Effect of the Philadelphia Chromosome on Minimal Residual Disease In

Effect of the Philadelphia chromosome on minimal residual disease in acute lymphoblastic leukemia MJ Brisco1, PJ Sykes1, G Dolman1, S-H Neoh1, E Hughes1, L-M Peng2, G Tauro3, H Ekert3, I Toogood4, K Bradstock5 and AA Morley1

1Department of Haematology, Flinders Medical Centre, Bedford Park, South Australia; 2Department of Laboratory Medicine, School of Medicine, West China University of Medical Sciences, Chengdu, People’s Republic of China; 3Department of Haematology, Royal Childrens Hospital, Parkville, Victoria; 4Department of Haematology/Oncology, Womens and Childrens Hospital, North Adelaide, South Australia; and 5Department of Haematology, Westmead Hospital, Westmead, NSW, Australia

The Philadelphia translocation is associated with a poor prog- number of leukemic cells remaining at the end of induction nosis in adults and children with acute lymphoblastic leukemia, provides a good approximation of the degree of drug resist- even though the majority of patients achieve remission. To test ance in vivo. the hypothesis that the translocation leads to drug resistance 6–10 in vivo, we studied 61 children and 20 adults with acute lym- Since the initial reports, a number of groups have phoblastic leukemia and used the level of minimal residual dis- developed sensitive methods to quantify minimal residual disease at the end of induction as the measure of drug resistance ease (MRD) by use of the polymerase chain reaction (PCR). in vivo. In children the presence of the translocation was asso- Our own approach has been to use the rearranged immuno- ciated with a significant increase in residual disease, indicating globulin heavy chain gene as a molecular marker for the higher drug resistance in vivo; five of seven Philadelphia-positive children but only five of 54 Philadelphia-negative children leukemic clone; to sequence the gene; to synthesise patient- had a minimal residual disease level Ͼ10؊3, a level which is specific PCR primers, which enable detection of the marker associated with a high risk of relapse in childhood acute lym- gene (and hence the leukemic cells); and to quantify leukemic phoblastic leukemia of standard risk. By contrast, in adults, cells by using limiting dilution analysis and Poisson stat- residual disease and hence drug resistance was already higher istics.11,12 In this paper, we have applied this approach to a than in children, and the presence of the Philadelphia translo- series of children and adults with ALL, in order to determine cation in seven patients had no obvious additional effect. We conclude that the Philadelphia chromosome may increase directly whether the Philadelphia translocation is associated resistance to drugs in vivo in children, but not detectably in with increased levels of residual disease at the end of induc- adults. tion, implying increased drug resistance in vivo. Keywords: minimal residual neoplasm; Philadelphia chromosome; acute lymphoblastic leukemia; drug resistance; polymerase chain reaction Materials and methods

Patients Introduction Patients were eligible for this study if: (1) they had B lineage It is well known that the presence of the Philadelphia translo- ALL; (2) their leukemia could be quantiﬁed by the techniques cation (t(9;22)(q34;q11)) markedly worsens the prognosis of described below; (3) they achieved morphological remission children and adults with acute lymphoblastic leukemia after induction treatment, and (4) there was enough material 1,2 (ALL). Philadelphia-positive (Ph+ve) patients achieve available for study. Patients were excluded if their cytogenetic remission less frequently or take longer to do so, and, even if testing was unsuccessful, or revealed a translocation other they achieve remission, most will relapse. The most obvious than the t(9;22): 23 otherwise eligible patients had no cyto- explanation is that the 9;22 translocation directly increases genetic information available, mainly due to technical prob- drug resistance even in patients who achieve remission. There lems; another 13 patients were not studied because they had is some evidence supporting this, as in several in vitro studies a translocation other than t(9;22) associated with higher on chronic myelogenous leukemia the presence of the bcr-abl relapse risk, eg t(1;19), t(4;11) or t(8;14). The 20 adult patients fusion gene prevented apoptosis following drug exposure.3–5 had been treated between 1984 and 1994 with the 7801-A However, other explanations of the poor outcome in Ph+ve protocol which involved a 5-week induction therapy of vincri- ALL are possible. For example, the 9;22 translocation might stine, prednisolone, daunorubicin, cytosine arabinoside, and increase genetic instability and drug resistance could be a asparaginase. The clinical features of most have been consequence of one or more secondary mutations. Alterna- described.13 The Philadelphia-negative adult patients (10 tively the 9;22 translocation might lead to a poor outcome by male, three female) had at diagnosis a median white cell mechanisms entirely unrelated to drug resistance. count 13 × 109/l (range 3–57 × 109/l) and median age of 21 In vivo drug resistance can be inferred from the rate of (range 15–38). Clinical data for Philadelphia-positive adults decline of leukemic cell number during treatment. However, are shown in Table 1. Three patients, including two with the since the number of leukemic cells in the body at diagnosis t(9;22), had white cell counts above 50 × 109/l (Table 1). The probably varies between patients by no more than one order 61 child patients were treated between 1985 and 1994 on of magnitude, whereas the number of leukemic cells at the Australian and New Zealand Children’s Cancer Group Studies end of induction varies by many orders of magnitude, the V (1985–92) and VI (1993–6), which involved the same 5- week induction therapy of vincristine, prednisolone, daunorubicin and asparaginase. The Philadelphia-negative child Correspondence: AA Morley, Department of Haematology, Flinders patients (28 male, 26 female) had at diagnosis a median white Medical Centre, Bedford Park, South Australia 5042 cell count of 10 × 109/l (range 1–252 × 109/l) and a median Received 27 January 1997; accepted 7 May 1997 age of 3 years 9 months (range 1 year 2 months to 15 years 11 Drug resistance in Ph؉ve ALL MJ Brisco et al 1498 Table 1 Clinical characteristics of patients with the Philadelphia Statistics translocation Comparisons between groups were made using the Mann– No. Sex Age WCC MRD Whitney U test, one-tailed unless otherwise indicated. (years) ×109/l at end of induction

Adults 1 F 74 25 5.2 × 10−2 Results 2 M 20 23 4.1 × 10−2 3 M 17 102 1.6 × 10−2 MRD was detected and quantified in 46 patients, at levels 4 M 33 10 4.9 × 10−3 from Ͼ10−1 to Ͻ10−6 (Figure 1, closed circles). For children, − 5 M 42 86 4.3 × 10 4 the MRD levels at the end of induction were significantly 6 F 19 2.8 6.6 × 10−6 + Ϫ × −6 higher in Ph ve patients than in Ph ve patients (medians 3 7 M 27 9.8 Ͻ3.3 × 10 − − 10 2 vs 8 × 10 5; P = 0.0008). For adults, the MRD levels at the end of induction in Ph+ve patients were not significantly Children −2 × −2 higher than in Ph−ve patients (medians 1.0 × 10 vs 4.2 × 8 M 8 7.7 6.7 10 − 9 F 2 181 3.9 × 10−2 10 3; P = 0.18; one-tailed). 10 F 14 2.3 3.2 × 10−2 11 M 5 73 2.1 × 10−2 12 M 15 186 8.4 × 10−3 13 M 3 17 Ͻ3.0 × 10−4 14 M 9 4 Ͻ1.4 × 10−4

months). Clinical data for Philadelphia-positive children are shown in Table 1. Twelve patients, including three with the t(9;22), had white cell counts above 50 × 109/l. This work was approved by the Flinders Medical Centre Committee on Clinical Investigation.

Tissue samples

Bone marrow was aspirated at diagnosis and at the end of induction treatment, and was usually available as ﬁxed, stained bone marrow smears, although unﬁxed smears or frozen marrow were sometimes used.

Quantiﬁcation of minimal residual disease

This was performed as previously described.11–14 Briefly, DNA was extracted by phenol-chloroform extraction. The rearranged immunoglobulin heavy chain gene, which acts as a marker for the leukemic clone, was amplified from the bone marrow at diagnosis, and sequenced. PCR primers were designed to the complementary determining region (CDR)3 and occasionally CDR2 regions and tested for specificity and sensitivity by ensuring that large amounts of amplified product of the correct size were obtained from one or a few molecules of the target DNA, and by demonstrating that no signal was obtained from 1 ␮g DNA from peripheral blood obtained from a healthy volunteer. Quantification was performed by using limiting dilution analysis. For leukemic genomes we used a nested PCR either with consensus primers or occasionally with one patient-specific primer and one consensus primer in the first round, followed by one or two patient-specific primers in the second round. For total amplifiable genomes we used primers for the N-ras gene in a one-round PCR. MRD levels were expressed as the proportion of nucleated cells which belong to the leukemic clone. Sequences of the rearranged immunoglobulin heavy chain Figure 1 Levels of minimal residual disease in bone marrow aspir- genes and details of the primers used for quantification can ates at the end of induction treatment in acute lymphoblastic leuke- be obtained from Genbank, and a list of accession numbers mia. (¼) MRD detected in marrow; (ș) MRD not detected; upper limit can be obtained from the authors. shown as derived from number of genomes examined. Drug resistance in Ph؉ve ALL MJ Brisco et al 1499 MRD was not detected in the other 35 patients, in part at from in vitro experiments which indicate that the bcr-abl least because too little DNA was available. However, it was translocation inhibits drug-induced apoptosis.3–5 If the bcr-abl possible to estimate an upper limit for the level of residual translocation were to lead directly to drug resistance then it leukemia from the number of genomes studied (Figure 1, open would be predicted that the degree of drug resistance would circles). If, conservatively, it is assumed that the level of MRD be relatively homogenous, both in the population of patients was actually equivalent to this upper limit, inclusion of these with the translocation and in the population of leukemic cells patients still indicated that Ph+ve children had significantly within any one patient. However, an indirect effect is also higher levels than Ph−ve children (P = 0.0008, one-tailed) possible if the translocation increases genetic instability. If a whereas Ph+ve adults did not have significantly higher levels second genetic change were to be responsible for develop- than Ph−ve adults (P = 0.37, one-tailed). ment of drug resistance, and if it were to occur at a higher Sufficient DNA was available for all patients to show frequency as the result of the bcr-abl translocation, then the whether a patient had MRD Ͼ10−3 (see Table 2). For children, second event could occur early in the course of the disease significantly more Ph+ve patients had MRD Ͼ10−3 than Ph−ve in the majority of Ph+ve patients but later in the majority of patients (Fishers exact probability test, P = 0.0007, one-tailed). Ph−ve patients. Consequently, Ph+ve patients would tend to For adults, the difference was not significant (P = 0.642, be relatively drug resistant at diagnosis whereas Ph−ve one-tailed). patients would tend to be relatively drug sensitive at diagnosis, Three of the seven Ph+ve children (43%) had at diagnosis but would develop drug resistance subsequently. Our data a white cell count above 50 × 109/l, compared to nine of the provide some support for this latter hypothesis, as a few of 54 Ph−ve children (17%). It might be argued that a higher our patients appeared at diagnosis to be still sensitive to ther- number of leukemic cells in Ph+ve patients, rather than any apy. Other data also indicate that the (9;22) translocation increase in resistance to treatment, led to high levels of leads to secondary mutations, at least in chronic myeloid leu- residual leukemia at the end of induction. To test this possi- kemia, as p53 mutations are commonly observed late in the bility, the three Ph+ve and nine Ph−ve children with high disease15,16 and at blast crisis, which probably has a genetic white cell counts were excluded, and the median MRD levels basis, develops in most patients. again compared. Ph+ve children still had a higher MRD level The results in adults were different from those observed in than did Ph−ve patients (patients with MRD detected: P = children. We have previously reported that at the end of 0.018; all patients: P = 0.011, one-tailed). A similar result was induction, the levels of MRD in Ph−ve adults were increased also obtained for the 12 patients with a high WCC (all compared to those in Ph−ve children.13 This is also seen in patients: Mann–Whitney U test, P = 0.001). This suggests that Figure 1 which shows that in adults there was no clear evi- a higher number of leukemic cells in Ph+ve child patients at dence of a further increase associated with presence of the diagnosis did not account for the higher level of MRD after 9;22 translocation. These observations suggest that in vivo induction. drug resistance is already present in most adults with Ph−ve To study the effect of age alone, the levels of MRD in adults ALL and is increased little if at all by the translocation. There and children were compared for patients in whom MRD was are at least three possible explanations for this difference detected. For Ph+ve patients there was no significant differ- between adults and children. Firstly, cells in adults may differ ence between adults and children (P = 0.27, two-tailed), from those in children in pathways, such as drug metabolism, whereas for Ph−ve patients, adults had a significantly higher which are related to drug resistance.17 Secondly, the leukemia MRD level than children (P = 0.0097, one-tailed). Inclusion in adults may differ biologically from that in children, eg in of the patients in whom MRD was not detected did not alter the prevalence of various subtypes or of mutations associated the conclusions significantly (Ph+ve: P ϭ 0.24; Ph−ve with drug resistance. An example of the latter is the t(12;21) patients: P = 0.004, one-tailed). which occurs commonly in childhood but rarely in adult ALL18,19 and which is associated with an excellent response to treatment. Thirdly, children and adults may have different Discussion forms of the t(9;22) and these forms may lead to different levels of drug resistance. Most children, but only about a third Children with Ph+ve ALL have significantly higher levels of of adults, produce a p190bcr-abl fusion protein, and the rest MRD at the end of induction than do children with Ph−ve produce a p210bcr-abl fusion protein. The two forms may have ALL. This difference is not due to a higher overall number different biological activities,20 and in transgenic animals of leukemic cells at diagnosis in the Ph+ve children and we there is evidence that they affect the course of disease differ- conclude that in children the (9;22) translocation increases ently.21 Finally, if the (9;22) translocation were to have only drug resistance in vivo. This increase may be a direct effect a small effect in adults, then such an effect could have been of the translocation, and supporting evidence for this comes missed in our study. The number of patients studied was small and, as cytogenetics occasionally fails to detect the translocation,19 occasional patients with the translocation could have Table 2 MRD level at the end of induction treatment in adult and − child patients been incorrectly scored as Ph ve. We have previously found that an MRD level of Ͼ10−3 at No. of children No. of adults the end of induction in Philadelphia-negative standard risk ALL patients carries a particularly poor prognosis.11,13 There MRD Ph+ve Ph−ve MRD Ph+ve Ph−ve are several possible explanations: a high level of leukemic cells at the end of induction therapy shows that the leukemic Ͼ10−3 55Ͼ10−3 45 cells are intrinsically resistant to drugs; a large population at Ͻ10−3 249Ͻ10−3 38 the start of post-induction therapy decreases the chance of eradicating the last clonogenic leukemic cell; a large leukemic All patients could be classified according to whether they had MRD population during treatment increases the probability of above or below 10−3. further mutations which increase drug resistance. In our study Drug resistance in Ph؉ve ALL MJ Brisco et al 1500 most Ph+ve children and adults had MRD levels Ͼ10−3 (Table 9 Brisco MJ, Tan LW, Orsborn AM, Morley AA. Development of a 2). These levels would be expected to predict for a poor out- highly sensitive assay, based on the polymerase chain reaction, come, and this prediction would agree with the known poor for rare B-lymphocyte clones in a polyclonal population. Br J Haematol 1990; 75: 163–167. prognosis associated with the presence of the Philadelphia 10 Jonsson OG, Kitchens RL, Scott FC, Smith RG. Detection of mini- chromosome. mal residual disease in acute lymphoblastic leukemia using immunoglobulin hypervariable region specific oligonucleotide probes. Blood 1990; 76: 2072–2079. Acknowledgements 11 Brisco MJ, Condon J, Hughes E, Neoh S-H, Sykes PJ, Seshadri R, Toogood I, Waters K, Tauro G, Ekert H, Morley AA. Outcome This study was funded by the Childhood Cancer Association prediction in childhood acute lymphoblastic leukemia by molecu- and the National Health and Medical Research Council. MJB lar quantification of residual disease at the end of induction. Lan- cet 1994; 343: 196–200. held a Peter Nelson Leukemia Research Fund Fellowship. We 12 Sykes PJ, Neoh SH, Brisco MJ, Hughes E, Condon J, Morley AA. thank Drs A Schwarer (Alfred Healthcare Group, Victoria), A Quantitation of targets for the PCR by use of limiting dilution. Grigg (Royal Melbourne Hospital, Victoria) B Webster and M Biotechniques 1992; 13: 444–449. Stevens (Children’s Hospital, Sydney, NSW) for contributing 13 Brisco MJ, Hughes E, Neoh SH, Sykes PJ, Bradstock K, Enno A, samples and clinical information. Szer J, McCaul K, Morley AA. Relationship between minimal residual disease and outcome in adult acute lymphoblastic leukemia. Blood 1996; 87: 5251–5256. References 14 Ralph QM, Brisco MJ, Joshua DE, Brown R, Gibson J, Morley AA. Advancement of multiple myeloma from diagnosis through pla- 1 Lestingi TM, Hooberman AL. Philadelphia chromosome-positive teau phase to progression does not involve a new B-cell clone. acute lymphoblastic leukemia. Hematol/Oncol Clin N Am 1993; Blood 1993; 82: 202–206. 7: 161–175. 15 Kelman Z, Prokocimer M, Peller S, Kahn Y, Rechavi G, Manor 2 Roberts WM, Rivera GK, Raimondi SC, Santana VM, Sandlund Y. Rearrangements in the p53 gene in Philadelphia chromosome JT, Crist WM, Pui CH. Intensive chemotherapy for Philadelphia- positive chronic myelogenous leukemia. Blood 1989; 74: 2318– chromosome-positive acute lymphoblastic leukemia. Lancet 1994; 2324. 343: 331–332. 16 Mashal R, Shtalrid M, Talpaz M, Kantarjian H, Smith L, Beran M, 3 Bedi A, Zehnbauer BA, Barber JP, Sharkis SJ, Jones RJ. Inhibition Cork A, Trujillo J, Gutterman J, Deisseroth A. Rearrangement and of apoptosis by BCR-ABL in chronic myeloid leukemia. Blood expression of p53 in the chronic phase and blast crisis of chronic 1994; 83: 2038–2044. myelogenous leukemia. Blood 1990; 75: 180–189. 4 Bedi A, Barber JP, Bedi GC, el-Deiry WS, Sidransky D, Vala MS, 17 Goker E, Lin JT, Trippett T, Elisseyeff Y, Tong WP, Niedzwiecki Akhtar AJ, Hilton J, Jones JR. BCR-ABL mediated inhibition of D, Tan C, Steinherz P, Schweitzer BI, Bertino JR. Decreased poly- apoptosis with delay of the G2/M transition after DNA damage: a glutamylation of methotrexate in acute lymphoblastic leukemia mechanism of resistance to multiple anticancer drugs. Blood blasts in adults compared to children with this disease. Leukemia 1995; 86: 1148–1158. 1993; 7: 1000–1004. 5 Maekawa T, Kimura S, Hirakawa K, Murakami A, Zon G, Abe T. 18 Romana SP, Poirel H, Leconiat M, Flexor MA, Mauchauffe M, Jon- Sequence specificity on the growth suppression and induction of veaux P, Macintyre EA, Berger R, Bernard OA. High frequency apoptosis by chronic myeloid leukemia cells by BCR-ABL anti- of t(12;21) in childhood B-lineage acute lymphoblastic leukemia. sense oligodeoxynucleoside phosphorothioates. Int J Cancer 1995; Blood 1995; 86: 4263–4269. 62: 63–69. 19 Shurtleff SA, Buijs A, Behm FG, Rubnitz JE, Raimondi SC, Hancock 6 d’Auriol L, Macintyre E, Gilbert F, Sigaux F. In vitro amplification ML, Chan GCF, Pui CH, Grosveld G, Downing JR. TEL/AML1 of T cell ␥ gene rearrangements: a new tool for the assessment of fusion resulting from a cryptic t(12–21) is the most common gen- minimal residual disease in acute lymphoblastic leukemias. Leuke- etic lesion in pediatric ALL and defines a subgroup of patients with mia 1989; 3: 155–158. an excellent prognosis. Leukemia 1995; 9: 1985–1989. 7 Hansen-Hagge TE, Yokota S, Bartram CR. Detection of minimal 20 Radich JR, Kopecky KJ, Boldt DH, Head D, Slovak ML, Babu R, residual disease in acute lymphoblastic leukemia by in vitro ampli- Kirk J, Lee A, Kessler P, Appelbaum F, Gehly G. Detection of BCR- fication of rearranged T-cell receptor ␦ chain sequences. Blood ABL fusion genes in adult acute lymphoblastic leukemia by the 1989; 74: 1762–1767. polymerase chain reaction. Leukemia 1994; 8: 1688–1695. 8 Yamada M, Wasserman R, Lange B, Reichard BA, Womer RB, 21 Voncken JW, Kaartinen V, Pattengale PK, Germeraad WT, Groffen Rovera G. Minimal residual disease in childhood B-lineage lym- J, Heisterkamp N. BCR/ABL P210 and P190 cause distinct leuke- phoblastic leukemia. New Engl J Med 1990; 323: 448–455. mia in transgenic mice. Blood 1995; 86: 4603–4611.