Abnormal telomere metabolism in Fanconi's anaemia correlates with genomic instability and the probability of developing severe aplastic anaemia


  • *

    Present addresses: Indiana University School of Medicine, Indianapolis, IN, USA.†CEA Saclay, France.‡IGR, Villejuif, France.

Professor E. Gluckman, Head of Bone Marrow Biology Laboratory and Bone Marrow Transplant Unit, Hospital Saint-Louis, 1 Avenue Claude Vellefaux, 75475 Paris cedex 10, France. E-mail: eliane.gluckman@sls.ap-hop-paris.fr


Summary. Fanconi's anaemia (FA) is an autosomal recessive disorder characterized by progressive bone marrow failure and a susceptibility to cancer. Haematopoietic stem cell transplantation is the only curative method for restoring normal haematopoiesis, and survival is improved if the transplant is carried out before severe complications occur. However, the evolution of FA is difficult to predict because of the absence of known prognostic factors and the unknown function of the genes involved. In studying 71 FA patients, a correlation was found between severe aplastic anaemia (SAA) and the individual annual telomere-shortening rate (IATSR) in peripheral blood mononuclear cells (P < 10−3). Spontaneous apoptosis was highest in SAA patients or patients with high IATSR (> 200 bp/year) (P < 0·01, n = 18). Univariate and multivariate analyses showed that significant relative risks for evolution towards SAA were high IATSR (P < 10−4), and that a high number of chromosome breakages occurred in the presence of nitrogen mustard (P < 0·001). A high IATSR was also associated with an increased frequency of malignancy (P < 0·01). Thus, these biological parameters were related to the spontaneous evolution of FA and could be used as prognostic factors. These data indicated that telomeres might play a role in the evolution of bone marrow failure and malignant transformation in FA.

Fanconi's anaemia (FA) is an autosomal recessive disease characterized by multiple congenital abnormalities, bone marrow (BM) failure and a susceptibility to cancer. The most common evolution is towards severe aplastic anaemia (SAA), which occurs at ≈ 8 years of age. Death usually results from complications of BM failure or from leukaemia during the first two decades of life (Alter, 1996). However, some patients are diagnosed with FA only upon discovery of secondary malignancies. To date, haematopoietic stem cell transplantation (SCT) is the only known therapeutic approach to restore normal haematopoiesis (Gluckman et al, 1995). The success of the transplant is highly dependent on both the source of haematopoietic stem cells and the patient's clinical status before transplantation. Indeed, without transplantation, patients with SAA have a poor survival rate, suggesting that it would be better to diagnose and transplant FA patients before aplasia becomes severe (Gluckman et al, 2000). However, an evolution towards SAA varies among FA patients, and no prognostic factors have been described.

The clinical manifestations of FA are remarkable in their variety. This variability is suspected to result from stochastic events caused by genomic instability and from the existence of eight complementation groups (A to G) of FA cells (Joenje & Patel, 2001). Based on the correction of the characteristic hypersensitivity of FA cells to DNA cross-linking agents such as mitomycin C (MMC) and diepoxybutane (DEB), eight different complementation groups have been identified in somatic cell hybrids (Joenje & Patel, 2001): FA-A, FA-B, FA-C, FA-D1, FA-D2, FA-E, FA-F and FA-G. The genes for FA complementation group A (FANC A), group C (FANC C), group D2 (FANC D2), group E (FANC E), group F (FANC F) and group G (FANC G) have been identified. Molecular cloning has localized FANC A on chromosome 16q24.3, FANC C on 9q22.3, FANC F on 11p13-p15 and FANC G on 9p13. Retroviral vectors containing cDNA for FA-A, FA-C and FA-G, the most frequent complementation groups in Europe and North America, allow rapid identification of the defective gene by complementation of primary T cells (Hanenberg et al, 2002). The involvement of FA proteins in a common pathway would explain similarities in the disease regardless of the complementation group, and could account for the severe symptoms that characterize the phenotypes associated with mutations in key residues (Faivre et al, 2000).

Telomeres, which identify chromosome ends, have been implicated in the control of both genomic stability and cell proliferation capacity (Hemann et al, 2001). Indeed, chromosome extremities must be protected from nuclease degradation and misidentification by repair machinery. One function of telomeres is to maintain chromosome-end integrity by means of a specialized nucleoprotein structure that requires a minimal telomere length. During DNA replication, chromosome extremities cannot be completely duplicated, leading to telomere shortening. A second function of telomeres is therefore to monitor the number of cell divisions that drive the cells progressively towards senescence, or apoptosis (Lansdorp, 1995). This occurs when telomeres become too short to ensure chromosome stability.

Thus, telomere length has been shown to diminish progressively with cell age in CD34+ haematopoietic cells, with a concomitant reduction in their proliferative potential (Vaziri et al, 1994). Studies have shown that telomeres are significantly shortened in FA cells (Ball et al, 1998; Leteurtre et al, 1999; Adelfalk et al, 2001; Hanson et al, 2001). In addition, we have previously found a correlation between the haematological severity of FA and telomere length (Leteurtre et al, 1999). In the present study, which included 71 FA patients, correlations were found between disease evolution and (1) the calculated kinetics of telomere shortening; (2) other biological features of FA, such as the amount of chromosome breakage in the presence of alkylants; and (3) the spontaneous level of apoptosis. Risk factors for severe complications in the evolution of FA are proposed, and a relationship between abnormal telomere metabolism and disease evolution is suggested.

Materials and methods

Patient characteristics.  Seventy-one FA patients treated in our department between 1982 and 2000 were studied. There were 44 males and 27 females, with a median age of 11 years (range 2–48 years). Diagnosis was confirmed in all cases by cytogenetic analysis. Complementation group data were obtained in 17 patients, which were 16 FANC A and one FANC C, whereas in eight cases, it was not determined because of the presence of somatic mosaicisms (Faivre et al, 2000). Blood samples were obtained during routine clinical visits after informed written consent by the patients or their guardians was obtained. Some patients were analysed retrospectively from previously frozen blood samples.

1. Patients were classified into severe aplastic anaemia (SAA) and non-severe aplastic anaemia (NSAA) groups. SAA patients satisfied at least one of the following criteria: a haemoglobin level < 8 g/dl, a platelet count < 20 × 109/l, an absolute neutrophil count < 0·5 × 109/l and a transfusion dependency for the last 3 months (Gluckman, 1996). NSAA patients fulfilled none of the above criteria.

2. Patients with > 5% blast cells on bone marrow aspirate analysis and myelodysplasia in more than one haematopoietic cell lineage were considered to have myelodysplastic syndrome (MDS) (Butturini et al, 1994). Specific dysgranulopoietic features were often observed in FA patients, which were an increase in myeloid cell size (especially of myelocytes and metamyelocytes) and abnormal inverted chromatins (apoptosis) in the granulocytes. Patients were diagnosed with acute leukaemia if there were > 30% blasts in their BM. Clonal abnormalities were defined by the presence of additional chromosome abnormalities in the BM cytogenetic analysis, as presented in Table I. All BM smears and cytogenetics results were reviewed blind by three investigators.

Table I.  Cytogenetics bone marrow abnormalities and/or malignant diseases for FA patients.
No.Age (years)SexIATSR (bp/year)Clonal abnormalitiesBM diagnosisSAA
  • No., patient number; AA, aplastic anaemia; RAEB, refractory anaemia with excessive blasts; AML, acute myeloid leukaemia (M4, subgroup M4); SAA, severe aplastic anaemia; MDS, myelodysplastic syndrome.

  • *

    Died of tongue cancer 6 years after BM transplant.

  • Had skin and head and neck cancers at age 32 years.

  • Had skin and uterine cancers at age 45 years.

  • §

    Died of head and neck cancer 15 years after BM transplant.

115M19045,XY−7, +11MDSYes
2 7M38647,XY, +14/45, XY,10, +14, −19AANo
417M13146,XY, +11(q23),−16,+marsMDSNo
513F19247,XY, +17AAYes
6 4M45047,XY, +8MDSYes*
718F 8947,XY, +21AA & dysgranulopoiesisYes
837F 7846,XX,i(7)(q10)/46,XXXAA & dyserythropoiesisNo
10 8F43846,XX,del13(q13q21)MDSYes
1113M16947,XX, +4(q35), +mars/47,XX,der (10)(p15), +marsMDSYes
1311M22747,XY, +1(1q),add(13p)RAEBNo
1424F12146,XX,−5,der(5)/46,XX,−22,der(22)/47,XX, +19MDSNo
1528M21147,XY, +21/47,XY, +7/46,XY,dup(1)AA & dyserythropoiesisYes
1615M19346,XY, +11(q23)AANo
17 6F33345,X/48,XXXXMDSNo
1910M19047,XY, +22AANo
2047F14747,XX, +8MDSNo
2111F42746,XX,der(1)dup(1)(p21p33)/46,XX,del(5), +21(qtr)MDS/AMLYes
22 9M522noAML-M4Yes
24 6F250noMDSNo

3. Malformations were scored according to the number of anatomical sites involved, as described previously (Guardiola et al, 2000). Only malformations of head, limbs and organs were scored. The malformation syndrome was considered to be extensive if at least three sites were involved, including at least one organ such as the kidney, gastrointestinal tract, urogenital tract or cardiovascular system.

Cytogenetics and chromosome breakage test.  Cytogenetic analyses of BM cells were carried out for every patient after in vitro cell culture. Peripheral whole blood cells were cultured for 72 h in Roswell Park Memorial Institute (RPMI)-1640 medium with 17% fetal bovine serum (FBS) and phytohaemagglutinin (PHA), both with and without nitrogen mustard (NM, 0·05 µg/ml added to the cells after 24 h of culture). Chromosome breaks were counted in 50 metaphases from each culture. FA diagnosis implied that more than five breaks were observed in the presence of NM, and less than five breaks in its absence (Berger et al, 1993). In this study, the presence of more than 150 breaks in 50 metaphases in the presence of NM was considered to represent a high number of chromosome breakages, which meant a > 30-fold increase in chromosomal breaks. These results were obtained only for the 56 patients for whom chromosome breakage studies were carried out in our hospital.

Mean terminal restriction fragment length (TRF) measurement.  Peripheral blood mononuclear cells were prepared by Ficoll/Isopaque density gradient centrifugation. Eight of 71 patients were tested two or three times over a period of 2–9 years. Telomere length was also measured in 42 healthy control individuals aged 4–55 years, and in nine control cord blood samples collected at birth. Telomere length in the mononuclear cells was determined by Southern blot analysis as described elsewhere (Vaziri et al, 1994; Leteurtre et al, 1999). The validity of TRF determinations was verified in all experiments by the incorporation of the same two quality controls: genomic DNA prepared from one cord blood and one middle-aged healthy individual.

Apoptosis.  Eighteen patients (nine SAA and nine NSAA) and 15 healthy control individuals were analysed for the spontaneous level of apoptosis in fresh peripheral blood nucleated cells. Analyses were carried out using the in situ terminal deoxynucleotidyl transferase-mediated dUTP nick end-labelling (TUNEL) assay, as specified by the manufacturer (ApopTag direct kit; Oncor, Gaithersburg, MD, USA). Cytospins from each sample were examined by fluorescence microscopy, and 200 cells per sample were scored for apoptosis (McGahon et al, 1994; Haneline et al, 1998).

Statistical analysis.  Computations were carried out using the Graph Pad prism software package (San Diego, CA, USA). Cumulative SAA probability and cumulative tumour or MDS probability were analysed using the Kaplan–Meier method. The date of blood collection for telomere analysis was chosen as the first clinical data entry and the closing date was April 2000. Fisher's exact or Mann–Whitney analyses were used to compare the various groups. Multivariate analysis was conducted using Cox analysis (SAS software, SAS Institute, Cary, NC, USA).


Among our 71 FA patients, 37 had SAA and 34 had NSAA. The mean age of the SAA patients was 10·6 years, compared with 14·6 years for the NSAA patients, suggesting that SAA evolved faster in a subpopulation of our patients. The total number of MDS plus acute myeloid leukaemia (AML) patients was 15, and the total number of patients with secondary solid tumours was four. Among these 18 patients with malignancies (one patient had both MDS and cancer), 15 had BM clonal abnormalities. BM clonal abnormalities were also observed in seven patients without malignant transformation (Table I). These results were in agreement with our previous findings, and highlight the heterogeneity of the haematological syndrome in FA (Gluckman et al, 1995, 2000). Thus, identification of the patients with fast evolution towards SAA would be of use in optimizing therapeutic options.

The individual annual telomere-shortening rate (IATSR) as a classification tool

In a previous study of 45 FA patients, we found a correlation between telomere length and BM failure. The SAA patients had shorter telomeres than the NSAA patients, who in turn had telomeres that were shorter than those of the controls (Leteurtre et al, 1999). Similar results were obtained in the present study, which included 26 additional FA patients. SAA patients, despite their lower age, had significantly shorter telomeres than NSAA patients (8·0 kb versus 8·5 kb, P = 0·01). The plot of TRF as a function of age revealed a reduction in telomere length with age. The telomere-shortening rates, obtained through linear regressions for the groups defined by their haematological status, were significantly higher among SAA patients (195 bp/year) than among NSAA patients (115 bp/year) and the control group (60 bp/year) (P < 10−3) (see the population annual telomere shortening rate in Table II) [SAA: TRF = (10·18 ± 0·25)–(0·195 ± 0·025) age, n = 42; NSAA: TRF = (10·25 ± 0·19)–(0·115 ± 0·013) age, n = 39; controls: TRF = (10·86 ± 0·19)–(0·060 ± 0·009) age, n = 42.]

Table II.  Characteristics of FA patients according to haematological and telomere-shortening classifications.
  Haematological classificationTelomere-shortening classification
SAANSAA P-value*> 200 bp/ year< 200 bp/ year P-value*
  • SD, standard deviation.

  • *

    Fisher exact or Mann–Whitney test for statistical analysis, bolding represents a statistically significant difference.

  • Except chromosome breakage: 56.

  • Number of patients in boldface, % within parentheses.

  • §

    Population annual telomere-shortening rate (PATSR).

  • Individual annual telomere-shortening rate (IATSR).

No. of patients713734 4724 
SexMale25 (68)16 (47) 27 (57)14 (58) 
Female12 (32)18 (53)0·0920 (43)10 (42)0·4
AgeMean ± SD (years)10·6 ± 4·414·6 ± 10·00·059·0 ± 4·319·3 ± 8·710−4
MalformationWith organ17 (46)11 (32) 22 (47)6 (25) 
Without organ20 (54)23 (68)0·3325 (53)18 (75)0·09
Telomere lengthMean ± SD (kb)8·0 ± 1·08·5 ± 1·20·01 
PATSR§bp/year195 ± 25115 ± 130·01 
IATSRbp/year312 ± 133212 ± 11210−3 
Chromosome breakageHigh24 (80)9 (35) 26 (70)7 (37) 
Moderate6 (20)17 (65)10−311 (30)12 (63)0·02
BM cell contentHypocellu- larity6 (43)6 (18)0·0215 (32)7 (29)1·0
Clonal anomaliesPresence of11 (30)11 (32)0·4011 (23)11 (46)0·06
Age (years)12·7 ± 5·818·8 ± 12·80·1611·2 ± 6·320·4 ± 11·50·03
MDS/tumoursPresence of12(32)6 (18)0·1812 (26)6 (25)0·45
Age (years)10·5 ± 3·120·5 ± 15·70·0110·5 ± 4·125·5 ± 13·70·01
Haematological statusSAA 31 (66)6 (25) 
NSAA 16 (34)18 (75)0·01

The haematological status does not predict progressive evolution towards pancytopenia or the severity of underlying disease. A group of NSAA patients might include patients with severe disease before the onset of aplasia, especially among the younger patients. Sequential measurements of telomere length in a small sample of patients had previously suggested that patients evolving towards SAA might have a higher telomere-shortening rate (Leteurtre et al, 1999). Therefore, we analysed the IATSR in every patient, hypothesizing that IATSR could reflect the severity of disease.

As sequential sampling of every FA patient is difficult to obtain, the IATSR was calculated. In a previous study, we found that heterozygous FA parents have a normal TRF, which suggests that FA homozygotes could have inherited a normal telomere length before the first mitosis (Leteurtre et al, 1999). Telomere length in FA patients at birth was thus considered to be approximately similar to the length found in control cord bloods, which was 10·9 ± 0·6 kb (n = 9). A theoretical IATSR was then calculated as follows: IATSR = (10·9–TRF length)/age. The validity of the IATSR concept implies that the loss of telomere length is constant with time. In agreement with previous reports (Frenck et al, 1998; Rufer et al, 1999), the calculated IATSR was constant in normal control subjects between 5 and 55 years old, as shown in Fig 1A. For the FA patients, IATSR analysis of the population would be misleading as the follow-up between diagnosis and bone marrow transplantation or death is around 10 years, and as patients with the most aggressive forms of the disease are diagnosed earlier in life (Fig 1A). Instead, the IATSR for the eight FA patients for whom sequential blood samples could be obtained were compared (Fig 1B). For the two FA patients with three IATSR determinations, telomere shortening was both constant and linear with time (Fig 1B, dotted lines).A comparison of the IATSR in the patients for whom we had at least two sequential measures of telomere length showed that IATSR values were constant over time, as shown graphically (Fig 1B) or through statistical analysis (Wilcoxon matched pairs test, P ≥ 0·35). Finally, the identical telomere erosion found in two homozygous twins suggests that the underlying disease (i.e. gene mutation) would be responsible for this accelerated erosion. Thus, the kinetics of telomere erosion in individual patients, as determined by IATSR, seems to be both linear and reproducible, and could be used to compare FA disease evolution in individual patients.

Figure 1.

Individual annual telomere-shortening rate (IATSR) as a function of age. (A) IATSR was constant in controls. IATSR was higher in FA patients than in controls. (B) The IATSR in FA patients was constant for those in whom sequential blood sampling could be obtained at least 1 year apart. The first, second and, when available, third IATSR determinations for the given patients are connected by a line (dotted line when three values of IATSR available). y, years.

Because the number of patients was too small to analyse the IATSR as a continuous variable, patients were divided into two groups according to their IATSR values. SAA individuals had higher IATSRs (median 267 bp/year) than NSAA patients (median 179 bp/year, P < 10−3) (Fig 1A). Thus, the threshold was set at an arbitrary value of 200 bp/year, which was close to the mean of the median IATSR in the SAA and NSAA groups, especially if the younger patients (aged under 6 years) were excluded to reduce the influence of the higher telomere-shortening rate described during very early childhood (Frenck et al, 1998; Rufer et al, 1999).

Patient characteristics relevant to their haematological and telomere-shortening rate status

To determine whether the IATSR could be a useful prognostic indicator of FA evolution, we examined FA patient characteristics according to their haematological status and their IATSR status. Patients with a high telomere-shortening rate (IATSR > 200 bp/year) mainly had SAA (66%), whereas those with a low telomere-shortening rate (IATSR < 200 bp/year) were mainly NSAA patients (75%). As a group, the 10 patients with an IATSR > 200 bp/year and NSAA were younger than the mean age of the NSAA patient group [8·0 ± 4·0 years (n = 10) versus 14·6 ± 10·0 years (n = 34), P < 0·01], suggesting that these young NSAA patients could be those rapidly evolving towards severe aplasia. In contrast, SAA patients with low telomere-shortening rates (IATSR < 200 bp/year) were older than their group mean age [14·1 ± 2·1 years (n = 6) versus 10·6 ± 4·4 years (n = 37), P < 0·01]. More interestingly, seven out of the eight gene mosaicisms were found in NSAA patients with a low telomere-shortening rate.

General characteristics

SAA patients were significantly younger than the NSAA patients (mean age 10·6 years versus 14·6 years, P = 0·05), and the difference between the IATSR groups (9 years versus 19·3 years, P < 10−4) was even more notable. In a comparison of the sex ratio between the two groups, SAA patients were more often male than female (25 versus 12, P = 0·09), whereas no sex ratio difference was found between the IATSR groups. Moreover, the presence of organ malformations was frequently found in patients with an IATSR > 200 (P = 0·09, Table II).

Chromosome breakage

In the SAA group, 80% of the patients (24 out of 30) had a high number of chromosome breakages. In the NSAA group, 65% of the patients (17 out of 26) did not show a high number of chromosome breakages, while most of those with a high number of breakages (seven out of nine) were below 8 years of age, suggesting that they were tested before evolving towards SAA. Thus, a significant correlation between the haematological status of FA patients and the magnitude of chromosome breakage was found (P = 10−3). Similar results were obtained when comparing the IATSR classifications (P = 0·02). Therefore, the number of chromosome breaks induced by NM correlated significantly with the severity of FA, as evaluated by both the haematological status and the individual telomere-shortening rate.

Spontaneous apoptosis in the peripheral blood cells of FA patients

FA cells have been shown to exhibit an increased spontaneous apoptosis rate (Ridet et al, 1997). The fraction of apoptotic cells was 16·0 ± 4·8% in the SAA patients (n = 9), 9·4 ± 4·9% in the NSAA group (n = 9) and 4·5 ± 1·7% in the healthy controls (n = 15; Fig 2A). These differences between groups were significant (P ≤ 0·01). There was also a significantly higher rate of apoptosis in the high versus low IATSR group (16·2 ± 4·4% versus 7·2 ± 2·7%, P < 10−3; Fig 2B). Interestingly, there was a linear correlation between the rate of telomere shortening and spontaneous apoptosis in the FA patients (Fig 2C). In addition, all SAA patients, all high IATSR patients and all patients with a high number of chromosome breakages had more than 10% apoptotic peripheral blood nucleated cells (Table II). The spontaneous apoptosis fraction was therefore strongly correlated with genomic instability, as quantified by chromosome breakage in the presence of NM, and with the severity of the disease, as evaluated by either haematological status or the individual telomere-shortening rate.

Figure 2.

Spontaneous apoptosis in white blood cells was correlated with the severity of FA and the telomere-shortening rate (IATSR). (A) Significant differences were found between FA patients and controls (P < 0·01) and between SAA and NSAA FA patients (P = 0·01). (B) A significant difference was found in the apoptotic fraction of FA patients' cells, according to the IATSR (P < 10−3). (C) The linear relationship between IATSR and apoptosis in FA patients: apoptotic fraction = (0·0385 ± 0·0066) IATSR + (3·34 ± 1·67), r2 = 0·67. The arrow indicates a 2-year-old patient for whom the very young age would increase the IATSR. y, years; SD, standard deviation; n, number of patients.

BM cellular composition and clonal abnormalities

The BM cellular composition was evaluated by BM aspiration, and 22 patients were found to display hypocellularity. Forty-three per cent (16 out of 37) of patients in the SAA group had hypocellular BM, whereas only 18% (six out of 34) of the NSAA patients exhibited this characteristic (Table II). The difference between the disease groups was significant (P = 0·02). This was as expected, as BM cellular composition is indicative of the severity of the haematopoietic status, although it is not necessarily very sensitive when compared with BM biopsies. In contrast, no correlation could be found between the BM cellular composition and the IATSR. Thus, the BM cellular composition describes the current haematopoietic status, whereas IATSR reflects disease evolution.

BM clonal abnormalities were present in 31% of the FA patients (22 out of 71; Table II), but no association was found with the classification of FA patients. However, when comparing patients by telomere-shortening classification, those with BM clonal abnormalities tended to be older (P = 0·03). When assessing premalignant and malignant transformation, 13 patients were found to have MDS, four had solid tumours, and two had AML (one patient had both MDS and a solid tumour). When classified by IATSR, 12 of these patients were in the high IATSR group, and only six were in the low IATSR group, with the latter group being older (P < 0·01, Table II).

Prognostic factors for disease evolution towards SAA and malignancy

Prognostic factors for the evolution towards SAA were first evaluated using univariate log-rank analysis (Fig 3). Although SAA tended to be more frequent among males, sex was not a significant risk factor. However, spontaneous apoptosis was a significant prognostic factor (P = 0·03), although the small number of patients analysed (n = 18) did not enable calculation of the relative risk. The presence of severe malformations (involving at least one organ) slightly increased the risk of SAA development [relative risk 1·8, 95% confidence interval (CI) 1·0–4·1, P = 0·06]. Moreover, a high number of chromosome breaks in the presence of NM was associated with a relative risk of 3·6 (CI 1·6–7·4; P = 10−3). Finally, a high IATSR was associated with a 7·2-fold increase in risk (CI 4·1–18; P < 10−4). Cox multivariate analyses for an evolution towards SAA were carried out on a population of 56 FA patients for whom both chromosome breakage quantification and the IATSR were available. The calculated unadjusted relative risks were 19·2 (CI 2·4–151·4, P = 10−4) in the presence of high IATSR and 15·7 (CI 1·9–128·3, P < 10−3) in the presence of a high number of chromosome breaks.

Figure 3.

The probability of SAA varied with different factors: presence or absence (w/o) of organ malformations (A), number of chromosome breaks in the presence of alkylating agent (B) and telomere-shortening rates (C). Kaplan–Meier cumulative probabilities and relative risk (RR) factors are indicated. A high number of chromosome breaks corresponded to at least > 30-fold increased chromosome breaks in the presence of NM. n, number of patients; y, years.

Prognostic factors for an evolution towards malignancy or premalignancy (MDS) were evaluated. Owing to the small number of events (n = 18), only univariate log-rank analysis was performed. Patients with a high IATSR had both higher frequency and earlier onset of malignant disease with a relative risk of 3·5 (CI 1·8–14·1, P < 0·01, Fig 4).

Figure 4.

The probability of malignancy varied as a function of age and IATSR. Kaplan–Meier cumulative probability of premalignancies and malignancies is shown as a function of IATSR groups. Patients with a high IATSR evolved more readily and rapidly towards malignancies than those with a low IATSR (P < 0·01). RR, relative risk; y, years; n, number of patients.


This study demonstrated a relationship between the telomere-shortening rate and an evolution of FA towards SAA and malignancies. Apoptosis and genomic instability, measured by the number of chromosome breaks after incubation with NM, were found to correlate with the severity of BM failure. We also investigated a possible role for telomere metabolism in FA pathophysiology.

Implications for the prognosis of the disease

Ideally, prognostic factors should be associated with a high relative risk and be based on parameters that remain stable throughout the course of the disease. The presence of organ malformations had a modest predictive value for the evolution of SAA from FA, with a relative risk of 1·8 (P = 0·06). On the other hand, a high individual telomere-shortening rate was associated with the highest relative risk, which was increased 7·2-fold (P < 10−4). The IATSR was the most sensitive parameter in multivariate analyses, but the absence of IATSR variation during the course of the disease could be more difficult to assess. Indeed, the rate of telomere shortening relies upon the balance between cell proliferation and telomerase activity, which is known to vary during the first 2 years of life (Frenck et al, 1998; Rufer et al, 1999; Robertson et al, 2000). However, among the eight patients for whom we obtained sequential TRF values, the calculated IATSR was statistically constant. The telomere-shortening rate might therefore constitute a valid prognostic factor for the evolution of FA towards both SAA and malignancies. The main influence of the IATSR in the absence of early disease onset was telomere length, which has been proposed as a prognostic factor for pathologies such as MDS (Ohyashiki et al, 1999), aplastic anaemia (Ball et al, 1998; Brummendorf et al, 2001) and chronic myeloid leukaemia (Brummendorf et al, 2000).

Chromosome breakage studies are commonly carried out during FA diagnosis. This parameter reflects the consequences of FANC gene mutations. The number of chromosome breaks was associated with the second highest relative risk of evolution towards SAA (3·6, P = 10−3). However, in a previous study, Gillio et al (1997) found a negative correlation between the number of chromosome breaks and the severity of disease in a subpopulation of FANC C patients. There are two points that might explain the differences between this study and our own. First of all, the majority of our French patients belong to the FANC A complementation group, according to our complementation data and a previous report (Faivre et al, 2000). Secondly, FANC C was shown to be involved in additional pathways that control apoptosis and cell division cycle 2 (cdc2) kinase, which is the kinase that allows progression to mitosis when chromosome breakage studies are performed (Kruyt et al, 1997; Kupfer et al, 1997; Cumming et al, 2001), leading to a possible selection bias because of the use of the different alkylating agents, diepoxybutane (DEB) and NM (the latter of which was used in our study).

The various risk factors examined in the present study were mainly based on parameters with threshold cut-off values. Telomere-shortening rate was found to be the most sensitive risk factor. Consequently, prognostic factors should result from the combination of several of these risk factors, especially when one parameter stands in close proximity to its cut-off threshold. In this retrospective study, combining the different risk factors provided a good description of the FA evolution of individual patients. The study also indicated the value of measuring the level of spontaneous apoptosis in peripheral mononuclear cells, which is a parameter that is easier to measure and normalize between different laboratories than telomere analysis. Additional studies are required in order to validate these prognostic factors in FA patients with known mutations.

Possible role of abnormal telomere metabolism in the disease evolution of FA

Telomeres play an essential role in normal cell growth, survival and chromosome stability, and a single short telomere may be sufficient to alter these parameters (Hemann et al, 2001). The direct physiological consequences of telomere shortening have only been analysed in knock-out mice for the telomerase RNA component (mTR–/–) (Lee et al, 1998; Rudolph et al, 1999). In that in vivo model, telomeres were shown to shorten at a rate of 4·8 ± 2·4 kb per generation. As the generation increased, an increasing number of cells displayed chromosome ends without detectable telomeres. Late-generation animals exhibited defects in highly proliferative tissues, including a compromised proliferative capacity of haematopoietic cells (Lee et al, 1998; Rudolph et al, 1999). A causal effect of telomere length on BM failure has been suggested after the comparison of two strains of knock-out telomerase mice, as aplasia appeared at an earlier time in the strain with the initial shortest telomeres (Herrera et al, 1999). More recently, patients with dyskeratosis congenita have been shown to display reduced telomerase activity, either via a mutation of the dyskerin protein bound to the RNA component of telomerase or through alteration of the RNA matrix of telomerase (Mitchell et al, 1999; Vulliamy et al, 2001a). Patients with dyskeratosis congenita have extremely short telomeres (our unpublished observations; Vulliamy et al, 2001b). In agreement with a role for telomere length in the appearance of BM failure, patients with dyskeratosis congenita have also been shown to evolve towards SAA (Marciniak & Guarente, 2001).

In FA patients, we previously found a significant correlation between short telomere length and haematopoietic status (Leteurtre et al, 1999). However, in a proportion of young patients, SAA occurred in the presence of moderate overall telomere shortening, at a telomere length at which normal elderly persons have normal haematopoiesis. In this study, we analysed the relationship between FA disease evolution and individual telomere-shortening rate. A high shortening rate correlated significantly with the evolution towards SAA. A correlation between IATSR and apoptosis was also found, suggesting a link between telomere-shortening rate and genomic instability. Therefore, high genomic instability, as revealed by the high telomere-shortening rate, might lead to the appearance of haematopoietic stem cell clones with at least one critical short telomere, leading to BM failure.

Another characteristic of FA patients is the high incidence of malignancies. Short telomeres increase genomic instability, and late-generation mTR–/– mice exhibit an increased incidence of spontaneous malignancies (Lee et al, 1998; Rudolph et al, 1999). Dyskeratosis congenita patients also have an increased incidence of cancer (Marciniak & Guarente, 2001). In FA patients, up to 32% of patients develop MDS, and at least 20% develop cancers (Butturini et al, 1994; Alter, 1996; Alter et al, 2000). Because many FA patients die of BM failure or other complications before the development of cancer, the actuarial cancer risk is probably even higher. Interestingly, in addition to a short telomere length, a high rate of telomere shortening was a significant risk factor for developing malignant disease during childhood. Thus, the telomere-shortening rate, probably as a marker of genomic instability, may also be one of the parameters driving FA patients towards BM failure and malignancies.

In conclusion, this study highlights the complex relationship between telomere metabolism, genomic instability, apoptosis and certain aspects of FA pathophysiology. Additional studies are required to examine the precise roles of FA proteins in genomic instability, including that of telomeres.


Research grant support was provided by Electricité de France, the Association pour la Recherche sur le Cancer, the Ligue contre le Cancer, the Etablissement Français des Greffes, Association de recherche sur la transplantation médullaire (ARTM) and the Association Franco-Chinoise pour la Recherche Scientifique et Technique.