Acute lymphoblastic leukemia (ALL) is a neoplastic disorder of lymphoid progenitor cells characterized by diverse cytogenetic and molecular abnormalities with peaks of prevalence for 2–5-year-old patients and those older than 50 [
Common cytogenetic subtypes abnormalities such as TEL-AML1 (ETV6-RUNX1)/t(12;21), BCR-ABL (BCR-ABL1)/t(9;22), MLL rearrangements/t(11q23), E2A-PBX1 (TCF3-PBX1)/t(1;19), MYC-IGH/[t(8;14), t(2;8) or t(8;22)], and hyperdiploidy/(50 chromosomes) are typically associated with ALL prognosis and together with age, white blood cell (WBC) count, leukemic cell immunophenotype, and time to respond to therapy define the risk stratification group [
Attending the need to improve relapse prediction, the aim of this study was the longitudinal validation of a comprehensive gene expression profile of 45 candidate genes involved in key cell signaling pathways and to determine whether their expression was associated with relapse in a group of Mexican B-ALL patients. In this panel, genes and pathways previously reported as altered by expression or copy number variations (CNV), known targets of sequence mutations in ALL, genes commonly mutated in other cancers, and ALL nonrelated genes involved in key signaling pathways were studied [
The patients included in this study were newly diagnosed with ALL by cytological examination, peripheral blood, or bone marrow aspirates analysis, and the B lineage lymphocytic precursor was determined by flow cytometry (see Table
General characteristics of 73 B-ALL patients.
Characteristic | Number of cases (%) | Mean (range) |
---|---|---|
Sex | ||
Male | 44 (60.3) | — |
Female | 29 (39.7) | |
Age | ||
0 to 18 | 55 (75.3) | 7.9 (1–18) |
Older than 19 | 18 (24.7) | 28.7 (19–44) |
MIC classification | ||
Early precursor B | 2 (2.7) | — |
ALL common | 22 (30.1) | |
ALL pre-B | 24 (32.9) | |
ALL B | 25 (34.3) | |
Diagnosis risk | ||
High | 39 (53.4) | — |
Regular | 34 (46.6) | |
White-cell count (/mm3) | ||
<10,000 | 49 (67.1) | 3,676.4 (6.67–9.1 × 103) |
10,000–49,999 | 16 (21.9) | 22,025 (1 × 104–4.7 × 104) |
50,000–100,000 | 2 (2.8) | 78,755 (7.8 × 104–7.9 × 104) |
>100,000 | 6 (8.2) | 212,150 (1 × 105–3.2 × 105) |
Percentage of leukemic blasts cells at B-ALL diagnosis | 31 (42.5) | 75.4 (17.5–95.0) |
A total of 73 patients provided peripheral blood at the time of diagnosis and at one and two months after the B-ALL diagnosis. All the samples were collected between September 2005 and February 2008 and were followed for 3 years. A total of 225 samples were used in the study, including 219 leukemic (73 at diagnosis, 73 at first, and 73 at second month, resp.) and 6 healthy peripheral blood controls. A complete blood count (CBC) was made for each patient and a molecular screening of BCR-ABL translocation was assessed by quantitative real-time polymerase chain reaction (qRT-PCR) according to standardized protocols of the Unidad de Diagnóstico Molecular del Departamento de Bioquimica y Medicina Molecular de la Facultad de Medicina de la UANL, Mexico.
Mononuclear cells were purified by Histopaque-1077 (Sigma-Aldrich, St. Louis, MO) and preserved in RNAlater Solution (Life Technologies, Carlsbad, CA) at −20°C. Total RNA was extracted from the mononuclear cells using Qiagen RNeasy Mini Kit (Qiagen, West Sussex, UK) according to the manufacturer’s protocol. RNA concentration and its quality were determined by measurement of the optical density at 260 nm and relation 260/280, respectively, using NanoDrop Spectrophotometer (NanoDrop Technologies, Wilmington, DE). Samples with a relation of 260/280 > 1.8 were used in the next steps of the analysis. RNA was stored at −80°C until use. cDNA was synthesized from 1.0
A total of 45 genes representing key signaling pathways and functional processes were selected for this study (Additional Table 2); 32 of them had previously been related to ALL (a list of references of these studies may be consulted in Additional Table 2). In addition, two reference genes RPL13A and HPRT1 were selected as internal quantitative controls. Gene specific primers for real-time qPCR assay were selected from the Primer Bank database (
The qRT-PCR analyses were performed with an ABI 7900 high throughput real-time PCR System (Applied Biosystems, Foster City, CA) in a total volume of 20
Simple comparisons of normalized ΔCq values (square of ΔCq values) between B-ALL cases and healthy controls were determined by Student’s
The study group included 73 B-ALL cases; 75.3% (55) were children (age range: 0 to 18 years) and 24.7% (18) adults (age range: 19 to 44 years). Morphologic, immunologic, and cytogenetic (MIC) working classification grouped 90% of the cases in the three most common B-ALL diagnoses: ALL common, Pre-B-ALL, and B-ALL, respectively (Table
Additional Table 3 shows the normalized ΔCq values obtained for the study groups. Out of 45 analyzed gene expressions, normalized ΔCq values of 23 genes at diagnosis, 23 at one month, and 28 at second month after starting treatment showed statistical differences between B-ALL cases and healthy controls (
Gene expression classifier for B-ALL diagnosis. ROC curves show the independent accuracy of the 5-probe-set classifier B-ALL diagnosis. (a) ROC curve for IL10, (b) ROC curve for WNT5A, (c) ROC curve for OPAL1, (d) ROC curve for CCND2, and (e) CASP8AP2 ROC curve, respectively.
Expression levels for the genes that showed ΔCq statistical differences at least one time between study groups were determined using healthy controls as calibrator (Additional Table 4). Additional Table 4 shows the descriptive statistics of expression level for 30 genes evaluated. In general, ΔCq differences observed between groups for IL10, WNT5A, OPAL1, CCND2, and CASP8AP genes reflected their overexpression with respect to the controls and the median of expression level for each gene was 2.2, 5.4, 7.7, 6.2, and 6.6, for IL10, WNT5A, OPAL1, CCND2, and CASP8AP, respectively.
The results of the identification of genes with statistical differences in their level of expression through time are displayed in Table
Differences in gene expression levels through time in 73 B-ALL patients.
Gene |
Times included in the comparison and significant |
||
---|---|---|---|
Time 1/time 2 | Time 1/time 3 | Time 2/time 3 | |
IL2RA | 0.011 | ns | ns |
IL10 | 0.012 | ns | ns |
CD10 | 0.016 | ns | ns |
CD3D | 0.018 | ns | ns |
HOXA9 | 0.041 | ns | ns |
PAX5 | 0.006 | ns | ns |
CTNNB | ns | 0.039 | ns |
GSK3B | ns | 0.003 | ns |
JAK2 | ns | 0.032 | ns |
FOS1 | ns | 0.007 | ns |
RAPGEF2 | ns | 0.044 | ns |
SORT1 | ns | 0.023 | ns |
HK2 | ns | 0.007 | ns |
PAX5 | ns | 0.002 | ns |
DEFA1 | ns | 0.011 | ns |
NOTCH1 | ns | 0.026 | ns |
S100A8 | ns | 0.002 | ns |
ns | ns | 0.037 | |
BCL11A | 0.023 | ns | ns |
ns | 0.012 | ns | |
STAT5 | 0.017 | ns | ns |
ns | 0.012 | ns | |
OPAL1 | 0.04 | ns | ns |
ns | 0.018 | ns | |
FLT3 | <0.001 | ns | ns |
ns | 0.009 | ns | |
RXRA | 0.01 | ns | ns |
ns | 0.002 | ns | |
BCL2A | 0.005 | ns | ns |
ns | <0.001 | ns | |
ns | ns | 0.005 |
ns: nonsignificant,
Positive associations between gene expression level and relapse/death.
Classification of study population | Gene | Time 1 | Time 2 | Time 3 | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Event- free group | Event group |
|
Event-free group | Event group |
|
Event- free group | Event group |
|
||
All the ALL cases | FLT3 | −0.049 | 0.434 | 0.022* | −0.598 | −0.075 | 0.028* | −0.242 | 0.023 | 0.02* |
IL2RA | 1.164 | 1.493 | 0.038* | — | — | — | — | — | — | |
DEFA1 | — | — | — | — | — | — | 1.99 | 4.012 | 0.01* | |
|
||||||||||
Children | FLT3 | — | — | — | −0.602 | 0.025 | 0.044* | — | — | — |
IL2RA | 1.193 | 1.558 | 0.038* | — | — | — | — | — | — | |
|
||||||||||
Adults | FLT3 | — | — | — | — | — | — | −0.49 | 0.03 | 0.046* |
SORT1 | — | — | — | — | — | — | 3.84 | 5.34 | 0.039* | |
|
||||||||||
Standard risk | FLT3 | — | — | — | −0.707 | −0.025 | 0.008* | — | — | — |
|
||||||||||
High risk | DEFA1 | — | — | — | — | — | — | 2.33 | 4.62 | 0.018* |
To determine the association between expression of IL2RA, SORT1, DEFA1, and FLT3 genes and B-ALL survival, the expression level cutoff values in which the sum of specificity and sensibility were the highest (nearest two) were determined for each gene by ROC analysis. In this test, the B-ALL cases were grouped as event-free patients and those that presented the event (relapse/death). Calculated cutoff values were used to construct and compare survival curves using B-ALL patients whose expression level was above the cutoff and B-ALL cases with expression levels equal to or below the calculated cutoff. The results obtained are shown in Figure
Survival analysis and OR values for differentially expressed genes between B-ALL relapsed patients and event-free cases.
Gene-time |
Group± |
|
Survival time |
Standard error | 95% CI |
|
OR | 95% CI |
|
---|---|---|---|---|---|---|---|---|---|
FLT3-T1 |
1 | 38/7 | 31.13 | 1.803 | 27.6–34.7 | 0.012* | 3.73 | 1.3–10.72 | 0.024* |
2 | 35/16 | 25.003 | 2.679 | 19.7–30.3 | |||||
|
|||||||||
IL2RA-T1 |
1 | 38/7 | 31.05 | 1.829 | 27.5–34.6 | 0.012* | 3.73 | 1.3–10.72 | 0.024* |
2 | 35/16 | 24.71 | 2.745 | 19.3–30.1 | |||||
|
|||||||||
FLT3-T2 |
1 | 32/4 | 31.47 | 1.901 | 27.7–35.2 | 0.004* | 6.05 | 1.8–20.36 | 0.005* |
2 | 41/19 | 25.19 | 2.398 | 20.5–29.9 | |||||
|
|||||||||
FLT3-T3 |
1 | 35/5 | 34.83 | 1.928 | 31.1–38.6 | 0.002* | 5.4 | 1.7–16.9 | 0.005* |
2 | 38/18 | 23.02 | 2.309 | 18.5–27.6 | |||||
|
|||||||||
DEFA1-T3 |
1 | 43/8 | 30.84 | 1.775 | 27.4–34.3 | 0.006* | 4.38 | 1.5–12.5 | 0.01* |
2 | 30/15 | 24.13 | 2.825 | 18.6–29.7 | |||||
|
|||||||||
IL2RA-T1-children |
1 | 25/3 | 32.76 | 2.153 | 28.5–37.0 | 0.062 | 4.03 | 0.98–16.6 | 0.088 |
2 | 30/10 | 28.57 | 2.857 | 23.0–34.2 | |||||
|
|||||||||
FLT3-T2-children |
1 | 24/1 | 33.83 | — | — | 0.003* | 14.53 | 1.7–122.1 | 0.008* |
2 | 31/12 | 27.04 | 2.844 | 21.5–32.6 | |||||
|
|||||||||
FLT3-T3-adults |
1 | 9/3 | 24.87 | 4.128 | 16.7–33.0 | 0.038* | 27 | 2.0–368.4 | 0.013* |
2 | 9/7 | 11.44 | 1.061 | 9.4–13.5 | |||||
|
|||||||||
SORT1-T3-Adults |
1 | 7/0 | — | — | — | <0.001* | 14 | 1.1–172.6 | 0.05 |
2 | 10/10 | 11 | 1.193 | 8.7–13.3 | |||||
|
|||||||||
FLT3-T2-standard risk |
1 | 18/0 | — | — | — | <0.001* | Undetermined | — | <0.001* |
2 | 16/8 | 22.51 | 3.625 | 15.4–29.6 | |||||
|
|||||||||
DEFA1-T3-high risk |
1 | 26/5 | 27.71 | 2.780 | 22.3–33.2 | 0.081 | 4.0 | 1.02–15.72 | 0.089 |
2 | 28/10 | 22.79 | 3.627 | 15.7–29.9 |
*Statistical significance.
Log-Rank Survival Test. The comparison between survival curves according to specific cutoff values for genes differentially expressed in relapse/death versus event-free cases is showed without classification for (a) FLT3 time 1, (b) IL2RA time 1, (c) FLT3 time 2, (d) FLT3 time 3, and (e) DEFA1 time 3, respectively. Considering B-ALL subgroups, figure shows in children (f) IL2RA time 1 and (g) FLT3 time 2, respectively, and in adults (h) FLT3 time 3 and (i) SORT1 time 3. In mild risk (j) FLT3 time 2 and finally survival curve for (k) DEFA1 time 3 in high risk subgroup, respectively.
After 2 months of starting treatment and compared to B-ALL cases with expression levels below or equal to 3.041, patients with expression level for DEFA1 > 3.041 had increased odds of the event by 4.38 times (95% CI: 1.5–12.5;
To evaluate the usefulness of IL2RA, SORT1, FLT3, and DEFA1 expression as independent predictors of B-ALL outcome a multivariate logistic regression using EFS/RFS as dependent variable was carried out. In the statistical modelling, WBC count, age, gender, age group (children/adult), B-ALL immunophenotype, and SORT-T1, DEFA1-T1, FLT3-T1-3, and IL2RA-T1 expression levels were included as independent variables. The significant B-ALL outcome predictors were WBC count (
With the aim to test if clinical data including differences in the counts of leukemic blasts cells recovered across samples had an impact on the expression levels of genes associated with survival, a correlation analysis was performed. In this analysis age, B-ALL diagnosis WBC counts, and diagnosis blasts data of 31 participants were included and their correlations with FLT3-T1, SORT-T1, and DEFA1-T1 expression levels were evaluated. Significant correlations between age, WBC, and counts of blasts cells with FLT3-T1, SORT-T1, or DEFA1-T1 expression levels were not observed (
Whilst gene expression profiling studies in ALL have identified gene expression signatures associated with recurrent cytogenetic abnormalities and
Normalized ΔCq values for 19 genes showed consistent differences between cases and controls at least until the second month of follow-up. Interestingly, observed differences were independent of the treatment given. IL10, WNT5A, OPAL1, CCND2, and CASP8AP2 normalized ΔCq presented the most marked differences with respect to the controls; WNT5A was the gene with the best sensitivity/specificity trade-off and therefore the best potential B-ALL diagnosis biomarker evaluated in this study. These findings are relevant because this set of 23 genes may be considered as a B-ALL related signature and their unknown role in the leukemogenesis process should be addressed in future studies. Compared to controls, normalized ΔCq differences of IL10, WNT5A, OPAL1, CCND2, and CASP8AP2 between the study groups were translated in an overexpression of these genes; a high expression of IL10, OPAL1, and CCND2 has been previously reported [
Comparisons of gene expression levels through time in B-ALL patients showed consistent differences among a set of 23 genes at least at two of the three time points evaluated and the differences in the expression levels of IL2RA, SORT1, DEFA1, and FLT3 genes in at least one of the times studied were associated with relapse and/or B-ALL-related death. Interestingly, the overexpression of FLT3 at B-ALL diagnosis and at one and two months after starting treatment was associated with relapse/death in our study group; these findings were established without regard to subclassification by clinical characteristics and were significant in children (time 2), adults (time 3), and standard risk (time 2) subgroups. IL2RA overexpression at the time of B-ALL diagnosis was significant in the children’s group. In the same sense, while DEFA1 overexpression was significant for all B-ALL cases (time 3) and in the high risk subgroup (time 3), SORT1 overexpression was only different in adults (time 3). Previous results indicate that differences through time in specific gene expression profiles may be useful to identify genes with capacity to pinpoint B-ALL cases with a higher potential risk to relapse.
Grouping the B-ALL cases as event-free patients and those that presented the event (relapse/death), we calculated level expression cutoff values for IL2RA, SORT1, DEFA1, and FLT3 by ROC analysis. These cutoff values allowed us to construct and compare survival curves (and calculate their respective ORs) for groups of B-ALL patients whose expression level was above the calculated cutoff and for patients with gene expression levels equal to or below their respective cutoff. In our study patients with expression values for IL2RA, SORT1, FLT3, and DEFA1 genes above the cutoff values have an increased risk of relapse/death ranging from 3.73 to 27 times compared to B-ALL patients with expression values equal to or below these cutoffs (
It is important to mention that in this study the presence of BCR-ABL rearrangement was evaluated in the B-ALL cases; only three patients (4.1%) were positive for this molecular abnormality. Considering that the number of patients with this translocation was low and that the screening of other molecular abnormalities was not included in the characterization of the cases, we decided not to remove them from the analysis; however, despite this limitation, the results obtained (without molecular abnormalities classification) showed an important association with B-ALL prognosis, and the expression of three genes demonstrated their usefulness to identify risk of relapse/death of patients with B-ALL (without incorporating additional clinical classification) and, considering that abnormalities in chromosome number or structure are present in 70–90% of the B-ALL cases [
We have shown that kinetics of early treatment response evaluated by molecular methods from B-ALL diagnosis at first and/or 2 months after starting the therapy are highly associated with the prognosis of B-ALL patients. Clinical application of these prognosis markers might allow the prospective identification of a significant subgroup of ALL patients with little chance for a cure if treated with contemporary chemotherapeutic regimens, potentially indicating the need for an early stem cell transplant or biologically targeted treatment. Further analysis of these expression profiles coupled with additional comprehensive genomic studies will hopefully lead to the successful identification of novel therapeutic targets and more effective therapies for these patients.
The kinetics of early treatment response assessed by molecular measures at one and two months after the beginning of therapy is highly associated with leukemia prognosis. Overexpression of FLT3 and DEFA1 genes retained independent prognostic significance for B-ALL outcome, reflected as increased risks of relapse/death among the study population.
Acute lymphoblastic leukemia
Chronic lymphoblastic leukemia
Acute myeloblastic leukemia
Chronic myeloblastic leukemia
Quantitative real-time polymerase chain reaction
Quantification cycle
Peripheral blood mononuclear cells
Receiver operating characteristic curve
Area under the curve
Event-free survival
Relapse-free survival.
The authors declare that they have no competing interests.
Conception and design of the study are done by Rocio Ortiz-Lopez and Idalia Garza-Veloz; acquisition of data is done by Idalia Garza-Veloz, Margarita L. Martinez-Fierro, Karol Carrillo-Sanchez, and Angel Lugo-Trampe; analysis and interpretation of data are carried out by Margarita L. Martinez-Fierro, Idalia Garza-Veloz, Maria Guadalupe Ramos-Del Hoyo, and Jose Carlos Jaime-Perez; drafting the paper and critical review is done by Margarita L. Martinez-Fierro, Idalia Garza-Veloz, Jose Carlos Jaime-Perez, Maria Guadalupe Ramos-Del Hoyo, Angel Lugo-Trampe, David Gomez-Almaguer, Cesar Homero Gutierrez-Aguirre, Oscar Gonzalez-Llano, Rosario Salazar-Riojas, Alfredo Hidalgo-Miranda, Augusto Rojas-Martinez, and Rocio Ortiz-Lopez; final approval of the paper is done by Idalia Garza-Veloz, Margarita L. Martinez-Fierro, Jose Carlos Jaime-Perez, Karol Carrillo-Sanchez, Maria Guadalupe Ramos-Del Hoyo, Angel Lugo-Trampe, Augusto Rojas-Martinez, Cesar Homero Gutierrez-Aguirre, Oscar Gonzalez-Llano, Rosario Salazar-Riojas, Alfredo Hidalgo-Miranda, David Gomez-Almaguer, and Rocio Ortiz-Lopez. Idalia Garza-Veloz and Margarita L. Martinez-Fierro equally contributed to this work.
The authors gratefully thank all the participants of this protocol. This work was supported by the CONACYT-SALUD Grant no. SALUD-2004-C02-26.