EP3155423A1 - Means and methods for predicting an extramedullary relapse in acute lymphoblastic leukemia (all) - Google Patents

Abstract

The invention relates to the field of cancer diagnosis and therapy design, in particular diagnosis of extramedullary (EM) involvement in a subject afflicted with acute lymphoblastic leukemia. Provided is a method comprising determining in ALL cells the relative level and/or absolute amount of Stearoyl-Co A desaturase (SCD; EC 1.14.19.1) protein; and correlating the level and/or absolute amount of the SCD protein with the risk of developing an EM relapse, wherein an increased level and/or absolute amount of SCD relative to a control sample is indicative of having an increased risk of developing an extramedullary relapse. Also provided are reagent compositions and a diagnostic kit.

Description

Title: Means and methods for predicting an extramedullary relapse in acute lymphoblastic leukemia (ALL).

The invention relates to the field of cancer diagnosis and therapy, in particular diagnosis of extramedullary involvement in a subject afflicted with acute lymphoblastic leukemia.

Acute lymphoblastic leukemia (ALL) is the most common cancer in children. Survival rates in pediatric ALL patients have considerably increased over the last two decades. This improved survival can be attributed to new treatment modalities, better supportive care, and implementation of better risk stratification strategies, particularly based on minimal residual disease (MRD) diagnostics 1-3. Nevertheless, relapses still occur in 15-20% of pediatric ALL patients 2. In the three recent DCOG ALL protocols, 36% of all relapses in B-cell precursor (BCP) ALL and 95% of all relapses in T-ALL occurred during the 2-year period of treatment (Figure 1A). Most relapses (-60%) are isolated bone marrow (BM) relapses, but isolated CNS relapses (-20%) or isolated testis relapses (-10%) also occur frequently. Clearly, extramedullary (EM) involvement is much more frequent at relapse than at initial diagnosis, since only a small subset of ALL patients (1-4%) present with overt CNS involvement at diagnosis 4-10. However, submicroscopic CNS involvement may appear in up to 40% of patients 11-13.

In current clinical treatment protocols, prophylactic CNS-directed therapy is included. However, because of the developing and maturing CNS in children, CNS treatment has a high chance of inducing late sequelae ^{2 14- 16}. Therefore, it is crucial to identify patients at risk of CNS relapse at an early stage, preferably already at diagnosis, so that early targeted therapy can be given only to those who need it. Whereas current strategies for MRD detection allow

identification of most ALL patients at risk of BM relapse, 17-19 MRD measurements in BM cannot reliably predict (isolated) CNS relapses. High white blood cell count (WBC), terminal deoxynucleotidyl transferase (TdT) positive cells in cerebrospinal fluid (CSF) at diagnosis, and T-ALL

phenotype are associated with CNS relapse 11 12 20 21 but have limited value for therapy stratification, because high WBC, TdT+ CSF cells, and T- ALL phenotype are found in many patients that do not develop CNS relapse 13 22-25.

Consequently, the present inventors aimed to identify cellular and molecular markers that allow early identification of CNS relapses. It was hypothesized that migration and homing of ALL cells to EM locations like the CNS is an active process, mediated by molecules expressed by the ALL cells. Consequently, relapsed ALL cells at EM locations should have distinct biological properties as compared to relapsed ALL cells from BM of patients without EM involvement. Until now, only few proteins have been associated with migration of ALL cells to EM sites 26-29, probably because all comparative studies in patients with and without EM involvement were limited to ALL cells in BM only; ALL cells at EM sites (like CNS) have not been studied yet. If indeed ALL cells at EM sites have special characteristics (as compared to ALL cells that are retained in BM), it is more appropriate to compare ALL cells from EM sites with ALL cells from BM and evaluate what proteins are typically (over)expressed by ALL cells at EM sites. In addition, it was speculated that a (small) subpopulation of ALL cells with an "EM protein profile" might already be detectable in BM at the time of primary diagnosis.

To that end, gene expression profiles of ALL cells from

cerebrospinal fluid (CSF) and ALL cells from BM were compared and confirmed by real-time quantitative PCR. For a selected set of overexpressed genes, protein expression levels on ALL cells in CSF at relapse and on ALL cells in diagnostic BM samples were evaluated by multicolor flow cytometry.

It was surprisingly found that CNS-derived ALL cells showed a clearly different gene expression profile than BM-derived ALL cells. In particular, it was found that the enzyme stearoyl-CoA desaturase (SCD), optionally in combination with the SIBLING glycoprotein osteopontin (OPN), are highly valuable markers at diagnosis to predict EM relapse in ALL. Moreover, flow cytometric analysis showed that a subpopulation of ALL cells with a "CNS protein profile" (SCD and/or OPN positive) was already present in BM at diagnosis in all ALL patients (n=12) who later developed a CNS relapse, but was absent in the vast majority of the other patients (19/21; 90%). The presence of a subpopulation of ALL cells with a "CNS protein profile" at diagnosis predicts CNS relapse. Such information is advantageously be used to design new treatment strategies that aim at prevention of CNS relapse with reduced toxicity.

Accordingly, in one embodiment the invention provides a method for predicting the risk in a human subject having acute lymphoblastic leukemia (ALL) of developing an extramedullary relapse, comprising the steps of a) providing an isolated sample comprising ALL cells of said subject; b) determining in said ALL cells the relative level and/or absolute amount of stearoyl-CoA desaturase protein; and c) correlating the level and/or absolute amount of the SCD protein with the risk of developing an extramedullary relapse, wherein an increased level and/or absolute amount of SCD relative to a control sample is indicative of having an increased risk of developing an extramedullary relapse.

Stearoyl-CoA desaturase (SCD; EC 1.14.19.1) is an iron-containing enzyme that catalyzes a rate-limiting step in the synthesis of unsaturated fatty acids. The principal product of SCD is oleic acid, which is formed by desaturation of stearic acid. The ratio of stearic acid to oleic acid has been implicated in the regulation of cell growth and differentiation through effects on cell membrane fluidity and signal transduction. SCD has been reported to affect migration and to be involved in carcinogenesis. However, a role in ALL and EM relapse has never been taught or suggested in the art.

According to the invention, increased SCD expression in ALL cells is used a prognostic indicator for developing an EM relapse, in particular a CNS relapse. As shown herein below, individuals showing a subpopulation of about 1-10% SCD-positive ALL cells at diagnosis developed an isolated CNS relapse. In one embodiment, a level of > 1% SCD-positive ALL cells is used as indicator of having an increased risk of developing an extramedullary relapse. Preferably, a level of > 2% SCD-positive ALL cells is used, such as > 3% or > 4% SCD-positive ALL cells.

A method of the invention comprises determining in an isolated sample comprising ALL cells obtained from a human subject the relative level and/or absolute amount of stearoyl-CoA desaturase protein in said ALL cells.

A sample comprising ALL cells according to the invention is biological material, which has been obtained from an individual. In one embodiment, the sample is isolated at early diagnosis of ALL. In particular, said term includes material obtained from blood, plasma, tissue, bone marrow or cerebrospinal fluid (CSF). Preferably, the sample is obtained from bone marrow, blood, or CSF. A sample can also be material indirectly obtained from an individual, such as cells obtained from the individual which have been cultured in vitro, prior to obtain a sample from these in vitro cultured cells. A sample can also be pretreated prior to analysis with the methods of the invention. Such pretreatments for example can be storage of the sample at various temperatures, such as room temperature, 4°C, 0° C, -20°C, -70°C, - 80°C, or other temperatures or storage on water ice or dry ice, or storage in liquid nitrogen or storage in other solid, liquid or gas media.

The sample can be obtained from a human subject of any age. For example, the subject is an infant, a child, an adolescent, an adult, or an elderly person. In a specific aspect, the subject is an infant or a child.

A method of the invention is particularly suitable for predicting the risk in a human subject having B-cell precursor ALL (BCP-ALL) of developing an extramedullary relapse.

In a preferred embodiment, the method further comprises determining in said ALL cells the relative level and/or absolute amount of the osteopontin (OPN) protein; and correlating the level and/or absolute amount of the OPN protein with the risk of developing an extramedullary relapse, wherein an increased level and/or absolute amount of OPN compared to a control sample is indicative of having an increased risk of developing an

extramedullary relapse. Osteopontin, also known as bone sialoprotein I (BSP-1 or BNSP), early T-lymphocyte activation (ETA-1), secreted

phosphoprotein 1 (SPP1), 2ar and Rickettsia resistance (Ric), is a human gene product, which is also conserved in other species. Osteopontin is a SIBLING (glycoprotein) that was first identified in 1986 in osteoblasts. OPN binds to several integrin receptors including α46ΐ, a96l, and α964 expressed by leukocytes and are known to induce cell adhesion, migration, and survival in immune cells including neutrophils, macrophages, T cells, mast cells, and osteoclasts. The fact that OPN interacts with multiple cell surface receptors that are ubiquitously expressed makes it an active player in many physiological and pathological processes including wound healing, bone turnover, tumorigenesis, inflammation, ischemia, and immune responses. Therefore, manipulation of plasma (or local) OPN levels has been proposed to be useful in the treatment of autoimmune diseases, cancer metastasis, bone (and tooth) mineralization diseases, osteoporosis, and some forms of stress. However, the use of OPN as prognostic marker for (CNS) relapse in ALL has heretofore never been disclosed or suggested.

Hence, also provided is a method for predicting the risk in a human subject having ALL of developing an EM relapse, comprising the steps of a) providing an isolated sample comprising ALL cells of said subject; b) determining in said ALL cells the relative level and/or absolute amount of SCD and OPN; and c) correlating the level and/or absolute amount of the SCD and OPN protein with the risk of developing an extramedullary relapse, wherein an increased level and/or absolute amount of SCD and OPN is indicative of having an increased risk of developing an

extramedullary relapse.

A further aspect of the invention relates to a method for the stratification of a subject being afflicted with ALL to determine the therapy regimen for the treatment of ALL. At present, CNS status at the time of diagnosis, e.g. diagnosis of childhood ALL, is commonly assessed by examination of CSF obtained by lumbar puncture. However, in the brains of children with ALL after autopsy CNS involvement in more than 50 % of the examined specimens was revealed. Further, without targeted CNS therapy more than 50 % of the patients will relapse in the CNS suggesting that CNS

involvement is present at diagnosis in the majority of these patients. The use of SCD detection, optionally combined with OPN, now allows to diagnose leukemia with CNS involvement more precisely and, thus, allow to adjust the toxic CNS-directed anti-leukemia therapy more accurate. The determination of the relative level and/or absolute amount of SCD/OPN not only allows to initially diagnose leukemia with CNS involvement but also enables stratification of the subjects afflicted with leukemia. Hence, in a further embodiment the present invention relates to a method for the stratification of a subject being afflicted with ALL to determine the therapy regimen for the treatment of ALL, comprising

a) determining the relative level and/or the absolute amount of SCD in a sample from said subject;

b) comparing the level and/or amount of SCD to the level and/or amount of SCD in a control sample, with the level and/or amount of SCD in samples obtained prior to the begin of the therapy of said subject, or obtained in earlier stages of the regimen of said subject .

With stratification of a subject according to the invention is meant to determine the therapy regimen for the treatment of ALL. In particular, said stratification includes determining whether said subject will benefit from irradiation, in particular, from cranial irradiation, and determining the intensity of CNS-directed therapy. Said stratification allows CNS risk assessment in patients afflicted with ALL. For example, a method of the invention allows to identify patients at risk of CNS relapse at an early stage, so that targeted therapy can be given only to those who need it.

In a preferred embodiment, the stratification method further comprises determining the relative level and/or absolute amount of OPN in said sample; and comparing the level and/or amount of OPN to the level and/or amount of OPN in a control sample, with the level and/or amount of SCD in samples obtained prior to the begin of the therapy of said subject, or obtained in earlier stages of the regimen of said subject.

The term "control sample" as used herein refers e.g. to a sample from an individual not afflicted with ALL, or an individual afflicted with ALL without later development of an extramedullary relapse.

The expression "level" or "amount" is meant to describe the relative

SCD/OPN level or the absolute amount of SCD/OPN relative to the same molecule in a control sample. Relative means that no distinct amounts such as mole or milligram per liter etc. are stated, but that for example is stated the sample contains more, less or the same amount of a certain molecule as compared to a control sample. The term "more, less or the same amount" in this situation includes also arbitrary units.

With "elevated" or "increased" according to the invention is meant, that the molecule is present in a sample in higher quantities, or amounts or concentration, as compared to another sample, or as compared to a control sample or as compared to a negative control sample or as compared to a reference value, regard as if these measurements are relative or absolute measurements. Preferably, the molecule which is increased is present in a concentration, or quantity or amount which is at least 10 %, at least 20 %, at least 30 %, at least 40 %, at least 50 %, at least 60 %, at least 70 %, at least 80 %, at least 90 %, at least 100 %, at least 200 %, at least 300 %, at least 400 %, at least 500 %, at least 1000 % or at least more than 1000 % above the value to which it is compared. In other words, preferably the molecule is increased at least 2-fold, 3-fold, at least 3-fold, at least 4-fold, at least 5-fold, at least 10-fold above the level and/or amount in the control sample.

Suitable methods to detect the relative level and/or absolute amount of SCD/OPN protein are well known in the art. For example, suitable methods includes suitable mass spectrometry methods, e.g. matrix assisted laser desorption ionisation (MALDI), continuous or pulsed electrospray ionisation (ESI) and related methods such as ionspray or thermospray or massive cluster impact. The ion sources can be matched with detection formats including linear or non -linear reflection time of light (TOF), single or multiple quadrupole, single or multiple magnetic sector, Fourier

transformation ion cyclotron resonance (FTICR), ion trap, and combinations thereof. Other mass spectrometric methods suitable are for example fast atom bombardment (FAB), mass spectrometry, surface enhanced laser desorption (ionisation) (SELDI), mass spectrometry, etc. Further suitable immunologic methods among other are enzyme link immunoassays (ELISA), sandwich, direct, indirect, or competitive ELISA assays, enzyme link immunospotassays (ELIspot), radio immunoassays (RIA), flow-cytometry assays (FACS), immunohistochemistry, Western Blot, fluorescence

resonance energy transfer (FRET) assays, protein chip assays using for example antibodies, antibody fragments, receptor ligands or other agents binding the molecules of the invention.

Preferably, a method of the invention comprises a flow cytometric method allowing for simultaneous detection of ALL cells and the SCD/OPN

expression levels. Provided is a multi-color flow cytometric method for predicting the risk in a human subject having acute lymphoblastic leukemia (ALL) of developing an extramedullary relapse,

comprising

(i) staining the sample with a panel of differentially labeled antibodies capable of detecting ALL cells, the SCD protein and, optionally, the OPN protein;

(ii) subjecting the sample to flow cytometry;

(iii) gating on ALL cells for expression of the selected markers (i.e. SCD/OPN) detected by the antibodies; and

(iv) determining the relative abundance of the SCD protein, and optionally the OPD protein, in the ALL cells based on the expression profile of the multiple markers. For example, the ALL cells are identified using one or more markers selected from the group consisting of CD 10, CD 19, CD20, CD34 and CD45, preferably at least CD 19 and CD45, more preferably at least CD 10, CD 19, CD20 and CD45. Also provided are reagent compositions for predicting the risk in a human subject having ALL of developing an extramedullary relapse. The reagent composition is characterized by (i) reagent for detecting ALL cells and (ii) reagent for detecting SCD protein. Preferably, the reagent composition further comprises (iii) reagent for detecting OPN protein. A reagent composition is advantageously in a method of the invention for predicting the risk in a human subject having ALL of developing an extramedullary relapse and/or in a method of the invention for the stratification of a subject being afflicted with ALL to determine the therapy regimen for the

treatment of ALL.

In one embodiment, the reagents are antibodies, preferably fluorochrome- conjugated antibodies. For example, a reagent composition comprises a panel of at least four distinct fluorochrome-conjugated antibodies against the ALL markers CD 10, CD 19, CD20 and CD45, preferably further comprising one or more antibodies against CD34 and/or CD38.

Antibodies for detecting the ALL markers and the "CNS protein profile" markers SCD and OPN are known in the art and available from commercial source. For example, the SCD monoclonal antibody clone CD.E10 can be obtained from Abeam, Cambridge, UK. Monoclonal antibody clone 223112 to Human Osteopontin, N-terminal (NT) is available from R&D,

Minneapolis, MN, USA

Exemplary reagent compositions comprise distinctly labelled fluorochrome- conjugated antibodies directed against one of the following combinations of markers:

(i) CD 19, CD45, SCD and OPN;

(ii) CD20, SCD, CD45, OPN, CD 19 and CD 10;

(iii) CD20, CD45, SCD, CD34, OPN, CD 19 and CD 10; Suitable fluorochromes for conjugating antibodies for use in the present invention against the recited markers are known in the art. As will be understood, the fluorochromes used within a reagent composition should be distinguishable from each other by flow cytometry. The fluorochromes are preferably selected for brightness, limited spectral overlap and limited need for compensation, stability, etc.

The following fluorochromes is are of particular use in a reagent

composition according to the invention: (1) pacific blue (PacB), brilliant violet 421 (BV421) or Horizon V450, (2) pacific orange (PacO), Horizon V500 (HV500), BV510, Khrome orange (KO) or OC515, (3) fluorescein

isothiocyanate (FITC) or Alexa488, (4) phycoerythrin (PE), (5) peridinin chlorophyl protein/cyanine 5.5 (PerCP-Cy5.5), PerCP or PE-TexasRed, (6) phycoerythrin/cyanine7 (PE-Cy7), (7) allophycocyanine (APC) or Alexa647, and (8) allophycocyanine/hilite 7 (APC-H7), APC-Cy7, Alexa680, APC-A750, APC-C750 or Alexa700.

In one embodiment, the invention provides a reagent composition for 4-color flow cytometry, for example CD19-FITC / SCD-PE / CD45- PerCPCy5.5 / OPN-APC (or equivalent fluorochromes).

In another embodiment, the invention provides a reagent composition for 6-color flow cytometry, for example CD20-FITC / SCD-PE / CD45-PerCPCy5.5 / OPN-APC / CD19-PC7 / CD10-APC-C750 (or equivalent fluorochromes) .

In yet another embodiment, the invention provides a reagent composition for 8-color flow cytometry, for example CD20-PB / CD45-PO / CD38-FITC / SCD-PE / CD34-PerCPCy5.5 / OPN-APC / CD19-PC7 / CD 10- APC-C750 (or equivalent fluorochromes).

Exemplary compositions for 10- or 12 -color flow cytometry comprise antibodies against markers like CD 123, CD66c, CD73, CD81, CD 123, CD304 (or combinations thereof), Immunoglobulin (Ig) kappa, Ig Lambda (or a combination thereof) and TdT may be added to provide better distinction between normal B-cell precursors and BCP-ALL cells.

Hence, in a further embodiment a reagent composition for detecting BCP- ALL cells further comprises one or more antibodies selected from the group of antibodies against CD 123, CD66c, CD73, CD81, CD304, SmlgK , SmlgA and terminal deoxynucleotidyl transferase (TdT). In particular, the antibody panel further comprises one or more sets of antibodies selected from (a) set of antibodies against CD66c and CD 123; (b) set of antibodies against CD304 and CD73; and (c) set of antibodies against SmlgK and SmlgA, wherein the antibodies within each set are conjugated to the same fluorochrome.

Also provided herein is a diagnostic kit for predicting the risk in a human subject having acute lymphoblastic leukemia (ALL) of developing an extrameduUary relapse, said kit comprising a reagent composition according to the invention, optionally together with instructions for use, buffer, and/or control samples.

Still further, the SCD protein and, optionally, the OPN protein, are suitably used as therapeutic target(s), for example in a method for prophylactic treatment strategies in a subject at high risk of CNS leukemia, and/or in a method for treatment of CNS relapse. To that end, an agent capable of reducing or downregulating, either directly or indirectly, the level of said protein(s) is administered to the subject. For example, the agent is an inhibitory nucleic acid or an antibody directed against SCD and/or OPN. For example, the inhibitory nucleic acid is an inhibitory RNA that specifically targets SCD, or which that specifically targets OPN. Suitable inhibitory RNAs include a siRNA, microRNA, shRNA, or ribozyme. In one embodiment, the inhibitory nucleic acid is a small hairpin RNA or short hairpin RNA (shRNA), which is an artificial RNA molecule with a tight hairpin turn that can be used to silence target gene expression via RNA interference (RNAi). Expression of shRNA in cells is typically accomplished by delivery of plasmids or through viral or bacterial vectors. shRNA is an advantageous mediator of RNAi in that it has a relatively low rate of degradation and turnover. In one aspect, the shRNA is directed against SCD. In another aspect, the shRNA is directed against OPN.

LEGENDS TO THE FIGURES

Figure 1. Cumulative proportions of relapse in BCP-ALL and T-ALL.

A. In BCP-ALL patients (closed squares), 36% of the relapses occurred within the first two years after diagnosis, i.e. during treatment, whereas this was 95% in T-ALL patients (open squares). B. In 57% of BCP-ALL patients with an isolated CNS relapse, the relapse occurred within the first two years after diagnosis, compared to 35% of isolated BM relapses and 22% of combined relapses. C. In all T-ALL patients with an isolated CNS relapse or combined relapse, the relapse occurred within the first two years after diagnosis, compared to 92% of BM relapses. BCP-ALL or T-ALL patients with an isolated testis relapse showed comparable kinetics as combined BM and CNS relapses (data not shown). Data are obtained from consecutive ALL patients enroUed into the DCOG-ALL8 (n=467), DCOG-ALL9 (n=859) and DCOG-ALLIO protocol (n=522 until 12 October 2009 to ensure a minimal follow-up time of 5 years).

Figure 2. Overview of the various patient cohorts used in the different phases and experiments. The study was divided in three subsequent phases: Discovery phase (Gene expression profiling to identify differentially expressed genes between ALL cells in BM and CNS); Confirmation and antibody selection phase (RQ-PCR and flow cytometric analysis of ALL cells in BM and CNS); and Validation phase (Flow

cytometric analysis of ALL cells in BM at diagnosis to evaluate whether (small) subpopulations of ALL cells with a "CNS protein profile" can be detected in patients who ultimately develop an isolated CNS relapse).

Figure 3. Gene expression profiling of ALL cells from bone marrow or CSF. A. Principle Component Analysis (PCA) mapped scatter plot with the global gene expression profiles of all probe sets of BM-derived ALL cells at diagnosis (BM-Dx; n=22; dots in right hand cluster) and CSF-derived ALL samples at time of isolated CNS relapse (CNS; n=8; dots in left hand cluster). B. OmniViz treescape showing the hierarchical clustering of 269 differentially expressed Affymetrix probe sets of BM-derived ALL cells at diagnosis (BM-Dx; n=22) and CSF-derived ALL samples (CNS; n=8). Heat map is shown for the 269 probe sets from BM-derived ALL cells at diagnosis (n=22) and CSF-derived ALL samples at time of isolated CNS relapse (n=8). Red (left hand upper and right hand lower quadrants) indicates upregulated genes compared to the geometric mean and blue (left hand lower and right hand upper quadrants) indicates downregulated genes compared to the geometric mean. The colour intensity correlates with the degree of change. C. PCA mapped scatter plot with the global gene expression profiles of all probe sets of BM-derived ALL cells at time of isolated BM relapse (BM-R; n=20; dots in right hand cluster) and CSF-derived ALL samples at time of isolated CNS relapse (CNS; n=8; dots in left hand cluster). Statistically significant differences (ANOVA, FDR 5%, FC>1.5) were observed for SCD (p=0.0007), OPN (p=lE-05) and LPAR5 (p=0.0005). D. PCA mapped scatter plot with the global gene expression profiles of five paired samples of BM- derived ALL cells at diagnosis (BM Dx; red dots) and CSF-derived ALL cells at isolated CNS relapse (CNS; green dots). Statistically significant differences (paired t-test) were observed for SCD (p=0.010), OPN (p=0.0061) and LPAR5 (p=0.038).

Figure 4. Gene and protein expression levels of SCD, OPN, and LPAR5 on ALL cells in BM and CSF. A. Gene expression levels by ALL cells in BM at diagnosis, BM at relapse, or CSF at relapse. Expression levels were determined by gene expression profiling. Data represent the

normalized expression of the markers. All selected genes showed a significant increase in expression in CNS ALL cells (CNS; n=8) versus BM ALL cells at diagnosis (BM Dx; n=22) or relapse (BM R; n=20)(ANOVA, FDR 0.05). B. Protein expression levels in ALL samples freshly obtained from BM or CSF. Freshly obtained ALL ceUs from BM (n=14) or CSF (n=4) were stained for the selected markers and analyzed by flow cytometry; data are presented as MFI. Reported p-values were obtained by two-sided Mann- Whitney test.

Figure 5. Retrospective flowcytometric analysis of BM cells at diagnosis, obtained from BCP-ALL patients who later developed no relapse (n=10), an isolated BM relapse (n=ll), or an isolated CNS relapse (n=12). A. Representative examples of SCD and OPN expression by ALL cells. B. Percentage of SCD-positive cells and OPN MFI values of ALL cells at diagnosis of patients who remained in complete remission, developed an isolated BM relapse, or developed an isolated CNS relapse. C. A "CNS protein profile" was defined as: >1% SCD-positive ALL cells and/or OPN expression with MFI>450. A "CNS protein profile" was found in all 12 patients with an isolated CNS relapse, but in none of the other 22 patients. EXPERIMENTAL SECTION

Methods

Analysis of relapse data

Data from 293 consecutive relapsed ALL patients originally treated according to Dutch Childhood Oncology Group protocols ALL8, 30 ALL9, 31 and ALL 10 32 were used to evaluate the type of relapse and time to relapse (see Tables 1 and 2).

Table 1. Characteristics of consecutive relapsed BCP-ALL patients treated on DCOG protocols ^a

³ 28 B-cell precursor ALL patients with relapses other than isolated bone marrow (BM), isolated central nervous system (CNS), or combined BM-CNS relapses are not included in this table; ^b mean, range in between brackets; ^c mean. Table 2. Characteristics of consecutive relapsed T-ALL patients treated on DCOG protocols ^a

2 T-ALL patients with relapses other than isolated bone marrow (BM), isolated central nervous system (CNS), or combined BM-CNS relapses are not included in this table; ^b mean, range in between brackets; ^c mean.

Patient samples

BCP-ALL patients were included based on availability of sufficient cell material, in concordance with the Declaration of Helsinki and the local or national medical ethics guidelines. Samples were obtained within the framework of the international BFM study group between 2002 and 2013. An overview of the various patient cohorts used for the different phases and experiments is shown in Figure 2. Characteristics for the patients included in the discovery phase and in the validation phase are shown in Table 1; characteristics of other patients are shown in Table 3. Table 3. Characteristics of BCP-ALL patients included in the various analyses of the confirmation phase.

³ Mean, range in between brackets; ^b Mean; ^c All BM samples from DCOG protocols ALL8 or ALL9); Five CSF samples were obtained from patients initially treated according to DCOG protocols (ALL9: 4 patients; ALL10: 1 patient), two patients were obtained from the BFM group (ALL-BFM 2000 protocol), and one CSF sample was obtained from a Singapore patient (WK-ALL 2000 protocol).; ^d Negative samples all from DCOG protocols, nine positive samples from DCOG protocols, two from Berlin, one from Singapore; ^e All samples from DCOG protocols; ^f No data available for one patient. BM=bone marrow; PB=peripheral blood; WBC=white blood cell count

CSF of pediatric BCP-ALL patients with an isolated CNS relapse was collected by a spinal tap. ALL cells were directly isolated from CSF by centrifugation. Mononuclear cells (MNCs) from pediatric BCP-ALL patients with an isolated bone marrow (BM) relapse were isolated from the BM samples by Ficoll-Pague (Pharmacia, Uppsala, Sweden) density

centrifugation.

RNA extraction, labeling and hybridization

RNA was extracted from the ALL cells using Qiagen RNeasy isolation kit according to the manufacturer's instruction (Qiagen, Hilden, Germany). The integrity of total RNA was checked using the Agilent 2100 Bio-analyzer (Agilent, Santa Clara, CA, USA). RNA was converted to cDNA and subsequently to biotinylated cRNA by standard methods as described33 34.

Microarray for discovery & PCR for confirmation

Microarray studies were performed using Affymetrix GeneChip Human Genome U133 Plus 2.0 Array (Affymetrix, Santa Clara, CA, USA) according to manufacturer's protocol and analyzed as described previously.35 36 Briefly, samples were normalized by Robust Multichip Analysis

normalization and batch effects were removed. To visualize the clustering of the samples, principal component analysis (PCA) was used. Differentially expressed genes were identified using ANOVA. Cutoff values for

significantly expressed genes were a false discovery rate (FDR) of 0.05 or less and a fold change of 1.5. Using OmniViz software version 6.1.2.0

(Instem Scientific, Inc.) hierarchical clustering of differentially expressed genes was performed. Functional annotation of the significantly expressed genes was done using IPA (Ingenuity® Systems, www.ingenuity.com).

Quantitative Real-Time RT-PCR

Selected differentially expressed genes were confirmed by RQ-PCR analysis. RNA was reverse transcribed into cDNA, and realtime quantitative PCR (RQ-PCR) was performed with newly designed primers in combination with FAM-TAMRA-labeled probes or Universal Probes (Universal Probe Library; see Table 4). The expression levels were normalized to the control gene glucuronidase (GUSB) as described previously37.

Table 4. Reverse, Forward primers and probe sequences for RQ-PCR

* Roche Universal probe library (Roche, Almere, The Netherlands)

Protein studies by flow cytometry for validation

Antibodies directed against the proteins encoded by selected genes (conjugated with PE or APC) were evaluated in 8-color EuroFlow-based flowcytometric protocols (see Table 5). Intracellular staining of SCD and OPN was performed using Fix&Perm (An der Grub, Vienna, Austria) according to EuroFlow protocols 38. Cells were measured on a LSRII flow cytometer (BD Biosciences), using standardized EuroFlow settings 38. Analysis was performed using Infinicyt software (Cytognos, Salamanca, Spain). OPN and cytokines levels were measured in CSF by ELISA. Table 5. Antibodies used for flowcytometric immunophenotyping

* Antibodies were conjugated ocally using Lightning Link Antibody labelling (Innova Biosciences, Cambrid ge, UK)

Statistical Analysis

The Mann-Whitney U test was used to determine differences in mRNA and protein expression between BM-derived and CSF-derived ALL cells; the paired t-test was used to determine differences in mRNA and protein expression between paired BM-derived and CSF-derived ALL cells. In all tests (two-sided), a p-value of less than 0.05 was considered significant.

Results

CNS relapses are the most common extramedullary relapses in BCP-ALL and T ALL We evaluated EM relapses in consecutive relapsed pediatric patients with BCP-ALL (n=246) or T-ALL (n=42) treated according to the DCOG-ALL8, ALL9 or ALL 10 protocols. In BCP-ALL patients, 72 relapses (29% of all relapses) occurred at EM sites: 28 were isolated CNS (39%), seven isolated testis (10%) and 31 combined EM and BM relapses (43%); six (8%) involved other isolated EM sites. In T-ALL, 18 relapses were EM (43% of all relapses): 11 were isolated CNS (61%) relapses and five combined EM and BM relapses (28%); two (11%) involved other EM sites. Because of the dominance of CNS relapses within the EM relapses, we focused our further study on CNS relapses.

Analysis of the time-to-relapse in BCP-ALL patients showed that isolated CNS relapses occurred mainly in the first two years after diagnosis, i.e.

during therapy (57%), whereas only 35% of the BM relapses and 22% of the combined relapses occurred during therapy (Figure IB).

In the consecutive DCOG-ALL8, ALL9, and ALL 10 treatment protocols (minimal follow-up period: 5 years), time-to-relapse gradually increased for all types of relapses, but isolated CNS relapses remained fastest: 78%, 58%, and 43% of all CNS relapses occurring within the 2-year treatment period, respectively (data not shown). A comparable pattern was seen in T-ALL patients (Figure 1C). Of note, in the MRD-based DCOG-ALLIO protocol, CNS relapses were relatively increased to approximately 25% of all relapses, compared to approximately 15% of all relapses in DCOG-ALL8. Therefore, early identification of patients at risk for CNS relapse becomes increasingly important.

T-ALL phenotype, high WBC and TdT positive cells in CSF are associated with CNS relapses

Isolated CNS relapses were twice more frequent in T-ALL (27% of all relapses) as compared to BCP-ALL (13% of all relapses). There was no significant difference in the median WBC between patients with an isolated CNS relapse and patients with an isolated BM relapse. Significantly more TdT -positive ALL cells were observed at diagnosis in CSF of BCP-ALL patients who developed an isolated CNS relapse (18/23) compared to patients without relapse (8/19; p=0.0038) and patients who developed an isolated BM relapse (7/27; p<0.0001). However, the prognostic significance of this parameter is limited due to considerable overlap between patients with or without an isolated CNS relapse and due to technical limitations (e.g. risk of blood contamination during CSF sampling and low cell numbers in CSF).

Discovery of genes differentially expressed between BM-derived and CSF-derived BCP-ALL cells

Although CNS relapses are relatively more common in T-ALL than in BCP- ALL, absolute numbers of T-ALL patients with CNS relapse are very low. Therefore we focused our further studies on BCP-ALL patients. Gene expression profiling was used to discover genes that were differentially expressed between BCP-ALL cells derived from CSF at relapse (n=8) and BCP-ALL cells derived from BM at diagnosis (n=22). Unsupervised clustering analysis showed that CSF-derived BCP-ALL samples perfectly clustered into one separate group (Figure 3A). Compared to BM-derived ALL cells at diagnosis, CNS-derived ALL cells showed differential expression (ANOVA Fold Change (FC) >1.5, either up-regulated or down- regulated; FDR 0.05) of 269 probe sets (data not shown). A heat-map of these 269 probe sets is shown in Figure 3B. Further evaluation of the differentially expressed genes using IPA showed particularly involvement in cellular development and cell death and survival and linkage to several signaling pathways. Three genes were selected based on FC, p-value, putative function and/or membrane expression: SCD (Stearoyl-CoA desaturase), OPN (osteopontin), and LPAR5 (Lysophosphatidic acid receptor 5). For these genes, transcript expression levels in BM and CSF samples are shown in Figure 4A.

To confirm the differential expression of the selected genes, the gene expression profiles of CSF-derived BCP-ALL cells (n=8) were also compared with BM-derived BCP-ALL cells at relapse (n=20) (Figure 3C).

Furthermore, for five patients the gene expression profiles of CSF-derived BCP-ALL cells could also be compared with paired BM-derived BCP-ALL cells at diagnosis (Figure 3D). The three selected genes were also

significantly differentially expressed in these two additional analyses.

Confirmation of increased expression of the three candidate genes in CSF-derived BCP-ALL cells

Differential expression of SCD1, OPN, and LPAR5 was first confirmed by RQ-PCR analysis. To evaluate whether the three selected genes were also differentially expressed at the protein level, flowcytometric

immunophenotyping was performed. As compared to BCP-ALL cells obtained from BM at diagnosis, BCP-ALL cells derived from CSF at relapse showed higher expression of SCD (p=0.0233), slightly higher OPN

expression (p=0.0295), and comparable LPAR5 expression (p=0.6762; Figure 4B). The relatively limited increase of OPN expression in CSF-derived BCP- ALL cells might be due to the fact that OPN is a secreted protein. Indeed, soluble OPN levels were significantly increased in CSF samples of patients with CNS relapse (data not shown).

Given the potential involvement of signaling pathways, CSF samples were evaluated for cytokines able to activate these pathways. A significant increase in several cytokines involved in the JAK-STAT pathway (namely IL-4, IL-6, G-CSF and IFN-γ) was seen in CSF of patients with an isolated CNS relapse compared to CSF from patients without CNS involvement, although differences were small and only present in a subset of patients. Subpopulations of OPN and/or SCD positive ALL cells are already present in bone marrow at diagnosis in ALL patients who develop a CNS relapse

To validate our gene expression results, BM samples at primary diagnosis from BCP-ALL patients with isolated CNS relapse (n=12), BM relapse (n=l l) and without a relapse (n=10) (matched for age, sex, and

immunophenotype) were analyzed by 8-color flow cytometry for the presence of (sub)populations of BCP-ALL cells expressing SCD, OPN, and/or LPAR5 proteins. In all patients with an isolated CNS relapse, a clearly detectable subpopulation of SCD-positive BCP-ALL cells (>1%) was present at diagnosis. In full contrast, such (sub)population was not detected or small (<1%) in patients without a relapse or with an isolated BM relapse (Figure 5). Furthermore, the mean fluorescence intensity for OPN was significantly higher in patients with a CNS relapse than in patients without a relapse (p=0.0051) or a BM relapse (p=0.0127)(Figure 5). No significant differences were observed between the different patient groups for LPAR5 protein expression.

These data indicate that detection of >1% SCD positive BCP-ALL cells and/or detection of BCP-ALL cells with increased OPN expression

(MFI>450) at diagnosis may predict an isolated CNS relapse.

In summary, the present invention demonstrates that already a small population of BCP-ALL cells with a "CNS protein profile" is present in BM at diagnosis. A 8-color flow cytometry identified clearly detectable

subpopulations of BCP-ALL cells with a "CNS protein profile" (SCD and/or increased OPN expression) in BM at diagnosis of all 12 patients who ultimately developed a CNS relapse, but not in patients without a CNS relapse. Small SCD-positive subpopulations (<1%) can be detected in patients who did not develop a CNS relapse, suggesting that such small subpopulations are less capable to penetrate and stably home into the CNS or that these small sub o ulations are more efficiently killed by the prophylactic CNS-therapy in the DCOG-ALL8, ALL9 and ALL 10 treatment protocols.

The present data provide the first direct evidence that a

(sub)population of BCP-ALL cells with a "CNS protein profile" is already present in BM at diagnosis and that detection of such population, using e.g. standardized EuroFlow-based immunophenotyping, 43 can predict an isolated CNS relapse. Moreover, the proteins in the profile can not only be used as diagnostic markers for prediction of CNS-ALL, but may also serve as attractive therapeutic targets for prophylactic treatment strategies in patients at high risk of CNS leukemia, as well as for treatment of CNS relapse.

REFERENCES

1. Pui CH, Evans WE. N Engl J Med. 2006;354 (2): 166-178.

2. Pui CH, Robison LL, Look AT. Lancet. 2008;371 (9617): 1030- 1043.

3. Johnston WT, Lightfoot TJ, Simpson J, Roman E. Cancer Epidemiol. 2010;34 (6):659-666.

4. Conter V, Arico M, Valsecchi MG, et al. 1982-1995. Leukemia. 2000; 14 (12):2196-2204.

5. Eden OB, Harrison G, Richards S, et al. 1980-1997. Medical Research Council Childhood Leukaemia Working Party. Leukemia. 2000; 14 (12):2307-2320.

6. Gaynon PS, Trigg ME, Heerema NA, et al. Leukemia. 2000; 14 (12):2223- 2233.

7. Kamps WA, Veerman AJ, van Wering ER, van Weerden JF, Slater R, van der Does-van den Berg A. Leukemia. 2000; 14 (12):2240-2246.

8. Pui CH, Boyett JM, Rivera GK, et al. 2000; 14 (12):2286-2294. 9. Schrappe M, Reiter A, Zimmermann M, et al. Leukemia. 2000; 14

(12):2205-2222.

10. Silverman LB, Declerck L, Gelber RD, et al. Leukemia. 2000; 14

(12):2247-2256.

11. Januszkiewicz DA, Nowak JS. Hematol Oncol. 1995;13 (4):201-206.

12. Hooijkaas H, Hahlen K, Adriaansen HJ, Dekker I, van Zanen GE, van Dongen JJ. Blood. 1989;74 (l):416-422.

13. Hagedorn N, Acquaviva C, Fronkova E, et al. Blood. 2007; 110 (12):4022- 4029.

14. Dongen-Melman V. Leukemia. 1997; 11 (8): 1197- 1200.

15. Hoelzer D, Gokbuget N, Ottmann O, et al. Hematology Am Soc Hematol Educ Program. 2002: 162- 192.

16. Pui CH, Cheng C, Leung W, et al. N Engl J Med. 2003;349 (7):640-649.

17. van Dongen JJ, Seriu T, Panzer-Grumayer ER, et al. Lancet. 1998;352 (9142): 1731-1738.

18. Willemse MJ, Seriu T, Hettinger K, et al. Blood. 2002;99 (12):4386-4393.

19. Van der Velden VH, Corral L, Valsecchi MG, et al. Leukemia. 2009;23

(6) :1073- 1079.

20. Burger B, Zimmermann M, Mann G, et al. J Clin Oncol. 2003;21 (2): 184- 188.

21. Gajjar A, Harrison PL, Sandlund JT, et al. Blood. 2000;96 (10):3381- 3384.

22. Nathan PC, Maze R, Spiegler B, Greenberg ML, Weitzman S, Hitzler JK. Pediatr Blood Cancer. 2004;42 (l):24-29.

23. Goulden N, Langlands K, Steward C, et al. Br J Haematol. 1994;87 (2):282-285.

24. Neale GA, Pui CH, Mahmoud HH, et al. Leukemia. 1994;8 (5):768-775.

25. O'Reilly J, Meyer B, Baker D, Herrmann R, Cannell P, Davies J.

Leukemia. 1995;9 (4):624-627.

26. Annels NE, Willemze AJ, van der Velden VH, et al. Blood. 2004; 103

(7) :2806-2808.

27. Crazzolara R, Kreczy A, Mann G, et al. Br J Haematol. 2001; 115 (3):545- 553. 28. Carlo G, Izraeli S, Teichert A, et al. J Clin Oncol. 2007;25 (30):4813-4820.

29. Ismail EA, Mahmoud HM, Tawfik LM, et al. Eur J Haematol. 2012;88 (2):118-127.

30. Kamps WA, Bokkerink JP, Hakvoort-Cammel FG, et al. Leukemia.

2002; 16 (6): 1099-1111.

31. Veerman AJ, Kamps WA, van den Berg H, et al. Lancet Oncol. 2009; 10 (10):957-966.

32. van der Veer A, Waanders E, Pieters R, et al. Blood. 2013; 122 (15):2622- 2629.

33. de Ridder D, van der Linden CE, Schonewille T, et al. Leukemia. 2005; 19 (4):618-627.

34. Staal FJ, van der Burg M, Wessels LF, et al. Leukemia. 2003; 17 (7): 1324- 1332.

35. Staal FJ, Cario G, Cazzaniga G, et al. Leukemia. 2006;20 (8): 1385- 1392.

36. Staal FJ, de Ridder D, Szczepanski T, et al. Leukemia. 2010;24 (3):491- 499.

37. Beillard E, Pallisgaard N, van der Velden VH, et al. Leukemia. 2003; 17 (12):2474-2486.

38. van Dongen JJ, Lhermitte L, Bottcher S, et al. Leukemia. 2012;26 (9):1908- 1975.

39. de Jongste AH, Kraan J, van den Broek PD, et al. Cytometry B Clin Cytom. 2014;86 (4):272-279.

40. Tun HW, Personett D, Baskerville KA, et al. Blood. 2008; 111 (6) : 3200- 3210.

41. Incesoy-Ozdemir S, Sahin G, Bozkurt C, Oren AC, Balkaya E, Ertem U. Turk J Pediatr. 2013;55 (l):42-49.

42. Igal RA. Carcinogenesis. 2010;31 (9): 1509- 1515.

43. Kalina T, Flores -Montero J, van der Velden VH, et al. Leukemia. 2012;26 (9):1986-2010.

Claims

Claims

1. A method for predicting the risk in a human subject having acute lymphoblastic leukemia (ALL) of developing an extrameduUary relapse, comprising the steps of:

a) providing an isolated sample comprising ALL cells of said subject; b) determining in said ALL cells the relative level and/or absolute amount of Stearoyl-CoA desaturase (SCD; EC 1.14.19.1) protein; and

c) correlating the level and/or absolute amount of the SCD protein with the risk of developing an extrameduUary relapse, wherein an increased level and/or absolute amount of SCD relative to a control sample of an ALL patient without later development of an extrameduUary relapse is indicative of having an increased risk of developing an extrameduUary relapse.

2. Method according to claim 1, for predicting the risk of developing an extrameduUary relapse in the central nervous system (CNS).

3. Method according to claim 1 or 2, wherein a level of > 1% SCD- positive ALL cells, preferably > 2% SCD-positive ALL cells, is indicative of having an increased risk of developing an extrameduUary relapse.

4. Method according to any one of claims 1 to 3, further comprising determining in said ALL cells the relative level and/or absolute amount of the osteopontin (OPN) protein; and correlating the level and/or absolute amount of the OPN protein with the risk of developing an extrameduUary relapse, wherein an increased level and/or absolute amount of OPN compared to a control sample of an ALL patient without later development of an extrameduUary relapse is indicative of having an increased risk of developing an extrameduUary relapse.

5. Method according to any one of the preceding claims, wherein the control sample is from an individual not afflicted with ALL, or an individual afflicted with ALL without later development of an extramedullary relapse.

6. A method for the stratification of a subject being afflicted with ALL to determine the therapy regimen for the treatment of ALL, comprising

a) determining the relative level and/or the absolute amount of SCD in a sample from said subject;

b) comparing the level and/or amount of SCD to the level and/or amount of SCD in a control sample, with the level and/or amount of SCD in samples obtained prior to the begin of the therapy of said subject, or obtained in earlier stages of the regimen of said subject.

7. Method according to claim 6, further comprising determining the relative level and/or absolute amount of OPN in said sample; and comparing the level and/or amount of OPN to the level and/or amount of OPN in a control sample, with the level and/or amount of SCD in samples obtained prior to the begin of the therapy of said subject, or obtained in earlier stages of the regimen of said subject.

8. Method according to any one of the preceding claims, wherein said isolated sample comprises ALL cells obtained from bone marrow, peripheral blood or cerebrospinal fluid (CSF).

9. Method according to any one of the preceding claims, wherein the subject has B-cell precursor ALL (BCP-ALL).

10. Method according to any one of the preceding claims, wherein said sample is isolated at early diagnosis of ALL.

11. Method according to any one of the preceding claims, wherein the ALL cells are identified using one or more markers selected from the group consisting of CD 10, CD 19, CD20, CD34 and CD45, preferably at least CD 19 and CD45, more preferably at least CD 10, CD 19, CD20 and CD45.

12. Method according to any one of the preceding claims, wherein the subject is an infant, a child, an adolescent, an adult, or an elderly person.

13. Method according to any one of the preceding claims, being a multi- color flow cytometric method comprising

(i) staining the sample with a panel of differentially labeled antibodies capable of detecting ALL cells, the SCD protein and, optionally, the OPN protein;

(ii) subjecting the sample to flow cytometry;

(iii) gating on ALL cells for expression of the selected markers detected by the antibodies; and

(iv) determining the relative abundance of the SCD protein, and optionally the OPD protein, in the ALL cells based on the expression profile of the multiple markers.

14. A reagent composition for predicting the risk in a human subject having acute lymphoblastic leukemia (ALL) of developing an

extramedullary relapse, said composition comprising (i) a reagent for detecting ALL cells and (ii) a reagent for detecting SCD protein.

15. Reagent composition according to claim 14, further comprising (iii) a reagent for detecting OPN protein.

16. Reagent composition according to claim 14 or 15, wherein said reagents are antibodies, preferably fluorochrome-conjugated antibodies.

17. Reagent composition according to any one of claims 14-16, comprising a panel of at least four distinct fluorochrome-conjugated antibodies against the ALL markers CD 10, CD 19, CD20 and CD45, preferably further comprising one or more antibodies against CD34 and/or CD38.

18. Reagent composition according to claim 17, wherein the panel further comprises one or more antibodies selected from the group of antibodies against CD 123, CD66c, CD73, CD81, CD304, SmlgK , SmlgA and terminal deoxynucleotidyl transferase (TdT).

19. Reagent composition according to any one of claims 14-18, comprising distinct fluorochrome-conjugated antibodies directed against one of the following combinations of markers:

(i) CD 19, CD45, SCD and OPN;

(ii) CD20, SCD, CD45, OPN, CD 19 and CD 10;

(iii) CD20, CD45, SCD, CD34, OPN, CD 19 and CD 10.

20. A diagnostic kit for predicting the risk in a human subject having acute lymphoblastic leukemia (ALL) of developing an extramedullary relapse, said kit comprising a reagent composition according to any one of claims 14- 19, optionally together with instructions for use, buffer, and/or control samples.

21. Use of the SCD protein and, optionally, the OPN protein, as therapeutic target(s) for prophylactic treatment strategies in patients at high risk of CNS leukemia, and/or for treatment of CNS relapse.