leukemia) who received SCT showed a 3-year EFS and OS of 56.8 and 80.2%, respectively, demonstrating a good prognosis, suggesting that SCT has more therapeutic advantages over conventional chemotherapy. The 10-year EFS rate and OS of children who received SCT in this study were 100%, which were higher than those who received chemotherapy alone at 54.32 ± 16.89% and 72.19 ± 16.88%, respectively, and SCT seemed to be a good independent prognostic factor affecting MLL-r-positive children. However, due to the small sample size, we cannot conclude that SCT can effectively treat MLL-r positive patients.
It has been suggested that alterations in MLL-r play an important role in the activation of oncogenes, while the role of partner genes fused to them is unclear, while some studies revealed that ALL with MLL-r positive were similar in most morphological and histochemical features, and childhood ALL with MLL-r positive, regardless of the type of partner gene, had an extremely poor prognosis . Previous studies have shown that MLL-AF4 fusions in ALL were associated with poorer survival . A large (n = 756), multicenter, retrospective Nordic study analyzing the prognosis of various types of MLL fusion gene leukemia showed that MLL-AF4 and MLL-AF6 had a very poor prognosis with 10-year EFS of 29 and 11% and 10-year OS of 27 and 22%, respectively, while MLL-AF9 had 10-year EFS and OS of 50 and 63%, with a relatively good prognosis . Previous studies [10, 11, 19] had also shown that MLL-AF4 had a worse prognosis than non-MLL-AF4 partner genes, and the above findings strongly suggested that different partner genes had distinct effects on the prognosis of patients with MLL-r positive leukemia. In the present study, we found that partner genes such as MLL-AF4, MLL-PTD and MLL-ENL were more common in MLL-r positive ALL, and further comparison of EFS and OS among the three groups showed statistically significant differences, indicating that MLL-AF4 positive ALL had the worst prognosis, and the results also indicated that the MLL fusion gene type (MLL-AF4) as an independent risk prognostic factor. Inconsistent with other reports in the literature, the most reported partner gene in our study was MLL-PTD, which had a 10-year EFS and OS of 85.71 ± 22.37%, showing a good prognosis, suggesting that MLL-PTD may be an indicator of relatively good prognosis, but most of the literature has not analyzed MLL-PTD, and the limited number of cases included in this study was not enough to reveal this, and the overall prognosis of children in this group deserves our further attention. We conducted long-term follow-up of 124 children with ALL included in this study, and eventually 5 children relapsed, with a significantly higher proportion of MLL-r positive children relapsing than in the negative group, and all of them relapsed in bone marrow alone and had high-risk factors for poor prognosis such as WBC ≥ 50 × 109/L, T-cell phenotype or positive D15 MRD, consistent with the literature [10,11,12,13,14,15].
Based on these findings, we believe that MLL-r positive should not be uniformly classified as a high-risk group in clinical practice, but that screening for these MLL partner genes is needed for accurate risk stratification at diagnosis: those positive for MLL-AF4 can be treated as a high-risk group, while other MLL partner genes need to be specifically combined with other prognostic factors (T-cell phenotype, WBC ≥ 50 × 109/L and D15 MRD positive) for a comprehensive evaluation. Moreover, it seemed that the relapse time for the MLL-r positive group is longer than MLL-r negative group (9.6 vs 3.9). We implied that the reason for the late relapse of MLL-r positive children was that the combination of drug chemotherapy will not kill the MLL gene clone formed in the fetal period of ALL children. The clone will undergo secondary transformation after treatment, which will eventually lead to the late relapse of ALL [12,13,14,15].
In conclusion, the remission rate of MLL-r positive ALL in children was moderate, but prone to relapse with low overall survival, and poor prognosis for those treated with chemotherapy-only. The prognosis of children with MLL-r positive ALL were closely related to the type of MLL-r, and clinical attention should be paid to screening for MLL partner genes and combining them with other prognostic factors for accurate risk stratification.