Seventy-four consecutive zero- to 19-year-old patients with a diagnosis of ALL were investigated. Patients were identified in three leading institutions in Belo Horizonte, Minas Gerais, Brazil: Hospital das Clínicas da Universidade Federal de Minas Gerais (UFMG) (n=52); Hospital da Baleia/Fundação Benjamin Guimarães (n=14); and Santa Casa de Misericórdia de Belo Horizonte (n=8). Bone marrow samples were collected from January 2010 to December 2012. Most patients (n=45) were treated according to the Grupo Brasileiro de Tratamento da Leucemia Infantil-leucemia linfoide aguda (GBTLI-LLA-99) protocol although 21 patients were treated according to GBTLI-LLA-2009, and eight patients using the Associazione Italiana Ematologia ed Oncologia Pediatrica (AIEOP-95) protocol. According to the GBTLI-LLA-99 and 2009 protocols, patients older than nine years at diagnosis and/or with a white blood cell (WBC) count above 50×109/L were assigned to the group with high risk of relapse and the others were assigned to the low-risk group and received less intensive treatment. Approval was obtained from the Ethics Committee of the three participating institutions and all guardians and/or patients gave their informed written consent to participate in the study according to the Declaration of Helsinki. No family or patient refused to sign the informed consent form.

Diagnosis of ALL was made by standard morphological analysis and by flow cytometry immunophenotyping. Karyotype analysis and reverse transcription PCR (RT-PCR) were also performed at diagnosis for the BCR-ABL, TCF3/PBX1, ETV6/RUNX1 and MLL-AF4 fusion genes.

Bone marrow samples were obtained from the patients at diagnosis (Day 0) and at the end of the induction period (Day 28 for those treated according to GBTLI-LLA-99 protocol; Day 33 for AIEOP-95; and Day 35 for GBTLI-LLA-2009).

Mononuclear cells were separated using Histopaque® (Sigma–Aldrich, Saint Louis, USA) centrifugation gradient and DNA was extracted with the NucleoSpin® Tissue Kit (Macherey-Nagel, Düren, Germany) according to manufacturer's instructions. The extracted DNA was quantified using the NanoDrop 2000™ Spectrophotometer. The quality of DNA was confirmed through amplification of the Fms-like tyrosine kinase 3 (FLT3) gene, according to Meshinchi et al.10

DNA from diagnostic samples was screened using 19 primer mixes according to the ALL subtype. For precursor B-cell acute lymphoblastic leukemia (pB-ALL), BIOMED2 primer sets for the complete and incomplete IgH (VH-(DH)-JH, DH-JH), IgK (Vk-Kde, Intron-Kde), TCRG (Vg-Jg1.3/2.3+Jg1.1/2.1) and incomplete TCRD gene rearrangements (Vd2-Dd3, Dd2-Dd3) were used.11 For T-ALL, BIOMED2 primer sets for the IgH (DH-JH), TCRG (Vg-Jg1.3/2.3+Jg1.1/2.1), TCRD (Vd-(Dd)-Jd1, Dd2-Jd1, Vd2-Dd3, Dd2-Dd3) gene rearrangements and for the Sil-Tal (Sil-Tal1, Sil-Tal2) microdeletion were used.11,12

PCR was carried out in 25μL reactions containing 25ng of DNA, 1U of Tth DNA polymerase (Biotools, Madrid, Spain), 10pmol of each primer, 2mM of MgCl2, and 100μM of each dNTP. The PCR amplification cycles have been previously described.11,12 Two negative controls were used in each PCR assay: one without DNA and the other containing pools of polyclonal DNA obtained from peripheral blood mononuclear cells (PBL) from ten healthy donors. PCR products were analyzed by homo-heteroduplex analysis on 12% acrylamide gels stained with Sybr Safe DNA gel stain (Invitrogen, USA), as previously described.13 Amplified gene rearrangements were characterized as clonal when a band of the expected size was visible,11 and not present in the PBL control. The band containing the clonal amplicon, according to the expected molecular size, was cut from the gel, dissolved in water and stored at −20°C for subsequent sequencing.

For MRD monitoring by the qualitative method, at least two clonal markers identified at diagnosis were tested, whenever possible. PCRs and homo-heteroduplex analyses were carried out as described above, except that 500ng of DNA were used. Day 28–35 samples, diagnostic DNA samples, as well as the polyclonal PBL DNA and the non-template controls were run in parallel. Follow-up samples were considered positive when they showed the same migration pattern and molecular weight as the samples at diagnosis.

Clonal PCR products from Day 0 that had been dissolved in water were re-amplified in a volume of 50μL using the same primer sets (but with T7 or M13 extensions) and reaction conditions as described above. Sequencing reactions were carried out using the BigDye Terminator Cycle Sequencing Reaction Kit (PE Applied Biosystems) and T7 and M13 primers. Sequences were run using the ABI Prism 3130 Genetic Analyzer (PE Applied Biosystems) and analyzed using the Chromas Lite 2.4 software (Technelysium Pty Ltd.). Patient-specific junctional region sequences were identified with the Blast tool (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and IMGT/V-QUEST (http://www.imgt.org/IMGT_vquest/share/textes/).

The Primer3 Biotools software (http://biotools.umassmed.edu/bioapps/primer3_www.cgi) was used to design patient-specific primers complementary to the junctional region sequence and compatible with primers and probes previously described for IgH,14IgK,15TCRG,16TCRD,17 and Sil-Tal.18 Two patient-specific primers were designed for each Ig/TCR region. GC rich (>80%) junctional regions were not used as targets.

Patient-specific primers were tested for specificity and sensitivity. RQ-PCR analysis was performed in duplicate, in a final volume of 25μL containing 100ng of DNA, 5μM sequence-specific TaqMan probes (Applied Biosystems), 7.5pmol of each primer, and TaqMan Universal Master mix (2×) (Applied Biosystems), on the StepOnePlus™ RQ-PCR System (Applied Biosystems). Results were analyzed with the StepOne software v2.3 and the sensitivity was defined as the point with the greatest dilution in which the cycle threshold (Ct) reached at least one Ct below the lowest Ct for polyclonal PBL. The primer was considered more specific the greater the difference between the sensitivity of Ct and that of PBL Ct.

RQ-PCR MRD analysis of Day 28–35 samples was performed and interpreted according to the guidelines of van der Velden et al.19 RQ-PCRs were carried out as described above for sensitivity tests, except that 500ng of DNA were used. Results were normalized using N-RAS as a control gene.18 MRD cut-off points were defined according to the GBTLI-2009 protocol, in which patients with results above 1×10−3 for pB-ALL and 1×10−2 for T-ALL are considered positive and classified as poor responders at the end of the induction therapy.

Overall survival (OS), event-free survival (EFS) and leukemia-free survival (LFS) curves were plotted employing the Kaplan–Meier method. The OS was calculated from the date of diagnosis to the date of death or last follow-up. The EFS was calculated from the date of diagnosis to the date of relapse or death. The LFS was calculated from the date of leukemia remission to the date of relapse (patients who died in remission was censored on the date of death). The cut-off date for censoring non-relapsed patients was 22 October 2014. Curves for different MRD groups were compared by the log-rank test according to age, WBC count at diagnosis, immunophenotype, risk group, gender, institution of origin and treatment protocol. All statistical analyses were performed using the Statistical Program for the Social Sciences (SPSS) software (version 20.0) with the level of significance set for p-values≤0.05. The association between LFS, qualitative MRD and RQ-PCR MRD results were adjusted for the effect of clinical and biological categorical variables in a multivariate Cox model.

At the end of the induction therapy, 9/57 patients (15.8%) had positive MRD by the qualitative assay, 12.8% (6/47) for pB-ALL and 30% (3/10) for T-ALL. Eight out of the nine MRD-positive patients had been assigned to the high-risk group at diagnosis; one of them was BCR-ABL-positive and another was MLL-AF4-positive. Sensitivity of the assay with the qualitative primers was 10−2 to 10−3.13

MRD analysis by RQ-PCR was performed in 14/44 (32%) patients using two markers and in 30 (68%) using just one marker. Twenty-seven IgH, 12 IgK, five TCRG and 11 TCRD rearrangements were used. Eleven out of the 44 patients (25%) had positive MRD at the cut-off level of 10−3 (pB-ALL) or 10−2 (T-ALL); 10/39 (25.6%) pB-ALL patients and 1/5 (20%) T-ALL patients. Eight out of the 11 had been assigned to the high-risk group at diagnosis; two of them were BCR-ABL-positive and another was MLL-AF4-positive. The observed sensitivity of the assay with the specific primers varied from 10−3 to 10−5.

According to MRD RQ-PCR cut-off points established by the GBTLI-2009 protocol, the agreement between the two methods using Kappa statistics was 40% for pB-ALL and 100% for T-ALL. When a cut-off point of 1×10−2 was used for pB-ALL, the Kappa coefficient was 75%, and considering a cut-off point of 1×10−3 for T-ALL, it was 50% (Table 2). Most divergent results between assays were patients with MRD loads between 10−2 and 10−3, which were positive in RQ-PCR but negative in conventional PCR.

The estimated 3.5-year probabilities of OS and EFS were 73.6% and 68.2%, respectively, while the estimated 3.5-year probability of LFS was 72.3% (Figure 1).

Figure 1.

Kaplan–Meier survival curve estimates for (A) overall survival and (B) event-free survival of all patients (n=74); and (C) leukemia-free survival in 72 patients.

(0,13MB).

The median time of LFS for children without relapse was 3.0 years (1.1–4.5 years) from the date of morphological bone marrow remission. The median time from remission to relapse was 1.2 years (0.5–2.5 years). The 3.5-year LFS was significantly higher in qualitatively MRD-negative children (84.1±5.6%) when compared to MRD-positive children (41.7±17.3%; p-value=0.004) (Figure 2). There was no significant association between any other analyzed clinical or biological variables and LFS. Even different protocols had no impact on survival (data not shown).

Figure 2.

Leukemia-free survival of 57 children with acute lymphoblastic leukemia according to minimal residual disease based on qualitative polymerase chain reaction.

(0,1MB).

LFS data analysis for qualitative MRD results was repeated considering only patients evaluated by both techniques, qualitative PCR and RQ-PCR (n=44). Again, qualitative MRD-negative patients had significantly higher LFS than MRD-positive children (p-value=0.032; Supplemental figure 1).

Cox's regression model was used to assess the prognostic impact of qualitative MRD on LFS on Days 28–35. After adjusting for the effect of gender, institution of origin, treatment protocol, risk group, immunophenotype, WBC count at diagnosis and age in a multivariate analysis, MRD was the only variable significantly associated with LFS (p-value=0.015) (Table 3).

After excluding non-significant variables, positive MRD by qualitative PCR on Days 28–35 was significantly associated with a lower LFS (p-value=0.009). The relapse risk for positive MRD patients on Days 28–35 was 4.6 higher than for those with negative MRD (95% confidence interval: 1.5–14.6).

There was no significant association between RQ-PCR MRD and LFS (Figure 3). Analyzing the individual data, only one of six children (all with pB-ALL) with positive RQ-PCR MRD and negative qualitative MRD relapsed so far, after one year of remission. The remaining five are alive and without relapsing for 3.2–4.0 years since the initial remission.

Figure 3.

Leukemia-free survival according to minimal residual disease based on real-time quantitative polymerase chain reaction at the end of induction in 44 children with acute lymphoblastic leukemia.

(0,09MB).

Cox's regression model was used to determine the prognostic impact of RQ-PCR MRD on Days 28–35 on LFS. After adjusting for the effect of gender, institution of origin, treatment protocol, risk group, WBC count at diagnosis, age and immunophenotype in a multivariate analysis, no variable was statistically associated with LFS, including RQ-PCR MRD (p-value=0.427; Table 4).