Mitochondrial D-loop variations in paediatric acute myeloid leukaemia: a potential prognostic marker


Dr Sameer Bakhshi, Associate Professor of Paediatric Oncology, Department of Medical Oncology, Dr. B. R. A. Institute Rotary Cancer Hospital, All India Institute of Medical Sciences, New Delhi, India.


The D-Loop region of mitochondrial DNA (mtDNA) is the regulatory region for its replication and transcription. There are two hypervariable regions (HV-I, HV-II) and the rate of mutation in these regions is 100- to 200-fold that of nuclear DNA. In the current study, the entire D-loop region of mtDNA was amplified in two overlapping polymerase chain reaction fragments and variations were evaluated in 44 paediatric acute myeloid leukaemia (AML) patients by direct DNA sequencing methods. Median age of the patients was 8·5 years (1–18 years) and the male:female ratio was 3·8:1. A total of 222 variations were observed at 118 positions in the D-Loop of 35/44 (79·5%) AML patients. The most common variations were T→C (24·6%) and C→T (21·4%) followed by A→G (15·8%). There was no significant difference in the event-free survival (EFS) of patients with or without any variations (P = 0·40). Three variations in HV-I, namely 16126T→C (P = 0·05), 16224T→C (P < 0·01) and 16311T→C (P < 0·001), were significantly associated with inferior EFS. In conclusion, this is the largest study to show a high frequency of mtDNA variations in paediatric AML and their potential relevance as a prognostic marker in this disease.

Mitochondrial DNA (mtDNA) is a 16569 base pair circular, maternally inherited double stranded DNA encoding 13 genes for respiratory chain subunits. It represents <1·0% of total cellular DNA with no introns and histones (Anderson et al, 1981; Taanman, 1999). The number of mitochondria and mtDNA varies in different tissues. The mtDNA content may be either increased or decreased in various cancers in comparison with controls (Lee et al, 2005). Mitochondria lack histone proteins and an efficient DNA repair mechanism; further, the presence of reactive oxygen species (ROS) and an oxidative environment in the mitochondrial matrix makes mtDNA prone to mutations. The mutation frequency in mtDNA is about 10- to 20-fold greater in comparison to nuclear DNA (Johns, 1995; Grossman & Shoubridge, 1996). The D-loop (1120 bp) region of mtDNA contains regulatory elements involved in its replication and thus mutations in the D-loop may affect mtDNA copy number. This region has been shown to be a mutational ‘‘hot spot’’ in human cancers wherein there are two hyper variable (HV) sites, HV-I (16024–16383) and HV-II (57–372) (Stoneking, 2000). This region regulates transcription and replication of mtDNA and this instability in the mtDNA region may be involved in carcinogenesis.

Acute myeloid leukaemia (AML) is a hematopoietic malignancy that often exhibits nonrandom chromosomal translocations. Structurally altered genes play important roles in cell proliferation, differentiation and gene transcription. In addition to these nuclear genetic changes, non-chromosomal, mitochondrial mutations may play a role in the progression of adult AML and myelodysplastic syndrome (MDS) (Linnartz et al, 2004). None of these studies have correlated somatic mitochondrial variations with outcome. This study aimed to evaluate mtDNA variations in the D-loop region of mtDNA in childhood AML and correlate the same with outcome.

Materials and methods

Patients sample and DNA isolation

Newly diagnosed AML patients ≤18 years of age treated at our cancer centre were selected prospectively for the study between February 2007 and January 2009. Patients were induced with daunorubicin 60 mg/m2 for 3 d and cytosine arabinoside 100 mg/m2 as continuous infusion for 7 d. After achieving remission the patients received three cycles of high dose cytosine arabinoside at 18 mg/m2. Prior to initiating therapy an informed consent was taken for the evaluation of peripheral blood or bone marrow for mtDNA variations. The study was approved by the institute ethics committee. Total DNA was isolated from 200 μl of whole peripheral blood or bone marrow samples using the QI-Aamp DNA Mini-Kit (Qiagen, Hilden, Germany) and its quality and quantity assessed by spectrophotometry.

Nested polymerase chain reaction (PCR)

The D-loop region was amplified by nested PCR instead of conventional PCR to increase the specificity. Outer primers were used to amplify the whole D-loop region (1120 bp) to 1298 bp using the forward primer OUTER F1 (5′- GCCCATACCCCGAAAATGTTG-3′) and reverse primer OUTER R1 (5′-GGTAGAACTGCTATTATTCATCC-3′) (Fig 1A). A total of 500 ng of genomic DNA was used for first round of nested PCR. The resulting PCR products were fractionated by 1·5% agarose gel. For evaluation of variations in the D-loop region, two PCR products at 855 bp (Fig 1B) and 674 bp (Fig 1C), were amplified from within the 1298 bp product using overlapping primers D1F 5′CTGTTCTTTCATGGGGAAGC3′and D1R 5′GCTGTGCAGACATTCAATTGTT3′; D2F 5′GAGCTCTCCATGCATTTGGT3′ and D2R 5′GGGGATGCTTGCATGTGTA3′. PCR conditions for amplification for the whole D-loop region were: initial denaturation at 94°C for 3 min; 35 cycles of denaturation at 94°C for 45 s; annealing at 58°C for 1 min; elongation at 74°C for 30 sThe PCR master mix contained 1·25 mmol/l of each dNTP (Fermentas,Glen Burnie, MD, USA), 20 pmol of each primer, 10 mmol/l Tris–HCl (pH 9), 50 mmol/l KCl, 1·5 mmol/l MgCl2 and Taq DNA polymerase (5 u/μl) (New England Biolabs, Ipswich, MA) in a total volume of 50 μl. PCR conditions for internal fragments of the D-loop were as follows: initial denaturation at 94°C for 3 min; 35 cycles of denaturation at 94°C for 30 s; annealing at 56°C for 30 s; elongation at 72°C for 45 s with a final extension at 72°C for 8 min (MJ Research DNA DYAD Thermal Cycler, Germany). PCR products were purified using a PCR purification kit (Qiagen) and separated on a 2·0% agarose gel.

Figure 1.

 Nested PCR for D loop region. (A) First round PCR of paediatric AML patients with outer primers (1298 bp). Lane 1, negative control, lane 2–6, AML patients. (B) Second round nested PCR for D1 region of 854 bp: Lane 1, negative control and lane 2–4, AML patients. (C) Second round nested PCR for D2 region of 674 bp of AML patients (lanes 1, 3 and 4). L = molecular weight ladder.

Nucleotide base sequencing

PCR product was eluted from the gel using a column based kit (Qiagen). The eluted purified template was quantified and DNA sequencing was performed (ABI 3730 XL; Applied Biosystems Instruments, Foster City, CA, USA). DNA sequencing was possible for up to 850–1000 bp in one reaction; as the D-loop (1120 bp) was sequenced in two overlapping fragments, the entire region of the D-loop was sequenced. Sequences and chromatograms obtained were examined by chromas pro bioinformatics sequence analysis software and aligned by Basic GeneBee Clustal W 1.83 tools ( All sequences were compared with Ensemble database (, mitochondrial database of Department of Genetics and Pathology, Uppsala University, Sweden ( and Cambridge mitomap (

Statistical analysis

The association of mitochondrial variations and patients’ baseline characteristics (white blood cell count [WBC], platelet count and cytogenetics) was analysed using Fisher’s exact test. The median WBC and platelet count were used as the cut-off values. Variations in the D-Loop region of mtDNA were correlated with median survival of patients. We considered only those variations present in at least four patients (10·0% of the studied population). Kaplan–Meier curves were obtained for survival analysis and followed by log rank test. We also assessed the difference in survival pattern of patients with or without any variation. Multivariate Cox proportional hazard model was applied for those D-loop variations variables that showed significance for event-free survival (EFS) (P ≤ 0·05) in univariate Cox regression model. The stepwise selection procedure in the multivariate Cox model was implemented by selecting the variables with entry probability (0·05) and removal probability (0·051). The model survival probability was predicted on the basis of number of significant variations present in the Cox model corresponding to different survival times. All analysis was implemented on Stata 9·2 software package (Stata Corp, College Station, Texas, United States). P value <0·05 was considered statistically significant.


A total of 52 de novo paediatric AML patients were registered during the study period; 6 of these patients did not receive treatment and 2 patients had an inadequate sample for any analysis. Thus, the study was performed on 44 subjects. The median age was 8·5 years (1–18 years); male:female ratio was 3·8:1 (Table I). Total of 222 variations were found at 118 positions in the D-Loop of 35/44 (79·5%) AML patients (Table II), with a median of 5 variations per patient (range: 0–14). These variations were not significantly associated with WBC (P = 0·45), platelet count (P = 0·45) and standard cytogenetics (P = 0·79). All variations were single nucleotide substitutions. The most common variations were T→C (24·6%) and C→T (21·4%) followed by A→G (15·8%) (Fig 2). Ninety-three out of 118 (78·8%) total variations sites were located in HV-I (16024–16383) and HV-II (57–372) hyper-variable regions of D-loop. Twenty-one of 118 variation sites in mtDNA had not been previously reported when compared with available databases updated to 26·10·09.

Table I.   Baseline Patients Characteristics (n = 44).
Median age, years (range)8·5 (1–18)
Sex (male:female)3·8:1
Median haemoglobin, g/l (range)79 (28–129)
Median white blood cell count, x109/l (range)21·75 (2·4–350)
Median platelet count, x109/l (range)40·5 (11–220)
Acute myeloid leukaemia subtype (n = 44)
 M01 (2·3%)
 M17 (16·0%)
 M218 (41%)
 M31 (2·3%)
 M46 (13·6%)
 M56 (13·6%)
 M61 (2·3%)
 M74 (9·1%)
Cytogenetics (n = 30)
 Good risk2 (6·7%)
 Intermediate risk20 (66·7%)
 Poor risk8 (26·6%)
Remission rate36/44 (81·8%)
Table II.   Sites of variations in D-Loop of mtDNA (n = 35 patients).
S. No.Position of nucleotideBase variationNo. of patientsS. No.Position of nucleotideBase variationNo. of patientsS. No.Position of nucleotideBase variationNo. of patients
  1. *Novel variations.

 116051A→G14216303G→A1 83198C→T 1
 2*16054A→G14316304T→C1 84199T→C 2
 3*16081G→ins14416311T→C8 85200A→G 1
 416083C→A14516318A→T1 86204T→C 3
 516092T→C24616319G→A3 87207G→A 1
 616093T→C14716325T→C1 88*210T → ins 2
 716111C→T14816335A→G1 89211A→G 1
 816113A→T149*16348C→G1 90214A→G 2
 9*16121T→C15016344C→T1 91*216T→A 1
1016124T→C15116352/53TC→CT2 92227A→G 1
1116126T→C45216355C→T1 93225T→C 1
1216129G→A65316356T→C2 94234A→G 1
1316145T→C15416359T→C1 95246T→C 1
1416147C→T15516362T→C2 96*240A→G 2
1516150C→T15616465C→T1 97252T→C 2
16*16151C→T15716497A→G1 98257A→G 1
1716196T→G158*16498C→T1 99259A→G 1
1816207A→G15916540A→C1100263G→A 1
1916206A→C26016500T→A1101315C→T 1
2016208G→T16116506T→C1102319C→T 1
2116209C→A16216509/10TT→CC1103320C→T 1
2216214C→A16316512/13A→C1104323C→G 3
2316218A→G16416519C→T9105*362C→A 1
2416224T→C146516524A→G1106*366G→A 1
2516225T→C166*16554A→G1107373A→C 2
2616228C→T167*16560C→del1108*411C→G 3
2716230A→G26862T→C1109414T→G 1
2816234C→T169*68G→T1110461C→T 1
2916239C→T17069C→del1111*462C→G 1
3016254A→G17172T→G1112482T→C 5
3216265A→C17376C→T1114493T→C 3
3316266C→T274*78C→ ins1115494T→C 3
3416270C→T275*85/86GC→CA1116514C→G 2
3516275A→G176*92G→T1117523A→del 7
3616278C→T177114C→A1118*572C→T 1
Figure 2.

 DNA sequence chromatograms of paediatric AML patients with variations 16126T→C (A) and 16311T→C (B).

Relationship of D-loop variations with outcome

Thirty-six of 44 (81·8%) patients achieved complete remission (CR) with induction chemotherapy. Fifteen patients relapsed/died after achieving remission. All eight patients who did not achieve CR had at least one or more mtDNA variations in the 118 sites and 9/36 (25·0%) patients with CR had no change in any of the location. Median follow up was 10·9 months. The 2-year EFS (±SE) and overall survival was 30·9 ± 10·8% and 36·2 ± 14·0% respectively. There was no significant difference between the survival of patients with or without any variations (P = 0·40) (Fig 3A). Of the 118 sites of variations, 10 were found on ≥4 patients. Three variations in HV-I, namely 16126T→C (P = 0·050), 16224T→C (P < 0·010), 16311T→C (P < 0·001), were significantly associated with inferior EFS (Fig 3B–D). Further, the hazard ratio for adverse outcome was 3·1 for 16126T→C, 3·1 for 16224T→C and 4·8 for 16311T→C (Table III). The hazard ratio progressively increased, from 1·68 to 3·72, with the increasing number of D-loop variations per patient (Table III). Using multivariate Cox regression model analysis for the presence or absence of the significant D-loop variations (16126T→C, 16224T→C and 16311T→C), the predicted 2-year EFS was 48·0% for subjects without any variation, 26·0% for those who had any one variation and 8·0% for those with any two variations (Table IV). None of the patients had all three variations together.

Figure 3.

 Kaplan–Meier event-free survival curves: (A) EFS of patients with or without any variation; (B) EFS of patients with or without variation 16126TC; (C) EFS of patients with or without variation 16224TC; (D) EFS of patients with or without variation 16311TC.

Table III.   Relationship between Event-Free Survival (EFS) and baseline patients characteristics, number of mtDNA D-loop variations and individual variations present in ≥ 4 paediatric AML patients.
VariableNo. of patientsEFS (months)*P valueHazard ratio (95% confidence interval)
  1. WBC, white blood cell count.

  2. *EFS in months for 25%, 50% and 75% of events.

  3. Bold values indicate P < 0·05.

Baseline patients characteristics
 WBC < 21·75 × 109/l217·519·026·00·691·00
 WBC ≥ 21·75 × 109/l232·912·51·19 (0·57 2·72)
 Platelet count < 40·5 × 109/l227·312·019·30·501·00
 Platelet count ≥ 40·5 × 109/l227·519·026·00·75 (0·33 1·75)
  Good + intermediate2261926·00·271·00
  Poor risk82·911·619·3 1·61 (0·57 4·54)
Number of mtDNA D-loop variations
 No variation911·60·411·00
 At least 1 variation35612·5301·68 (0·49 5·78)
 <2 variations1011·60·301·0
 ≥2 variations343412·5261·91 (0·56 6·61)
 <3 variations1211·619·319·30·201·00
 ≥3 variations322·912262·03 (0·68 6·06)
 <4 variations1812·519·30·011·00
 ≥4 variations262·28·13193·72 (1·36 10·16)
Individual mtDNA variations present in ≥4 patients
 16126 T→C Absent407·319260·051·00
 16126 T→C Present41·67·510·933·05 (1·01 9·22)
 16129 G→A Absent387·3314·8326·00·801·00
 16129 G→A Present68·1312·00·80 (0·25 2·98)
 16224 T→C Absent3010·9319·3326·0<0·011·00
 16224 T→C Present142·2714·833·11 (1·28 7·55)
 16311 T→C Absent368·131926<0·0011·00
 16311 T→C Present81·07226·74·79 (1·87 12·25)
 16519 C→T Absent357·7714·8319·330·761·00
 16519 C→T Present96·08·1326·01·17 (0·43 3·22)
 146 T→ C Absent388·014·8326·00·261·00
 146 T→ C Present6661·86 (0·62 5·55)
 151 T→C Absent318·019·019·330·671·00
 151 T→C Present136·012·4726·01·20 (0·50 2·92)
 482 T→C Absent397·3314·830·701·00
 482 T→C Present57·510·93191·20 (0·37 4·36)
 489 T→C Absent327·3314·830·551·00
 489 T→C Present121·610·93191·31 (0·54 3·24)
 523 A→Del Absent377·512·470·801·00
 523 A→Del Present76·7319·026·00·80 (0·29 2·60)
Table IV.   Predicted Event-Free Survival Probability (%) on the basis of number of significant mtDNA variation (16126T→C, 16224T→C, 16311T→C).
Time (months)Number of significant mtDNA variations
0 (%)1 (%)2 (%)


The D-loop regulates the replication and transcription of mtDNA. Variations in this region have been observed in different solid malignancies and their incidence varies from 20·0 to 81·0% in different malignancies (Alonso et al, 1997; Ivanova et al, 1998; Nomoto et al, 2002; Tan et al, 2002, 2006; He et al, 2003; Suzuki et al, 2003;Grist et al, 2004; Lièvre et al, 2005;Guo & Guo, 2006; Yao et al, 2007; Wulfert et al, 2008; Yu et al, 2009) but only a single study has evaluated their significance as a prognostic marker (Linnartz et al, 2004) (Table V.) mtDNA was previously shown to be amplified 2- to 80-fold in 25 adult AML subjects as well as in 20 chronic granulocytic leukaemia patients; this amplification normalized at remission in these patients (Boultwood et al, 1996). This amplification might have been due to increased number of mitochondria or mtDNA copies per cell. As the D-loop region controls its replication, a change in this region may alter the mtDNA copies and thus this region may be a reason to study if it is altered in acute leukemias. It has also been observed that mtDNA variations in the D-loop region may be responsible for the transformation of normal or dysplastic cell to leukemic blast phenotype (Linnartz et al, 2004).

Table V.   Mitochondrial variations/mutations in different studies.
S. No.ReferenceMalignancyNo. of patientsRegion of mtDNAMethodFrequencyPrognostic significance
  1. S.No, serial number; DGGE, denaturing gradient gel electrophoresis; PCR-RFLP, polymerase chain reaction restriction fragment length polymorphism; PCR-SSCP, polymerase chain reaction single stranded conformation polymorphism; dHPLC, denaturing high-performance liquid chromatography; MDS, myelodysplastic syndrome; AML, acute myeloid leukaemia; ALL, acute lymphoblastic leukaemia; CLL, chronic lymphoblastic leukaemia; CML, chronic myeloid leukaemia.

 1Alonso et al (1997)Colorectal cancer13Whole D-loopPCR SSCP23%Not evaluated
 2Ivanova et al (1998)ALL30Whole mtDNAPCR-RFLP36·6%Not evaluated
 3Tan et al (2002)Breast cancer19Whole mtDNADirect sequencing74% (81% of total in D-loop region)Not evaluated
 4Nomoto et al (2002)Hepatocellular carcinoma19Whole D-loopDirect sequencing68%Not evaluated
 5Suzuki et al (2003)Lung cancer28 cell lines
55 patients
Whole D-loopDirect sequencing61%
Not evaluated
 6He et al (2003)AML, ALL, CLL, CML24Whole mtDNADirect sequencing40%Not evaluated
 7Linnartz et al (2004)MDS10Whole mtDNADirect sequencing50%Not evaluated
 8Grist et al (2004)ALL, AML22 AML
26 ALL
Partial D-loop (16111-190)DGGE36% AML
58% ALL
Not evaluated
 9Lièvre et al (2005)Colorectal cancer365Whole D-loopDirect sequencing38·3%Poor prognostic factor
10Guo and Guo (2006)Osteosarcoma20Whole D-loopDirect sequencing70%Not evaluated
11Tan et al (2006)Oesophageal cancer20Whole mtDNADirect sequencing55% (64% of total in D-loop region)Not evaluated
12Yao et al (2007)AML18Whole mtDNADirect sequencing60%Not evaluated
13Wulfert et al (2008)MDS104Whole mtDNAdHPLC56%Not evaluated
14Yu et al (2009)Ewing sarcoma17Whole D-loopDirect sequencing70·6%Not evaluated
15Present studyAML44 PaediatricWhole D-loopDirect sequencing79%Poor prognostic factor

Previously Ivanova et al (1998) screened for mutations in the mtDNA of 30 ALL patients using restriction fragment length polymorphism (RFLP). However, we have used a direct DNA sequencing method, which has the advantage over RFLP of detecting all possible single nucleotide mutations. Another study, using denaturing gradient gel electrophoresis for the D-loop region from 16111 to 190 bp, detected variations in 8/22 (36·0%) adult AML and 15/26 (58·0%) adult ALL patients at diagnosis (Grist et al, 2004). We, however, detected mtDNA variations in the D-loop region in 35/44 (79·5%) childhood AML subjects. The increased detection in our study may be due to the fact that we studied childhood AML as compared to adult leukemias in the previous studies. Further, we evaluated the entire D-loop region whereas Grist et al (2004) evaluated a portion of this region for variations.

The accumulation of somatic mutations is greater in mtDNA compared to nuclear DNA because nuclear DNA replicates only once at the time of cell division and undergoes proofreading by DNA polymerase. On the other hand, turnover of mtDNA is high, as degradation and replication is a continuous process in mitochondria in one cell cycle, and mtDNA polymerase γ does not have the ability to proofread (Shadel & Clayton, 1997). Defective mtDNA will result in abnormal energy yield but only high energy producing cells are likely to be selected during the pathogenic clonal expansion of neoplastic cells. These superior, energy-producing cells will accumulate more ROS as a byproduct and this may result in more defects in its surrounding mtDNA. This in turn could also damage the nuclear genome and contribute to cancer initiation and progression. mtDNA is composed of two strands, namely, heavy and light strands, based on their nucleotide density. There are two sites of origin (OH and OL) and two promoters (heavy strand promoter and light strand promoter) for the replication and transcription of heavy and light strands respectively. Light strand promoter and one origin site of the heavy strand, OH, is in close proximity to HV-I region of the D-loop (Shadel & Clayton, 1997). Thus, any change in HV-I region may have an impact in the regulation of mtDNA transcription and replication.

The present study showed, for the first time, the impact of D-loop variations on EFS of paediatric AML. Three variations (16126T→C, 16224 T→C, 16311 T→C) in HV1 region of D-loop were found to be associated with inferior EFS in paediatric AML subjects. Variations, 16224 T→C and 16311 T→C, are located at positions where the mitchondrial transcription factors, namely Tfam, bind (16220–16325 bp) and regulate the transcriptome of mitochondria (Choi et al, 2005). If there is any change in binding sites or a conformatory change in mtDNA, it may alter its binding capability of Tfam; this in turn may alter the strength of mtDNA-protein complex, which may result in aberrent expression of mitochondrial genes and development of mitochondria of cancer phenotype.

We compared the D-loop sequence of our subjects with known databases for mitochondrial polymorphisms and mutations. The available databases also include the polymorphisms and mutations observed in normal Indian individuals that overlaps with those observed in other races; however, these studies did not include any risk evaluation for disease in those individuals (Palanichamy et al, 2004; Thangaraj et al, 2008). There is no separate database for mitochondrial D-loop polymorphisms in Indian subjects. Further, we did not evaluate germline samples of our subjects and therefore, we preferred to designate the D-loop sequence alterations from the database in our subjects as variations instead of mutations or polymorphisms. Out of the 10 variations found in 4 or more patients of our cohort, 9 had been previously reported as polymorphisms, but their role in the pathogenesis of leukaemia had not been ruled out in the previous phylogenetic studies (Palanichamy et al, 2004; Thangaraj et al, 2008).

In conclusion, this is the largest study to show a high frequency of mtDNA variations in paediatric AML and their relevance as a potential prognostic marker in this malignancy. This is an indicative study for the potential inclusion of D-loop variations in large prospective clinical trials so as to evaluate their prognostic significance in AML patients.


This study was funded by Terry Fox Foundation – Cancer Research. Surender Kumar Sharawat is grateful to Indian Council of Medical Research (ICMR) New Delhi, India for a Senior Research Fellowship.