Letter to the Editor | Published:

Validation of NG2 antigen in identifying BP-ALL patients with MLL rearrangements using qualitative and quantitative flow cytometry: a prospective study

Leukemia volume 22, pages 858861 (2008) | Download Citation

Timely diagnosis of mixed lineage leukemia (MLL) rearrangements in pediatric patients with B-cell precursor acute lymphoblastic leukemia (BP-ALL) is highly relevant in patient stratification. The presence of t(4;11)-MLL rearrangement is the most important criteria for high-risk stratification in protocols for childhood precursor BP-ALL. Current tests to identify t(4;11)-MLL involve karyotyping, fluorescence in situ hybridization and reverse transcription-PCR. The tests, however, are expensive both in terms of personnel and reagents. They do not provide a fast diagnosis and may fail due to technical reasons or high variability in the breakpoint that does not allow for standard reverse transcription-PCR detection. Therefore, a fast and reliable test that identifies MLL rearranged BP-ALL is highly welcome, particularly in low-income countries.

Advances in identification of leukemia subtypes have been recently achieved using gene expression and flow cytometry (FC) techniques. These approaches were shown to be particularly relevant in the characterization of patients with MLL rearrangements, a recurrent disease that accounts for 5% of all leukemic subtypes and more than 50% in infants. Armstrong et al.1 were able to separate patients with MLL translocations from other leukemic subtypes using microarray experiments, showing distinct gene expression profiles that characterize MLL rearranged specimens. Despite remarkable results in oncology research, microarray technology is currently an expensive test not yet introduced in clinical use for leukemia diagnosis.

Flow cytometry is a cheap and widely accepted diagnosis method that provides accurate identification of leukemia lineage and subclasses using a panel of antibodies. In a cohort of 533 patients with acute leukemia, Wuchter et al.2 applied a qualitative FC approach to successfully isolate MLL(+) samples using monoclonal antibody 7.1 which recognizes a cell surface chondroitin sulfate homolog of the rat NG2-surface molecule. De Zen et al.3 inspected the quantitative FC expression profiles (molecules of equivalent soluble fluorescein (MESF) values) of NG2 antigen as the main key marker for the presence of 11q23 rearrangements comparing 46 MLL+/ALL to 99 MLL−/ALL pediatric cases. De Zen et al.3 demonstrated high performances of NG2 antigen in characterizing MLL-positive samples using computational analysis on quantified molecules of equivalent soluable fluorescein (MESF) values for 16 marker proteins. Parting from these results, in this prospective study we want to validate the accuracy of qualitative (negative, low, bright signals) and quantitative (MESF values) FC methods using the most significant marker, NG2, as a single criterion for identification of MLL rearrangements. Here we tested the performance of NG2 expression comparing immunophenotypic data to cytogenetics, molecular biology and fluorescence in situ hybridization information as ‘gold standard’ (GS) references for detection of MLL translocations. In our study cohort of 1461 pediatric B-ALL patients, diagnosed from 1 January 2002 to 31 December 2005 (median age 5 years), 86 (5.89%) specimens were classified by immunophenotyping as pre-preB, 995 (68.11%) as CALL, 17 (1.16%) as pre-B/B, 321 (21.97%) as pre-B and 42 (2.87%) as B mature. Sixteen out of 33 (48.5%) MLL-positive infants were registered in the entire cohort. Table 1 summarizes the sample distribution within B-ALL phenotypes, including MLL diagnosis after GS response and the presence of CD10 antigen. All samples were enrolled in the AIEOP ALL-2000 protocol according to diagnostic guidelines. Diagnosis for MLL rearrangement was made by qualitative FC criteria using an immunological gate (CD19 vs side scatter) to identify all blast cells, by quantitative FC expression of NG2 antigen using MESF values and by GS data combining cytogenetic (Q-banding technique), fluorescence in situ hybridization and molecular biology. As previously reported4, a case was considered MLL positive by quantitative FC when it resulted in NG2 (PE) MESF value higher than 400 (±2 s.d.). NG2 (PE) expression values have been calculated by immunophenotyping analyses on 33 samples selected by combining two lists of patients: one list included all positive samples for NG2 marker according to qualitative FC criteria for suspicion of MLL gene rearrangements (n=30); the other group enrolled all positive cases according to GS criteria for diagnosis of 11q23 translocations (n=30) (Figure 1).

Table 1: Cohort of 1461 ALL patients distinguished by phenotype.
Figure 1
Figure 1

Comparison between flow cytometry (FC: QL=qualitative, QN=quantitative) and gold standard (GS) responses for mixed lineage leukemia (MLL) diagnosis.

Comparisons between qualitative and quantitative FC results have been represented in Figure 2. A case was considered concordant when qualitative FC showed a low or bright signal and quantitative FC MESF value was over 400 (QL.FC+/+QN.FC, case A) as well as when a negative signal was related to MESF value below 400 (QL.FC−/−QN.FC, case D). A case was considered discordant when low or bright signals were associated to MESF values below 400 (QL.FC+/−QN.FC, case B), while quantitative calculations were not feasible if no positive blast cells were detected (QL.FC−/+QN.FC, case C). As NG2 antigen is negative in MLL(−) samples, a very low amount of positive blast cells for this marker is sufficient to suspect the presence of MLL rearrangements by qualitative FC. In this case, the quantified geometric mean for NG2 antigen could not be affected (producing a negative value for quantitative FC) because the sample shows only few positive events (case B).

Figure 2
Figure 2

Four possible combinations between qualitative and quantitative flow cytometry (FC). G(m) represents the geometric mean for NG2 values. (a) NG2 positive for both QL and QN FC. (b) NG2 positive for QL, negative for QN FC. (c) Case not feasible. (d) NG2 negative for both QL and QN FC.

The performance of the two FC methods has been calculated (Table 2). Twenty-seven (90.0%) and 26 (86.7%) out of thirty cases were correctly diagnosed as true positive according to GS MLL identification using qualitative and quantitative FC, respectively (true positive (TP)). MESF analysis correctly identified 1430/1431 (99.93%) true negative cases while 1428/1431 (99.79%) showed no suspect of MLL rearrangement for qualitative FC (true negative (TN)). Three patients have been misclassified as MLL(+) by qualitative FC, while only one has been incorrectly detected by MESF calculation (false positive (FP)). Three and four negative samples for qualitative and quantitative FC methods, respectively, were shown to be positive after GS response (false negative (FN)). Qualitative FC identified a total number of six cases (0.41%) not concordant with GS responses while in total five samples (0.34%) were misclassified by quantitative FC. To test how well NG2 analyses match GS results, we calculated the sensitivity and specificity of both methods. The results show no significant differences between the two methods but indicate that qualitative FC slightly increases the identification of true MLL(+) patients (sensitivity: 0.90 vs 0.87) while quantitative FC recognizes MLL(−) samples better (specificity: 1.00 vs 0.99). Furthermore, (positive predictive value (PPV)), proportion of patients with positive FC results that are correctly diagnosed) and (negative predictive value (NPV)), proportion of patients with negative FC results that are correctly diagnosed) predictive values have been retrieved for each FC approach: no significant differences were shown between the two methods, but quantitative FC performed slightly better considering both PPV (0.96 vs 0.90) and NPV (1.00 vs 0.99) outcomes. The accuracy measure, a comprehensive value reflecting the proportion of correctly predicted MLL(+) and MLL(−) patients, indicates that both methods perform similarly in predicting MLL presence in ALL patients (QL.FC=0.99, QN.FC=1.00) (Table 2).

Table 2: Accuracy table for qualitative and quantitative FC according to GS results

MLL aberrations have been proved to share common features in different hematological diseases identifying NG2 antigen as an accurate marker for the presence of MLL rearrangements.2, 5, 6, 7 Attarbaschi et al.8 recently published a study, on behalf of the BFM cooperative study group, in which they report absolute sensitivity and 0.78 specificity for absence of CD10 expression (CD10−) in pro-B and pre-B ALL as phenotypic marker for predicting MLL rearrangements. Our results show that the absence of CD10 in BP-ALL patients with MLL rearrangements has very low specificity (28/86, 32.56%), while the presence of NG2 accounts for high sensitivity (87–90%) and specificity (99–100%). Furthermore, in 2 out of 30 MLL-positive cases analyzed by GS methods, the rearrangement was found in CD10(+) samples: one in a patient with CALL and one in a patient with pre-B ALL. While Attarbaschi et al.8 question the use of NG2 as a marker for MLL rearranged BPC-ALL, we point out that NG2 expression, evaluated by a combination of qualitative and quantitative FC analysis, is a highly reliable marker for MLL diagnosis, more than the absence of CD10. By collecting immunophenotypic NG2 data from 1461 B-ALL patients using both qualitative and quantitative methods, we tested FC performances according to GS response: sensitivity, specificity, predictive values and accuracy for both approaches produced comparable results with no significant differences in MLL diagnosis. Quantitative and qualitative FC produced five and six discordant results, respectively, compared to GS outcomes. Two out of five patients were considered negative by quantitative FC because of the low number of NG2-positive blast cells, enough to suspect MLL rearrangements by qualitative FC. This acknowledgment reduces the number of discordant cases to three in a comprehensive cohort of 1461 specimens. These findings suggest that the most accurate immunophenotyping approach to diagnose MLL rearrangements is given by the evaluation of both FC methods on NG2 antigen data combining subjective procedures (gating processes by qualitative FC) and objective measures (MESF values by quantitative FC). Moreover, taking into account that FC is undoubtedly a fast, cheap and easy technique compared to molecular diagnostics, our results support the introduction of immunophenotyping as a fundamental tool in the diagnosis of leukemias and the reliability of NG2 antigen as an accurate marker for MLL diagnosis. Immunophenotypic NG2 criterion for the diagnosis of MLL rearrangements provides a simple method with sufficient sensitivity and specificity to be applied for upfront clinical use in low-income countries.


  1. 1.

    , , , , , et al. MLL translocations specify a distinct gene expression profile that distinguishes a unique leukemia. Nat Genet 2002; 30: 41–47.

  2. 2.

    , , , , , et al. Detection of acute leukemia cells with mixed lineage leukemia (MLL) gene rearrangements by flow cytometry using monoclonal antibody 7.1. Leukemia 2000; 14: 1232–1238.

  3. 3.

    , , , . Computational analysis of flow-cytometry antigen expression profiles in childhood acute lymphoblastic leukemia: an MLL/AF4 identification. Leukemia 2003; 17: 1557–1565.

  4. 4.

    , , , , , et al. Quantitative multiparametric immunophenotyping in acute lymphoblastic leukemia: correlation with specific genotype. I. ETV6/AML1 ALLs identification. Leukemia 2000; 14: 1225–1231.

  5. 5.

    , , , , , et al. MLL gene rearrangement, cytogenetic 11q23 abnormalities, and expression of the NG2 molecule in infant acute myeloid leukemia. Blood 1997; 89: 3801–3805.

  6. 6.

    , . Antigen expression patterns reflecting genotype of acute leukemias. Leukemia 2002; 16: 1233–1258.

  7. 7.

    , , , , , . Expression of the human homologue of rat NG2 in adult acute lymphoblastic leukemia: close association with MLL rearrangement and a CD10(−)/CD24(−)/CD65s(+)/CD15(+) B-cell phenotype. Leukemia 2003; 17: 1589–1595.

  8. 8.

    , , , , , et al. Mixed Lineage Leukemia—rearranged childhood pro-B and CD10-negative pre-B acute lymphoblastic leukemia constitute a distinct clinical entity. Clin Cancer Res 2006; 12: 2988–2994.

Download references

Author information


  1. Department of Pediatrics, Hemato-Oncology Laboratory, University of Padua, Padua, Italy

    • A Zangrando
    • , F Intini
    • , G te Kronnie
    •  & G Basso


  1. Search for A Zangrando in:

  2. Search for F Intini in:

  3. Search for G te Kronnie in:

  4. Search for G Basso in:

Corresponding author

Correspondence to G te Kronnie.

About this article

Publication history




Further reading