Acute Leukemias

The homeobox gene CDX2 is aberrantly expressed and associated with an inferior prognosis in patients with acute lymphoblastic leukemia

Article metrics


Molecular characterization of acute lymphoblastic leukemia (ALL) has greatly improved the ability to categorize and prognostify patients with this disease. In this study, we show that the proto-oncogene CDX2 is aberrantly expressed in the majority of cases with B-lineage ALL and T-ALL. High expression of CDX2 correlated significantly with the ALL subtype pro-B ALL, cALL, Ph+ ALL and early T-ALL. Furthermore, high expression of CDX2 was associated with inferior overall survival and showed up as a novel and strong risk factor for ALL in bivariate analysis. Functional analyses showed that overexpression of Cdx2 in murine bone marrow progenitors perturbed genes involved in lymphoid development and that depletion of CDX2 in the human ALL cell line Nalm6 inhibited colony formation. These data indicate that aberrant CDX2 expression occurs frequently and has prognostic impact in adult patients with ALL.


With intensive risk adapted chemotherapy, stem cell transplantation and targeted therapies, the outcome of adult acute lymphoblastic leukemia (ALL) has improved in the past decades from <10 to 40–50%.1, 2 In addition, great progress has been made in understanding the biology of ALL: recently, genome-wide analyses of ALL patients showed deletions and mutations of genes associated with lymphoid development, such as LEF1, TCF, PAX5, IKAROS and NOTCH1.3, 4, 5 However, for the majority of ALL cases the molecular mechanisms driving the malignant transformation are unknown. Recently, data from experimental models and gene expression profiling in patients with acute myeloid leukemia (AML) have identified CDX2 as a powerful oncogene when aberrantly expressed in adult hematopoietic progenitor cells:6, 7, 8 the ‘caudal related homeobox gene’ CDX2 belongs to the family of so-called ‘ParaHox genes’, which also includes CDX1, CDX4 and the GSH2 homeobox gene.9 Normally expression of CDX2 is tightly regulated in the adult organism, with expression in the intestine but no expression in adult hematopoietic tissue.10 It was shown that CDX2 is among the most frequent aberrantly expressed proto-oncogenes in AML, with up to 89% of AML cases with normal karyotype expressing CDX2.8, 11 In AML patients, high expression levels of CDX2 were closely associated with HOX gene deregulation.11 However, it was shown that besides Hox genes Cdx2 regulates other stem cell regulatory genes, such as Scl, Gata1 and Runx1.12 Thus, aberrant expression of CDX2 might perturb the stem cell regulatory network at different levels.

We now report that CDX2 is aberrantly expressed in 81% of adult patients with ALL, and that high expression levels of this proto-oncogene predict poor treatment outcome in patients with this disease.

Materials and methods

Patient samples

Bone marrow or peripheral blood samples from 57 newly diagnosed adult patients with ALL were analyzed. Of these, 31 patients were enrolled between 2000 and 2004 in the protocols 06/99 and 07/03 of the German Multicenter Study Group for Adult ALL (GMALL).13 Sixteen patients were enrolled in other multicenter protocols (GMALL Elderly 12/96, GMALL Elderly 01/03 and GMALL B-ALL/NHL 02). Ten patients were treated outside a study protocol. Patients gave written informed consent according to the Declaration of Helsinki. The studies were approved by the ethics board of the University Frankfurt am Main, Germany.

Quantitative PCR

Expression analyses were performed by TaqMan qRT-PCR using the Applied Biosystems 7900HT fast real-time PCR system with pre-designed gene expression assays purchased from Applied Biosystems (Assay IDs CDX2: Hs01078080_m1, HOXA7: Hs00600844_m1, HOXA9: Hs00365956_m1, HOXB6: Hs00980016_m1, TATA box binding protein (TBP): 4333769F; Applied Biosystems, Foster City, CA, USA). Reactions were run with 1 μl of cDNA containing the equivalent of 50 ng total RNA in a total volume of 20 μl. ΔCT values were obtained by normalization to the housekeeping gene TBP. For patients with undetectable ΔCT values (negative samples), the lowest possible ΔCT was calculated by subtracting the CT value of TBP from the maximum number of cycles (45 cycles). Note that ΔCT values are inversely correlated to gene expression levels. For the ‘low density assay’ (custom design LDA, array configuration 7), 92 different genes involved in self-renewal, proliferation and differentiation were selected. Each sample was run in duplicate and fold expression was calculated using the ΔΔCT method after normalizing to β-actin.

Retroviral infection and in vitro assays

Primary mouse BM cells were transduced as described earlier.7 For transduction of Cdx2, cells were co-cultured with irradiated (4000 cGy from a 137Cs γ-radiation) GP+ E86 Cdx2 producer cells. Clonogenicity of short hairpin RNA (shRNA) transduced Nalm6 cells was analyzed in the CFC assay as described earlier (Methocult H4434, Stemcell Technologies, Vancouver, British Colombia, Canada).14

Short hairpin RNA

Short hairpin RNA against CDX2 (NM_001265) were designed using the following website Four different shRNAs were obtained and cloned into the MSCV/LTRmiR30-PIG retroviral vector (kindly provided by Scott W Lowe, Howard Hughes Medical Institute, New York, USA). Short hairpin RNA sequences are available on request. Cell lines were retrovirally transduced as described earlier in Ahmed et al.14

Methylation analysis

Quantitative DNA methylation status of the CpG island surrounding the transcription start site of CDX2 (−181 to +163)15 was assessed by pyrosequencing of bisulfite-treated genomic DNA. After bisulfite treatment of genomic DNA, the region of interest was amplified by the primer set CDX2_F (5′-IndexTermTTGGTGTTTGTGTTATTATTAATAGAGTTTTGTAAATAT-3′) and CDX2_R (5′-biotin-IndexTermATCCCAAAACAAACCTCACCATACTA-3′). PCR product was immobilized to Streptavidin Sepharose HP beads (GE Healthcare, Waukesha, WI, USA) followed by annealing to the sequencing primer. For sequencing three different primers were used, which are available on request. CpG analysis was done with Pyro Q-CpG software (Biotage, Uppsala, Sweden).

Statistical analysis

Correlations of CDX2 expression level with other patient variables were analyzed using Fisher's exact test and the Mann–Whitney U-test. For this, patients were divided into groups with high and low/absent CDX2 expression using the median ΔCT value (7.14) as cut-off. Data from patients enrolled in the GMALL 06/99 and 07/03 studies were used for correlation of CDX2 expression level with different patient parameters as these patients underwent comparable risk stratification and received similar treatment regimens. Univariate survival analysis was performed with the log-rank test and Cox regression, CDX2 expression levels together with other potential risk factors for ALL were analyzed in a bivariate Cox proportional hazards model for overall survival. P-values <0.05 were considered statistically significant. Statistical analyses were performed using SPSS (SPSS Inc, Chicago, IL, USA).


CDX2 is aberrantly expressed in the majority of patients with acute lymphoblastic leukemia

Fifty-seven samples from adult patients with newly diagnosed ALL (Table 1) were analyzed for CDX2 expression levels by qRT-PCR. In contrast to human normal hematopoietic bone marrow, CD34+7, 8 or peripheral blood mononuclear cells (PBMCs), 46 of 57 ALL patients (81%) were positive for CDX2 expression. Among the CDX2-positive cases, expression levels varied substantially between different ALL subgroups: the median expression level was 16-fold higher in pro-B ALL compared with mature B-ALL and 29-fold higher in early T-ALL compared with thymic T-ALL (Figure 1a, Table 1). When high and low CDX2 expression levels were defined as below or above the median ΔCT value, none of the patients with mature B-ALL or thymic T-ALL had high CDX2 expression, whereas all the patients with pro-B ALL, 8 of 10 patients with cALL and 8 of 11 patients with Ph+ ALL (10 pre-B/c-ALL and 1 pro-B ALL) showed high CDX2 expression (P<0.001) (Figure 1b). Of note, ALL subtypes such as pro-B ALL expressed CDX2 more than four-fold higher than AML cases with normal karyotype, described as the AML subgroup with highest CDX2 expression levels11 (Figure 1a).

Table 1 Patient characteristics
Figure 1

Quantitative expression of CDX2 and HOX genes in ALL patients. (a) Expression analysis of CDX2 in different acute lymphoblastic leukemia (ALL) subtypes and acute myeloid leukemia (AML) with normal and abnormal karyotype was performed by qRT-PCR. ΔCT values were calculated by normalization to the housekeeping gene TATA box binding protein (TBP). Dots represent individual patients, bars indicate the median expression level of CDX2 in the different ALL subgroups (±0.5 × interquartile range (IQR)). The dashed line indicates the median expression level of CDX2 for all patients measured. Note that ΔCT values are inversely correlated to expression level. (b) Distribution of patients with low/absent versus high CDX2 expression levels in different ALL subgroups. Patients with no CDX2 expression or expression above the median (ΔCT of 7.14) were considered as low/absent, patients with CDX2 expression equal to or below the median as high. (c) Expression of HOX genes in CDX2-positive ALL patients and normal healthy individuals. (MNC, mononuclear BM cells, CD34+, CD34+BM cells) quantified by qRT-PCR. Columns represent average expression levels±s.e.m. The number of patient samples is indicated.

Although HOX gene expression was detectable in CDX2-positive ALL cases, there was no correlation between expression of CDX2 and expression of the leukemogenic HOXA7 or HOXA9 genes. There was a correlation between CDX2 and HOXB6 expression, but of borderline significance (P=0.048, Mann–Whitney U-test) (Figure 1c).

We tested promoter hypomethylation as a potential cause for aberrant CDX2 expression, but did not detect any difference in methylation of a region surrounding the transcriptional start site of CDX28, 15 between CDX2-positive and CDX2-negative ALL cases (Table 2). We next analyzed whether shRNA-mediated CDX2 depletion would affect the CDX2-positive human pre-B ALL cell line Nalm6. In the CFC assay, Nalm6 cells transduced with HU_672_CDX2 (inducing a 49% reduction of CDX2 expression at the mRNA level by qRT-PCR) generated 33% fewer colonies compared with empty vector control and the non-active shRNA HU_673_CDX2 (data not shown), indicating an important role of CDX2 for the growth of this cell line.

Table 2 Methylation analysis of ALL samples

Besides Hox genes,11 enforced expression of Cdx2 in normal murine progenitors significantly upregulated genes involved in lymphopoiesis, such as Lef1 (3.9-fold), Tcf3 (13.3-fold), Flt3 (4.2-fold) and Id3 (10.2-fold) (Figure 2).

Figure 2

Differential gene expression in Cdx2-transduced bone marrow. Differential gene expression in Cdx2-transduced-bonemarrow cells. Fold expression levels of different genes in murine progenitor enriched 5-FU bone marrow after ectopic expression of Cdx2 determined 72 h after the end of transduction. Fold expression difference compared with empty vector-transduced bone marrow cells was calculated by averaging three independent experiments. Only genes with significant up- and downregulation are shown.

High CDX2 expression levels are associated with poor treatment outcome in ALL

When we analyzed the expression of CDX2 in different risk groups of ALL patients (according to Raff et al.13), most very high risk (VHR) and high risk (HR) patients showed high CDX2 expression, in contrast to the standard risk (SR) group (P=0.007 according to Fisher's exact test) (Table 3).

Table 3 Correlation of CDX2 expression levels with clinical patient characteristics

In a univariate analysis, high CDX2 expression levels were significantly associated with inferior OS (hazard ratio (HR) 4.2, 95% CI: 1.1–15.7) (Figure 3): survival rate two years after diagnosis was 86% of patients with low/absent and 52% of patients with high CDX2 expression levels (P=0.019, log rank test). In bivariate Cox regression, the CDX2 expression level remained a significant risk factor even after adjusting for the risk factors age and presence of molecular markers. The hazard ratio for CDX2 did not change substantially after adjusting for age (HR for CDX2 3.3, 95% CI: 0.86–12.85), presence of molecular aberrations (HR for CDX2 4.0, 95% CI: 1.07–15.13), ALL subtype (HR for CDX2 1.8, 95% CI: 0.36–8.96), ALL risk category (HR for CDX2 2.9, 95% CI: 0.69–12.25), presence of karyotype aberrations (HR for CDX2 4.7, 95% CI: 0.99–12.45), gender (HR for CDX2 6.2, 95% CI: 1.58–24.24) or leukocytes at presentation (HR for CDX2 3.8, 95% CI: 0.97–15.18).

Figure 3

High expression of CDX2 is associated with poor treatment outcome in acute lymphoblastic leukemia (ALL). Kaplan–Meier plot for overall survival of patients with high versus low/absent CDX2 expression levels.


This report describes for the first time that the proto-oncogene CDX2 is aberrantly expressed in the majority of patients with B-lineage or T-lineage ALL. Thus, it extends our knowledge about the deregulation of this transcription factor that was reported earlier to be one of the most frequent ectopically expressed proto-oncogenes in AML,8, 11 and underlines that aberrant expression of this homeobox gene is not restricted to myeloid malignancies. This aberrant expression of CDX2 in human leukemias stands in clear contrast to highly related genes such as CDX4 or CDX1, which were not expressed in our series of AML patients or were detected only in a minor fraction of patients with AML.11, 16 Furthermore, analyses of CDX4 expression in our series of ALL samples did not show expression of this ParaHox gene (data not shown). The reason for this is unknown. But data from experimental murine models indicate that acquiring aberrant CDX2 expression might provide a greater growth advantage for the cell than acquiring aberrant CDX4 expression as constitutive expression of Cdx2 seems to be much more leukemogenic than Cdx4 in a murine BM transplantation model.16

So far the leukemogenic potential of Cdx2 has been linked to its ability to deregulate HOX gene expression: it has been shown that Cdx2 is able to upregulate expression of 5′ located Hox genes known to be leukemogenic and that this ability depends on its N-terminal transactivation domain.11 However, it is unclear whether CDX2 can also exert its oncogenic effect by other pathways. This might be particularly true in adult patients with ALL, in which HOX gene deregulation is comparably rare compared with AML.5 This could indicate that in ALL the leukemogenic effect of CDX2 does not depend on HOX gene perturbation to the same extent as in AML, and that CDX2 mediates its oncogenic potential through different gene networks in ALL. When we analyzed gene transcription after retroviral induction of aberrant Cdx2 expression in murine BM progenitor cells, we observed that besides Hox genes also factors involved in lymphoid development such as Lef1 were upregulated. Recently, we showed that aberrant expression of this factor caused AML and also B-lineage ALL in transplanted mice.17 Therefore, the aberrant CDX2 expression might initiate or support myeloid as well as lymphoid leukemogenesis, probably depending on the differentiation stage of the cell initially affected by CDX2 overexpression. The underlying mechanism for aberrant CDX2 expression in human leukemias is unknown: we did not detect any difference in promoter methylation of the CDX2 gene between CDX2-positive and CDX2-negative ALL cases. The same was reported for patients with AML, where also no mutation in the coding region or gene amplification was detected.8, 15

That aberrant CDX2 expression has functional relevance in ALL was supported by the observation that shRNA-mediated CDX2 depletion compromised the clonogenic growth potential of the CDX2-positive human pre-B ALL cell line Nalm6. Furthermore, high expression of CDX2 was mostly found in ALL patients with a high-risk profile. In line with this, high expression of CDX2 was associated with a significantly inferior OS. Remarkably and despite the limited patient number available for this analysis, CDX2 expression level remained a significant and strong risk factor even after adjusting for the risk factors age and presence of molecular markers in bivariate analyses.

Altogether, these results show that aberrant expression of CDX2 is a frequent event in adult patients with ALL. Delineating the underlying molecular mechanisms of aberrant CDX2 expression and characterizing factors mediating CDX2-induced leukemogenesis will help to understand important steps in leukemogenesis, which are common in both lymphoid and myeloid leukemias.


  1. 1

    Pui CH, Evans WE . Treatment of acute lymphoblastic leukemia. N Engl J Med 2006; 354: 166–178.

  2. 2

    Gokbuget N, Hoelzer D . Treatment of adult acute lymphoblastic leukemia. Hematol Am Soc Hematol Educ Program 2006; 1: 133–141.

  3. 3

    Armstrong SA, Look AT . Molecular genetics of acute lymphoblastic leukemia. J Clin Oncol 2005; 23: 6306–6315.

  4. 4

    Mullighan CG, Goorha S, Radtke I, Miller CB, Coustan-Smith E, Dalton JD et al. Genome-wide analysis of genetic alterations in acute lymphoblastic leukaemia. Nature 2007; 446: 758–764.

  5. 5

    Paulsson K, Cazier JB, Macdougall F, Stevens J, Stasevich I, Vrcelj N et al. Microdeletions are a general feature of adult and adolescent acute lymphoblastic leukemia: unexpected similarities with pediatric disease. Proc Natl Acad Sci USA 2008; 105: 6708–6713.

  6. 6

    Chase A, Reiter A, Burci L, Cazzaniga G, Biondi A, Pickard J et al. Fusion of ETV6 to the caudal-related homeobox gene CDX2 in acute myeloid leukemia with the t(12;13)(p13;q12). Blood 1999; 93: 1025–1031.

  7. 7

    Rawat VP, Cusan M, Deshpande A, Hiddemann W, Quintanilla-Martinez L, Humphries RK et al. Ectopic expression of the homeobox gene Cdx2 is the transforming event in a mouse model of t(12;13)(p13;q12) acute myeloid leukemia. Proc Natl Acad Sci USA 2004; 101: 817–822.

  8. 8

    Scholl C, Bansal D, Dohner K, Eiwen K, Huntly BJ, Lee BH et al. The homeobox gene CDX2 is aberrantly expressed in most cases of acute myeloid leukemia and promotes leukemogenesis. J Clin Invest 2007; 117: 1037–1048.

  9. 9

    Brooke NM, Garcia-Fernandez J, Holland PW . The ParaHox gene cluster is an evolutionary sister of the Hox gene cluster. Nature 1998; 392: 920–922.

  10. 10

    Suh E, Traber PG . An intestine-specific homeobox gene regulates proliferation and differentiation. Mol Cell Biol 1996; 16: 619–625.

  11. 11

    Rawat VP, Thoene S, Naidu VM, Arseni N, Heilmeier B, Metzeler K et al. Overexpression of CDX2 perturbs HOX gene expression in murine progenitors depending on its N-terminal domain and is closely correlated with deregulated HOX gene expression in human acute myeloid leukemia. Blood 2008; 111: 309–319.

  12. 12

    Wang Y, Yabuuchi A, McKinney-Freeman S, Ducharme DM, Ray MK, Chawengsaksophak K et al. Cdx gene deficiency compromises embryonic hematopoiesis in the mouse. Proc Natl Acad Sci USA 2008; 105: 7756–7761.

  13. 13

    Raff T, Gokbuget N, Luschen S, Reutzel R, Ritgen M, Irmer S et al. Molecular relapse in adult standard-risk ALL patients detected by prospective MRD monitoring during and after maintenance treatment: data from the GMALL 06/99 and 07/03 trials. Blood 2007; 109: 910–915.

  14. 14

    Ahmed F, Arseni N, Glimm H, Hiddemann W, Buske C, Feuring-Buske M . Constitutive expression of the ATP-binding cassette transporter ABCG2 enhances the growth potential of early human hematopoietic progenitors. Stem Cells 2008; 26: 810–818.

  15. 15

    Yuasa Y, Nagasaki H, Akiyama Y, Sakai H, Nakajima T, Ohkura Y et al. Relationship between CDX2 gene methylation and dietary factors in gastric cancer patients. Carcinogenesis 2005; 26: 193–200.

  16. 16

    Bansal D, Scholl C, Frohling S, McDowell E, Lee BH, Dohner K et al. Cdx4 dysregulates Hox gene expression and generates acute myeloid leukemia alone and in cooperation with Meis1a in a murine model. Proc Natl Acad Sci USA 2006; 103: 16924–16929.

  17. 17

    Petropoulos K, Arseni N, Schessl C, Stadler CR, Rawat VP, Deshpande AJ et al. A novel role for Lef-1, a central transcription mediator of Wnt signaling, in leukemogenesis. J Exp Med 2008; 205: 515–522.

Download references


We thank Bianka Ksienzyk and Nicole Behm for their excellent technical assistance, and the members of the animal facility at the Helmholtz Center Munich for their excellent breeding and maintenance of the animals. VPSR and CB and their work were supported by a grant of the DFG (SFB 684 project A7), MF-B by the Deutsche Krebshilfe (70-2968-Fe I to MFB) and the DFG (SFB 684 project A8), SKB by the DFG (SFB 684 project A6), CB, MF-B and SKB by the Bundesministerium für Bildung und Forschung (NGFN2 Grant 01GS0448) and DH by the Deutsche Krebshilfe (70-2657-Ho2).

Author information

Correspondence to C Buske.

Rights and permissions

Reprints and Permissions

About this article


  • CDX2
  • acute lymphoblastic leukemia
  • HOX genes
  • prognostic factor

Further reading