annotation
Defining the correct role of minimal residual disease tests in the management of acute lymphoblastic leukaemia Giovanni Cazzaniga,1 Maria Grazia Valsecchi,2 Giuseppe Gaipa,1 Valentino Conter3,4 and Andrea Biondi4 1
M.Tettamanti Research Centre, Paediatric Clinic, University of Milano-Bicocca, San Gerardo Hospital, Monza, 2Department of Clinical Medicine and Prevention, Centre of Biostatistics for Clinical Epidemiology, University of Milano-Bicocca, Monza, 3Department of Paediatrics, ‘‘Ospedali Riuniti’’, Bergamo, and 4Paediatric Clinic, University of Milano-Bicocca, San Gerardo Hospital/Fondazione MBBM, Monza, Italy
Summary Minimal residual disease (MRD) has acquired a prominent role in the management of childhood and adult Acute Lymphoblastic Leukaemia (ALL) for its high prognostic value. Several studies have demonstrated the strong association between MRD and risk of relapse in childhood and adult ALL, irrespective of the methodology used. MRD is now used in clinical trials for risk assignment and to guide clinical management. Negativity at early time points may be considered to decrease treatment burden in patients who are likely to be cured with reduced intensity regimens. On the other hand, high MRD levels at late time points (end of consolidation) define ALL subgroups which deserve investigation of more effective treatments. The predictivity of MRD as a measurement of drug response in vivo opened new perspectives for its use in clinical decision, to deliver risk-based treatments, and possibly as a surrogate for efficacy in the evaluation of novel therapeutic approaches. Keywords: acute lymphoblastic leukaemia, minimal residual disease, prognostic factors. After the initial demonstration that bone marrow samples in morphological remission contain measurable levels of leukaemia (‘minimal residual disease’ – MRD) (Bradstock et al, 1981), the use of MRD tests has become prominent in the management of acute lymphoblastic leukaemia (ALL) (reviewed in Campana, 2010). The main reasons for this development were the progressive improvement of standardized methodologies that were applicable to virtually all patients, and the achievements of clinical studies that used MRD evaluation as a marker of in vivo early response to
Correspondence: Andrea Biondi, Department of Paediatrics, University of Milano-Bicocca, San Gerardo Hospital/Fondazione MBBM, Via Pergolesi, 33, 20900 Monza, Italy. E-mail:
[email protected]
allocate patients to different risk-based treatments and to improve outcome.
Methodologies for MRD detection A reliable technique to detect MRD should have the following prerequisites: (i) sensitivity of at least 10)4 (one malignant cell within 10 000 normal cells), although the target value of sensitivity depends on the clinical question to be addressed by the MRD assessment; (ii) specificity, to discriminate between malignant and normal cells (to prevent false-positive results); (iii) be quantifiable within a large dynamic range; (iv) stability over-time of leukaemia-specific markers, to prevent falsenegative results, particularly in long-term studies; (v) reproducibility between laboratories (essential for multicentre trials); (vi) careful standardization and quality control checks; (vii) rapid availability of results (in time for clinical usefulness) (Cazzaniga et al, 2003; van der Velden et al, 2003, 2007a,b; Campana, 2008; Flohr et al, 2008). In ALL, the most reliable methods for MRD detection include flow cytometric (FCM) profiling of leukaemia-associated immunophenotypes and real-time quantitative polymerase chain reaction (RQ-PCR) amplification of fusion transcripts or Immunoglobulin (Ig) and T-cell receptor (TcR) genes. These approaches are now widely used for MRD monitoring in clinical trials because they have all the necessary requisites (Bruggemann et al, 2010). Table I summarizes the features and applicability of the main methods for MRD detection.
PCR-MRD monitoring RQ-PCR analysis of fusion transcripts. Somatic chromosome aberrations can be used as PCR targets for MRD detection because they are tumour-specific and stable during the disease course. However, most ALL cases do not have specific chromosome aberrations that can be detected by PCR (Table I). Moreover, there are several potential pitfalls to be taken into account, including the risk of RNA degradation and
ª 2011 Blackwell Publishing Ltd, British Journal of Haematology, 155, 45–52
First published online 4 August 2011 doi:10.1111/j.1365-2141.2011.08795.x
Annotation Table I. Features and applicability of the main methods for MRD detection in acute lymphoblastic leukaemia. Method
Target
Applicability
Sensitivity
Advantages
Disadvantages
Flow cytometry
Leukaemic immunophenotypes
95%
10)4
Widely applicable; rapid; accurate MRD quantification
PCR
Fusion gene transcripts
40%
10)4
Rapid; direct link with the leukaemic clone
Ig/TcR gene rearrangements
95%
10)5
Widely applicable; highly sensitive; accurate MRD quantitation
Operator-dependent: lack of appropriate training may lead to false-positive and false-negative results. Need for standardization Phenotypic shifts require the use of multiple markers Unknown number of transcripts per cell (uncertain quantitation of MRD). Need for standardization RNA degradation may produce false-negative results; risk of cross-contamination (false-positive results) Labourious and costly (because reagents are patient-specific). Need for experienced personnel and standardization Oligoclonality and clonal evolution may produce false-negative results
PCR, polymerase chain reaction; Ig, immunoglobulin; MRD, minimal residual disease; TcR T-cell receptor.
the need for checking in parallel an appropriate housekeeping gene. More importantly, the number of transcript molecules per cell cannot be defined, and thus a clear quantification of leukaemia cells is not possible. A consensus regarding the reagents and minimal requirements for MRD monitoring by RQ-PCR of fusion transcripts has been reached by several European laboratories collaborating on a ‘Europe Against Cancer’ programme (Gabert et al, 2003). Currently, the use of RQ-PCR of fusion genes is limited to Philadelphia chromosome-positive (Ph+) ALL. Early studies have shown that MRD level at the end of induction and after consolidation is a powerful indicator of prognosis (Pane et al, 2005; Wassmann et al, 2005). However, a more recent study of a large prospective series of adult Ph+ ALL treated with imatinib-combined chemotherapy showed that negative MRD at the end of induction therapy was not associated with longer relapse-free survival (RFS) or a lower relapse rate (Yanada et al, 2008). Moreover, among the 29 patients showing MRD elevation during haematological complete remission (CR), 10 of the 16 who had undergone allogeneic haematopoietic stem cell transplantation (HSCT) in first CR were alive without relapse, whereas 12 of the 13 who had not undergone allogeneic HSCT experienced a relapse. Taken together these findings indicate that RQ-PCR of BCRABL1, represents an important tool for the development of novel treatment concepts in MRD-based therapy protocols of Ph+ALL, as currently ongoing in paediatric and adult trials. RQ-PCR analysis of Ig/TcR gene rearrangements. Ig/TcR gene rearrangements can be used as universal PCR-MRD targets. In combination with fluorescent probes, Allele-Specific Oligonucleotide (ASO) primers can be designed complementary to the patient- and tumour-specific junctional region sequence of each target. A large standardization, quality control and guidelines on RQ-PCR analysis and interpretation have been 46
assessed within the ‘Euro-MRD’ group (previous ‘European Study Group for MRD detection in ALL’, ESG-MRD ALL) (van der Velden et al, 2007a). Equally important is the care on the logistics of MRD monitoring in large prospective multicentric clinical trials, including shipping and processing of patient samples, MRD measurement in centralized reference laboratories as well as rapid data transfer to the study management (van der Velden et al, 2003, 2007b; Flohr et al, 2008). FCM-MRD monitoring. Leukaemia-associated immunophenotypes can be assessed and quantified by using multicolour flow cytometry to distinguish leukaemic lymphoblasts from their normal counterpart (Coustan-Smith et al, 1998; Dworzak et al, 2002; Basso et al, 2009). The feasibility of a multicentric standardization of FCM-MRD measurement in ALL, consisting of technical alignment (sample preparation, reagent selection, instrument set-up), personnel education (data analysis and interpretation), and continued quality control, has been demonstrated by the Associazione Italiana di Ematologia e Oncologia Pediatrica (AIEOP)-Berlin-Frankfurt-Mu¨nster (BFM)-ALL-FCM-MRD-Study Group (Dworzak et al, 2008). The achievable quantifiable range by FCM-MRD can generally be defined as the ability to detect 30 clustered MRD events in 3 · 105 total cellular events (0Æ01%). However a cluster of at least 10 events with leukaemia-associated immunophenotype and back-gating light scatter can be sufficient to define a sample as ‘MRD-positive’ (Dworzak et al, 2008). Combined use of PCR- and FCM-MRD. The combined application of Ig/TcR PCR- and FCM-MRD on more than 1500 childhood ALL patients enrolled within the multicentric AIEOP-BFM ALL 2000 trial showed an overall concordance of 80%, although different results were observed according to the various follow-up time points (from 70% concordance at day
ª 2011 Blackwell Publishing Ltd, British Journal of Haematology, 155, 45–52
Annotation 33, to more than 85% at days 15 and 78) (Gaipa et al, 2008). The combined analysis suggests a potential complementary role of the two technologies in optimizing risk stratification in future clinical trials. FCM is more time and cost effective whereas PCR enables higher sensitivity to be achieved. Thus, FCM or PCR, or both can be used in MRD-based treatment protocols depending on the available expertise, resources and specific protocol design. Critical issues in MRD studies. Several parameters are critical for the conception and interpretation of MRD studies, including therapeutic context, specific features of the target population, timing of sampling, target markers, required sensitivity of the assay, inter-laboratory standardization (particularly relevant in multicentric studies), and the retrospective or prospective design of the study. Another critical issue is the discordance between PCR and FCM results; this may be mostly due to the limited number of total cells available in the FCM test-tube, which reduces the sensitivity and thus the accuracy of FCM at low MRD levels (below 10)4). However discrepancies (qualitative or quantitative), can be also explained by different factors affecting either PCR or FCM, such as, (i) quality of clonal PCR-markers; (ii) nonspecific amplification of normal DNA; (iii) oligoclonality and/ or clonal evolution; (iv) age–related (Lucio et al, 1999) or therapy–related BM B-cell precursors regeneration status (Dworzak et al, 1997; van Wering et al, 2000); (v) immunophenotypic changes between diagnosis and relapse (van Wering et al, 1995; Borowitz et al, 2005), (vi) druginduced immunophenotypic modulation (Gaipa et al, 2005).
MRD in childhood and adult ALL: where are we? Several studies in childhood and adult ALL identified MRD as the most relevant and independent prognostic factor for the duration of CR (reviewed in Campana, 2010). The current status of European ALL trials was comprehensively discussed in a recent review (Bruggemann et al, 2010). Childhood ALL. There is a close association between the quality of the molecular remission, i.e. clearance of leukaemia blasts, and the final outcome, independently of the applied treatments in childhood ALL. It is still unknown why the exposure to drugs during the early phases of treatment (induction or consolidation) produces different in vivo chemosensitivities, which influence the final treatment outcome. Several attempts have been made to correlate MRD with gene expression profiles of leukaemic cells (Flotho et al, 2007; Bhojwani et al, 2008; Kang et al, 2010) or to germline or leukaemia-associated gene polymorphisms (Stanulla et al, 2005; Yang et al, 2009). Whether MRD studies could eventually be replaced by novel risk factors based on presenting features is still unknown, but it is unlikely to
occur because MRD measurements reflect the combination of leukaemia-presenting features as well as the efficacy of the therapy. In the largest prospective study of 3184 B-lineage ALL patients enrolled in the AIEOP-BFM ALL 2000 protocol, MRD response, detected with a highly sensitive and well-standardized PCR technique at the end of induction (day 33; timepoint [TP]1) and consolidation (day 78; TP2), was highly predictive of relapse, thus markedly reducing the importance of conventional prognostic factors, such as age, white blood cell count at diagnosis, genetic abnormalities (ETV6/RUNX1positivity and DNA index) and also prednisone response, which is a measure of early in vivo sensitivity used in BFM protocols (Conter et al, 2010). These findings confirm and further extend previous reports on more limited retrospective or prospective series (reviewed in Campana, 2009). In the AIEOP-BFM ALL 2000 study, unfavourable cytogenetics [hypodiploidy, t(4;11) and t(9;22) translocations] still retained independent prognostic value. In patients with favourable cytogenetics [such as t(12;21) translocation or hyperdiploidy], MRD can be used to investigate the safety of treatment reduction. MRD negativity at the end of induction is the strongest predictor for excellent outcome [5-year event-free survival (EFS) >90%]. Interestingly, among 1358 patients who presented this feature, only 61 relapsed, of whom 29 presented with an extramedullary relapse, either isolated (n = 17) or combined (n = 12), which may not be likely to be predicted by bone marrow MRD response, especially for the cases with isolated relapse (Conter et al, 2010). MRD positivity at the end of induction, even at very low levels (