Routine application of a novel MLPA-based first-line screening test uncovers clinically relevant copy number aberrations in haematological malignancies undetectable by conventional cytogenetics

Abstract

Objective

The presence of numerical and/or structural chromosomal abnormalities is a frequent finding in clonal hematopoietic malignant disease, typically diagnosed through routine karyotyping and/or fluorescent in situ hybridization (FISH) analysis. Recently, the application of array comparative genomic hybridization (aCGH) has uncovered many new cryptic genomic copy number imbalances, most of which are now recognized as clinically useful markers of haematological malignancies. In view of the limitations of both FISH and aCGH techniques, in terms of their routine application as a first line screening test, we designed a new multiple ligation-dependent probe amplification (MLPA) probemix for use in addition to classic karyotype analysis.

Methods

A novel MLPA probemix was developed to interrogate copy number changes involving chromosomal regions: 2p23-24 (MYCN, ALK), 5q32-34 (MIR145A, EBF1, MIR146A), 6q21-27, 7p12.2 (IKZF1), 7q21-36, 8q24.21 (MYC), 9p24 (JAK2 V617F point mutation), 9p21.3 (CDKN2A/2B), 9p13.2 (PAX5), 10q23 (PTEN), 11q22.3 (ATM), 12p13.2 (ETV6), 13q14 (RB1, MIR15A, DLEU2, DLEU1), 17p13.1 (TP53), and 21q22.1 (RUNX1/AML1) and was applied to DNA extracted from 313 consecutive bone marrow patient samples, referred for routine karyotype analysis.

Results

More than half of the samples originated from newly investigated patients. We discovered clinically relevant genomic aberrations, involving a total of 24 patients (8%) all with a normal karyotype, which would have remained undiagnosed.

Discussion

Our data clearly indicate that routine application of this MLPA screening panel, as an adjunct to karyotype analysis, provides a sensitive, robust, rapid and low-cost approach for uncovering clinically important genomic abnormalities, which would have otherwise remained undetected.

Keywords:

• Haematological malignancies
• Leukaemia
• Multiple ligation-dependent probe amplification
• Copy number aberrations

Introduction

Karyotype analysis in haematological disorders has traditionally been a valuable first-line test for disease diagnosis and classification as well as for prognosis during treatment and monitoring of patients. Over the years, this ‘whole genome scan’ at ∼5–10 Mb resolution, has led to the discovery of new bio-markers, which typically involve large-scale chromosomal imbalances and/or recurrent translocations. However, conventional chromosome analysis in haematological malignancies has several technical limitations, a very important one being the inability to detect sub-microscopic genomic aberrations. Also karyotyping can be sometimes difficult due to the low mitotic index of leukemic cells or due to the low quality of the metaphases, and in this case an adjunct genetic analysis with another technique is needed.

Timely detection of chromosomal abnormalities in leukemias is of great importance, affording accurate diagnosis, prognosis and monitoring of residual disease.¹ Traditionally, the application of fluorescent in situ hybridization (FISH) in haematological investigations has enabled detection of recurrent chromosomal alterations, mostly balanced translocations, in interphase nuclei and metaphase chromosomes. Although the method has proven extremely useful in this field, it generally requires prior detailed clinical information or a differential diagnosis to guide the appropriate choice of probes to be used, which is often not the case for newly investigated patients. Furthermore, in clinical practice FISH is a laborious and costly procedure, highly selective (4–5 loci interrogated at one time), with a low throughput and is only capable of detecting imbalances of sequences larger than 50 kb.^2–5 Recently, the application of array comparative genomic hybridization (aCGH) has provided extremely valuable clinical data through its implementation in haematological malignancies.^6–12 However, the routine application of aCGH in a clinical setting, as a routine first-line test in haematological malignancies, is still hampered by relatively low throughput, the inability to detect balanced chromosomal aberrations, high costs and the need for careful and expert data analysis, and is therefore limited to specific patient–disease cohorts and/or follow-up studies.

Multiplex ligation-dependent probe amplification (MLPA)¹³ is a polymerase chain reaction (PCR)-based technique, which allows the simultaneous detection of small scale genomic imbalances across multiple loci (>50) in a single reaction generating results within 24 hours and it has been successfully applied in the investigation of a wide variety of genetic disorders. A number of published studies have described the application of specific MLPA probe mixes for haematological malignancies, e.g. chronic lymphocytic leukaemia (CLL),^14–18 for acute lymphoblastic leukaemia (ALL),^18–21 for myelodysplastic syndrome (MDS)²² and have thereby demonstrated the sensitive and accurate detection of disease-specific clinically significant copy number changes. Furthermore, a high degree of concordance between MLPA and FISH has been established, provided that the abnormality is present in >20% of the cells.^{14,16,18,19,23,24}

In our efforts to increase the diagnostic yield provided by standalone karyotype analysis, in a timely and cost-effective manner, we developed a novel MLPA probemix targeting genomic loci involved in a variety of haematological malignancies (not disease-specific), with an aim to be used as a high-throughput adjunct to classic karyotype analysis. In this study, we report our findings and experience with the newly designed probe mix from screening 313 consecutive patient samples, the majority of which were newly investigated patients awaiting definitive diagnosis.

Materials and methods

Patient cases and samples

Following the initial evaluation period, parallel MLPA analysis was offered and requested by all referrals, from June 2010 to October 2011, as an addition to standard bone marrow or peripheral blood karyotype analysis. Samples analysed during an initial 4-month evaluation period of the new P377 MLPA probemix are not included in the study. Approximately 56% (175) of all referrals involved newly investigated patients, for a haematological abnormality awaiting definitive diagnosis. Indications were recorded as stated on the test requisition form, completed by the referring physician, typically broad and not very well defined.

A summary of all the specimens analysed and the recorded clinical indications is presented in Table 1.

Table 1. Cases studied according to primary clinical indication

Karyotype and FISH analysis

Karyotype analysis, as part of the routine cytogenetics service, was performed following short-term (24–48 hours) flask culture of bone marrow cells or peripheral blood (72 hours) using established R-banding protocols. Standard interphase FISH analysis was performed using commercially available probes (Vysis, Abbott Molecular, Des Plaines, IL, USA) or appropriately custom labelled BAC clones selected from the chromosomal region harbouring the abnormality (Empire Genomics LLC, Buffalo, NY, USA, Bluegnome Ltd., Cambridge, UK).

MLPA analysis and data evaluation

DNA were extracted from a fraction of the cultured cell pellets using the QIAamp DNA Blood Mini Kit (Qiagen, CA, USA) and subjected to MLPA according to the standard protocol with minor modifications. We chose to extract DNA from the same sample used for karyotype analysis, as the results of the two tests are evaluated in parallel. The new probemix, P377-A1, includes 54 MLPA probes for the detection of copy number abnormalities involving the following chromosomal regions/genes: 2p (MYCN, ALK), 5q (MIR145, EBF1, MIR146A), 6q, 7p12 (IKZF1), 7q, 8q24 (MYC), 9p (MTAP, CDKN2A, CDKN2B, PAX5), 9p (JAK2 V617F mutation-specific probe), 10q23 (PTEN), 11q23 (ATM), 12p (ETV6), 12q, 13q (RB1, MIR15A, DLEU2, DLEU1), 17p (TP53), 17q, chr 18 and chr 19, 21q22.1 (RUNX1). A detailed description of the probemix is presented in Table 2.

Table 2. Genomic regions interrogated by new MLPA probemix P377

Patient samples, in parallel with a ‘standard’ control DNA consisting of a mixture of 10 normal DNA samples, were analysed on an ABI 3730xl automated sequencer (Applied Biosystems, Foster City, CA, USA) and data were exported to the GeneMarker software (Softgenetics LLC, PA, U.S.A.) for copy number evaluation, with population normalization of peak height ratios and cut-off values of 0.8 for deletion and 1.20 for duplication detection.

Samples exhibiting probe imbalances were re-analyzed, using the same DNA sample, through subsequent MLPA reactions with different probe mixes, where available, containing a multitude of probes for the aberrant chromosomal region.

Results

Samples and clinical indications

The vast majority of samples analysed (301 of 313, 96%) were bone marrow samples, the remaining being peripheral blood samples. The clinical indication for each sample (Table 1) was recorded as stated in the accompanying test requisition form, often broad and not well defined. The majority of samples (174 of 313, 56%) originated from newly investigated patients, with an indication of a (suspected) haematological malignancy awaiting definitive diagnosis. Furthermore, approximately half the samples (155 of 313, 49%) had as sole indication an abnormal result, following routine haematology testing (e.g. elevated platelets count, and elevated WBC).

MLPA design rationale

The new MLPA probemix contains probes for several genes and chromosomal regions known to have a diagnostically or prognostically significant role in haematological malignancies (Table 2). The probe mix was designed based primarily on: (a) several published reports, mostly from aCGH studies, indicating the presence of several clinically relevant copy number gains and losses in leukaemia patients (e.g. RUNX1, ETV6, IKZF1(IKAROS), EBF1, PAX5, see below); (b) our own previous experience from approximately 300 cases as well as the results from published studies describing MLPA analysis in CLL, ALL, and MDS patients (see Introduction for references); and (c) the concept that this specific MLPA probemix is intended to be used in combination with standard karyotype analysis, for screening patient DNA samples (first-line test) for common and diagnostically significant copy number changes associated with haematological malignancies, including ALL, acute myeloid leukaemia (AML), CLL, chronic myeloid leukaemia, MDS, and various lymphomas.

For example, loss of 5q, especially 5q32-34, is a frequent event in MDS and in AML. Recently, MIR145 and MIR146a were shown to be key mediators of the 5q syndrome,²⁵ while deletions of EBF1 have been reported to be more frequent in high-risk ALL cases.^26–28

Deletions of the IKZF1 (Ikaros) gene are detected in ALL, especially in cases that also carry the BCR-ABL1 gene fusion.²⁹ IKZF1 deletions in ALL have been associated with relapse and very poor clinical outcome³⁰ and deletions of IKZF1 might also be involved also in other hematologic malignancies, since Ikaros proteins are active throughout human B-cell differentiation.³¹

Deletions at 9p21.3, encompassing the tumour suppressor genes CDKN2A and CDKN2B genes, encoding p16(INK4a), p14(ARF) and p15(INK4b), are especially frequent in ALL, occurring at >20% in B-cell precursor ALL and approximately 50% T-ALL patients.³² The 9p21.3 deletions have been suggested to associate with unfavourable outcome in both adult ALL and paediatric ALL, although the prognostic impact of CDKN2A deletions in pediatric ALL appears controversial.^32–44 Deletions of 9p21.3 are also detected in other hematologic malignancies.

9p13.2 (PAX5) deletions are frequent in B-ALL and in BCR-ABL1-positive ALL cases,^19,45 and PAX5 deletions in ALL can be large and extend sometimes into the CDKN2A/2B genes. Amplifications of PAX5 exon 2 and 5 have been suggested to be an alternative mechanism of PAX5 inactivation.^19,26,46

ETV6 (12p13.2) deletions are detected in childhood ALL and in AML/MDS with normal karyotype.⁴⁷ ETV6 deletions are also frequently associated with leukemic transformation of Philadelphia chromosome-negative myeloproliferative neoplasms.⁴⁸

Finally, amplifications of RUNX1 at 21q22.1, mostly due to intrachromosomal amplification of chromosome 21 (iAMP21), have been reported in childhood ALL and clinical studies have shown that ALL patients with RUNX1 copy gains have an increased risk of relapse and have significantly inferior survival compared to patients without this genetic change.^20,49–51

A mutation-specific probe for the detection of the JAK2 V617F mutation was also included, as its presence is quite common in myeloid malignancies in AML, MDS and myeloproliferative neoplasms, with the highest frequency (∼95%) occurring in polycythemia vera.^52,53

The reliable detection of genomic aberrations in malignancies, such as those commonly encountered in leukaemia, depends on the level of mosaicism, i.e. the percentage of cells harbouring the abnormality relative to the normal cells. Therefore, during the validation phase each probe was carefully selected so that the average probe-specific variation, using the GeneMarker software population normalization, was 1.00 ± 0.1 for normal samples. In addition, each probe was tested for (a) compatibility of probes within the mix, (b) sensitivity to sub-optimal hybridization temperature, (c) sensitivity to polymerase activity, (d) sensitivity to salt in DNA samples, (e) compatibility of probes with various DNA concentrations (20 ng/500 ng), and (f) sensitivity of probe signals to evaporation during the hybridization and PCR. All probes included in this probemix fulfilled the above criteria. This careful validation was particularly important due to the absence of reference probes in the probemix. The absence of reference probes was justified with the following reasoning: this probemix contains probes for more than 10 different loci, all on different chromosomes, and so the probability that more than 25% of all probes included in the mix are affected by a copy number change per sample is extremely low. The availability of the karyotype results aided significantly the analysis and provided extra reassurance. Our normal range for population normalization was subsequently set to a cut-off of 0.8 for deletion and 1.2 for duplication, allowing the detection of mosaicism at >25% and this was true for all probes in the probemix.

Karyotype analysis

A karyotype result was not available for 4 of the 313 samples (1%), due to absence of mitotic activity, but MLPA analysis was successful in 3 of the samples. An abnormal karyotype result was recorded for 74 specimens (24%), which included 18 cases with an abnormal karyotype coupled to an abnormal MLPA result (due to a common gross chromosomal abnormality). The remaining 56 cases with abnormal karyotype harboured chromosomal abnormalities not detectable by the probemix (e.g. balanced translocations, complex rearrangements, and other trisomies). Finally, 210 samples (67%) had normal results from both karyotype and MLPA analysis.

MLPA analysis

MLPA analysis was performed on DNA extracted from short-term cultured cells, except in three cases where there was culture failure and DNA was extracted directly from the primary sample. Although acceptable MLPA results were also obtained in our initial trials from DNA of uncultured samples, it was decided that optimal diagnostic sensitivity would best be obtained following short-term culture. Fully interpretable MLPA results were obtained in 308 of the 313 samples, as 5 samples harboured multiple chromosomal abnormalities interfering with MLPA normalization.

Most importantly, clinically relevant genomic aberrations were detected in 24 patient samples (8%), all with normal karyotype or with unrelated karyotypic abnormalities, which would have remained otherwise totally undetected (Table 3).

Table 3. Genomic aberrations detected by the new MLPA probemix, all with normal karyotype or unrelated karyotypic abnormalities

Major findings in this group included:

Deletion of 9p21.3: CDKN2A/2B was the most common finding, present in a total of seven cases (7 in 24, 29%), four of which had bi-allelic loss and were confirmed by FISH. All of these patients, four of which were paediatric, were newly investigated referred with an indication of AL or possible ALL. It is noteworthy that in one case there was no karyotype result available due to the absence of mitotic activity.

Copy gain or loss of 21q22.1: RUNX1(AML1) was the second most common finding (6 in 24, 25%). A gain was detected in 4 cases (21%), all newly investigated paediatric patients with an indication of acute leukaemia. All results were confirmed by repeat analysis using MLPA probemix (P327-iAMP21-A1) containing multiple probes for the region. Interestingly, we also observed two heterozygous micro-deletions of the region in patients with an indication of possible MDS. Although loss of RUNX1 is not a common finding, inactivating gene mutations have been associated with AML and copy loss of RUNX1 in mice has been shown to lead to leukemic transformation⁵⁴ and RUNX1 has been shown to be deleted in a AML patient at relapse following a bone marrow transplantation.⁵⁵

Deletion of 12p13.2: ETV6 was detected in four cases and two of these patients harboured additional genomic aberrations. All were referred with an indication of acute disease .The deletion was confirmed by repeat MLPA analysis with a probemix (P335) containing additional probes for ETV6. The additional abnormalities in the two patients involved deletion of 7p12.2-Ikaros (IKZF1) and deletion of 17p13-TP53, both associated with aggressive disease. The deletion was confirmed by repeat MLPA analysis with a probemix (P335) containing additional probes for ETV6.

In case 23 (Table 3), a newly investigated child with an indication of AL, we detected a copy gain of 6q21-q27, a region harbouring the MYB oncogene. Deletions of 6q are commonly found in various lymphoid malignancies, while reciprocal MYB duplications have been described primarily in T-ALL.⁵⁶ Other common copy number changes, for example in 13q14, 11q23, and 8q24, known to occur primarily in CLL, were also detected. It is of interest that in one out of the three cases with a 13q14 deletion (Table 3, case 24), the extent of the deletion was smaller and did not include the RB1 gene, and this would have been missed by FISH analysis. It is worth noting that the extent of the deleted area in 13q14 correlates with the extent of treatment, as patients with small deletions require longer treatment.

All of the detected aberrations presented with MLPA copy number ratios ranging from 0.2 ± 0.12 (called as bi-allelic deletions) to 0.6 ± 0.16 (called as apparently heterozygous deletion) and 1.4 ± 0.15 (called as copy number gain), relative to the control sample. It is beyond the scope of this study to establish concordance between FISH and MLPA, as this has already been proven through numerous published studies (see Introduction). However, all the bi-allelic deletions, occurring exclusively at 9p21.3, were confirmed as such by FISH analysis, while the remaining aberrations were confirmed either by FISH or through repeat analysis with different MLPA probemixes containing multiple probes in the affected regions.

Discussion

The data presented herein extend previous evidence and provide additional proof that MLPA is a highly robust, useful and cost-effective technique for the characterization of known recurrent genomic lesions occurring in haematological malignancies. Moreover, the development of this new probemix permits its utilization as a first line screening test in all patient samples being investigated for a haematological malignancy, irrespective of indication. It interrogates imbalances occurring potentially in a variety of leukemias, unlike previous MLPA studies which describe applications targeted for CLL, ALL, or MDS patients.

In our unselected cohort of patients, 24% had an abnormal karyotype and 8% had an abnormal MLPA undetectable by standard karyotype analysis. From our data, it is apparent that at least 1 in 15 patients would have received a completely normal karyotype result, whereas a clinically significant genomic abnormality was present. Concomitantly, most probably they would not be afforded an accurate assessment in terms of proper diagnosis, risk stratification and prognosis, especially since the majority are newly diagnosed cases. Published studies have suggested the application of MLPA, for example in CLL or MDS, as a standalone test without karyotype analysis.¹⁷ However, close inspection of our results shows that at least 51 patients, with an abnormal karyotype exhibiting gross chromosomal abnormalities not interrogated by MLPA, would remain undetected. Combined analysis through standard karyotype coupled to the MLPA panel affords a detection level of 32% for clinically important chromosomal anomalies, at least in our series of patients. Bearing in mind the fact that more than half of the patients were newly diagnosed cases, it is even more important that they receive, within reason, the highest possible level of diagnostic sensitivity.

The MLPA technique by itself is not designed to optimally detect differences in ploidy owing to inherent limitations in the technology, while genomic copy number changes present within a small fraction of the (mosaic) aberrant cell population might also be missed. However, the presence of low-level mosaicism in the vast majority of cases will be missed by most molecular techniques, including aCGH, unless prior knowledge of the case under investigation points to the application of a specific FISH panel. While the detection of low level mosaicism (<20%) may be viewed as a potential drawback, our screening approach, which utilizes MLPA coupled to karyotype analysis, overcomes to a large extent these limitations. The substantially reduced cost relative to FISH allows the assay to be used routinely in conjunction with karyotyping, affording the simultaneous detection of gross numerical chromosomal aberrations as well as sub-microscopic genomic imbalances. Additionally, as FISH analysis is limited by the number of probes used in a single assay, it will miss many important prognostic markers. Furthermore, the availability of complementary MPLA probe mixes containing multiple probes in the affected regions permits further confirmation of the results as well as fine-mapping of the alterations, not feasible by FISH.

The application of aCGH in the investigation of haematological malignancies has been described, albeit in rather limited and highly selected series of patients.^{9–11,57–59} Although direct comparison between our results with the combined results from aCGH studies is not strictly accurate, it is nonetheless evident that only a very small number of clinically significant recurrent genomic aberrations would have been missed by our approach. At the same time, aCGH would have missed, for example, a significant number of important recurrent balanced chromosomal rearrangements (translocations, inversions) detectable by karyotype analysis. Taking into consideration the above, as well as the relatively high cost of aCGH, it does not appear that routine application of aCGH as a standalone first line screening test for all samples can be justified in a clinical setting. However, ongoing research efforts utilizing aCGH may uncover novel recurring genomic aberrations with clinical significance, which may then be easily incorporated into MLPA panels or in the design of novel FISH probes.

In summary, first-line screening by MLPA in this patient cohort combined with karyotyping provides a substantial increase in diagnostic yield by uncovering clinically significant recurrent genomic abnormalities, which would have remained otherwise undetected by classic karyotype analysis. The approach described herein may provide a rapid screening test for known recurrent copy number abnormalities, particularly in patients without a specific disease diagnosis, and is designed to supplement classic karyotype analysis. It is cost-effective, robust, and rapid and has several advantages over FISH and aCGH in terms of throughput, cost and routine clinical application in all haematological samples. Finally, drawing from the experience of numerous published studies over the years, the number of clinically significant recurrent genomic abnormalities occurring in haematological malignancies will most probably remain within the capability of routine MLPA assay detection, by including probes for new loci, thus complementing and facilitating targeted confirmatory and follow-up studies with FISH or other emerging high-resolution molecular techniques.

Acknowledgements

The authors wish to thank K. Pispili for valuable technical assistance in the MLPA analysis. C.K. conceived and designed the study, analysed the data and drafted the manuscript. S.S. and J.P.S. contributed to the design of the study, contributed essential reagents and contributed to the drafting of the manuscript. S.K., A.M., M.K., and B.H. performed cytogenetics and MLPA assays. S.P., M.A., M.V., N.A.V., E.V., A.K., and K.Z. contributed in the collection and characterization of patient samples and critical reading, C.P. coordinated the study and contributed to the drafting of the manuscript. All authors approved the submitted manuscript.