Main

Ataxia-telangiectasia (A-T) is a rare autosomal recessive neurological disorder, typically diagnosed in early childhood and characterised by progressive cerebellar degeneration and oculocutaneous telangiectasia. Almost all cases of A-T are caused by homozygous or (more commonly) compound heterozygous mutations in the ATM gene (MIM 607585) (Savitsky et al, 1995; Taylor and Byrd, 2005). Ataxia-telangiectasia patients show a predisposition to the development of different types of lymphoid tumours as children or as young adults and, less frequently, they may also develop brain and other tumours (Spector et al, 1982; Morrell et al, 1986; Taylor et al, 1996). The spectrum of leukaemias in children with A-T is different to that in the general population in which infant and childhood leukaemias are mostly pre-B and B-precursor ALL, involving a stage before V(D)J recombination. About half of them involve Bcr-Abl, Tel-AML, MLL and other rearrangements (Greaves and Wiemels, 2003). There appears to be no role of ATM in protecting against B-precursor ALL, as there is no observed increase in these tumours in A-T patients. The proposed role of ATM in V(D)J recombination (Bredemeyer et al, 2006, 2008; Huang et al, 2007; Vacchio et al, 2007; Hewitt et al, 2009) is consistent with the predisposition to lymphoid tumour development that involves chromosome translocations, derived through errors in immune system gene rearrangement, occurring in A-T patients often at an earlier age than the same tumour in the normal population (Taylor et al, 1996).

In contrast, heterozygous female ATM mutation carriers in A-T families have an increased risk of developing breast cancer, without any evidence for an increased risk of lymphoid tumours (Swift et al, 1991; Geoffroy-Perez et al, 2001; Olsen et al, 2001; Thompson et al, 2005). Ataxia-telangiectasia shows substantial phenotypic heterogeneity, both neurological (Hiel et al, 2006; Verhagen et al, 2009) and immunological (Staples et al, 2008). There is also substantial allelic heterogeneity, and a consideration of possible mechanisms for tumour development in A-T patients must take into account the expression of either different mutant ATM proteins, or combinations of two ATM proteins with or without kinase activity. We sought to establish the relationship between absence or presence of ATM kinase activity and cancer risk, based on a large cohort of A-T patients from the UK and the Netherlands.

Materials And Methods

Patient cohort

Ethical approval was given by the National Research Ethics Committee (REC reference 07/H1210/155). Ataxia-telangiectasia patients from the UK (n=244) and the Netherlands (n=52) were included in the analysis (Table 1). Participants were recruited on the basis of a clinical diagnosis of A-T and without reference to their cancer status. The follow-up period was defined as beginning at birth and continuing until the earliest of death, cancer diagnosis, loss of contact or database completion. Time period of follow-up ranged from 10 months to 66 years (median=18 years, IQR=11–28 years), with a total of 6239 person-years of follow-up.

Table 1 Details of the study cohorts
Full size table

Identification of A-T patients whose cells expressed ATM kinase activity

As many as possible ATM mutations were identified mainly by genomic sequencing (primer data available on request), in all UK and Dutch A-T patients, both with tumours (ATM mutations categorised in Supplementary Tables S1–S5) and without tumours (mutations not shown). Cell lines were available from all UK A-T patients and most Dutch A-T patients. In conjunction with sequencing, we also used western blotting on all cell lines to determine whether or not there was expression of ATM protein. As our hypothesis was that expression of residual ATM kinase protected against tumour development in A-T patients, we also determined whether cells showed ATM kinase activity as assessed by detection of phosphorylated substrates of ATM, using phospho-specific antibodies (see Materials And Methods below). Using a combination of ATM mutation identification and western blotting for ATM protein detection, the following were used as criteria by which cells were deemed not to have any ATM kinase activity.

Criteria by which cells were deemed not to have any ATM kinase activity

  1. 1)

    The presence of two truncating ATM mutations. Truncating ATM mutations result in instability and loss of the ATM protein from the cell. Therefore, those A-T patients with two truncating ATM mutations were assumed not to express ATM protein (and therefore no ATM kinase activity). Nevertheless, western blotting was used to confirm, in most cases, whether or not cell lines expressed any ATM protein. We could show that where there were two truncating ATM mutations there was always absence of ATM protein (Supplementary Table S1).

  2. 2)

    Where ATM was found to be expressed, the mutational cause was always investigated. Where ATM protein was expressed, ATM kinase activity was measured. Absence of ATM kinase activity was associated with following types of mutation producing ATM protein (Supplementary Table S2, Supplementary Figure S1).

  1. i)

    Some founder mutations (e.g., c.7638_7646del9; p.Arg2547_Ser2549del) resulted in loss of three in-frame amino acids (Stankovic et al, 1998); previous work showed expression of a low level of mutant protein and absence of ATM kinase activity (Stewart et al, 2001).

  2. ii)

    Splice site mutations, resulting in exon skipping, that allowed expression of a very low level of ATM protein, were tested for ATM kinase activity. In all the cases, measurable kinase activity was absent. These occurred, for example, in 25 out of 66 A-T patients with tumours (Supplementary Table S2).

  3. iii)

    A small group of A-T patients carried missense mutations that did not result in expression of any ATM protein (for A-T patients with tumours, see Supplementary Table S1).

  4. iv)

    A small group of A-T patients carried missense mutations that resulted in expression of low-level mutant protein but without any activity (for A-T patients with tumours, see Supplementary Table S2).

  5. v)

    Mutation in the initiating methionine of the coding sequence leading to likely use of the next downstream methionine, as judged by the slightly truncated mutant protein (for A-T patients with tumours, see Supplementary Table S2).

  6. vi)

    A small group of A-T patients carried mutations that resulted in expression of an almost normal level of mutant ATM protein but without any kinase activity. In the non-tumour patients, these included c.9022C>T (p.Arg3008Cys) (Angele et al, 2003) and in 3 out of 66 tumour patients, two with c.8293G>A (p.Gly2765Ser) (Supplementary Table S3, Supplementary Figure S2).

In total, 135 out of 185 A-T patients without tumours showed absence of ATM kinase activity and out of 66 patients with tumours 51 did not show kinase activity.

Criteria by which cells were deemed to have ATM kinase activity

With respect to the presence of ATM kinase activity, this was always observed in cells from A-T patients with or without tumours who carried:

  1. i)

    The c.5763-1050A>G (p.Pro1922fs) leaky splice site mutation (all from the UK and Ireland). Where not tested, cells with this mutation were assumed to be expressing a low level of normal ATM with kinase activity as previously shown (McConville et al, 1996; Stewart et al, 2001, Supplementary Table S4, Supplementary Figure S4).

  2. ii)

    The c.1066-6T>G p.Val356fs mutation. This is assumed to be a leaky splice site mutation (Austen et al, 2008) (Supplementary Table S4, Supplementary Figure S4).

  3. iii)

    Some, but not all, missense mutations. These could express either a relatively high level of mutant ATM protein (e.g., c.7271T>G; p.Val2424Gly) (Stankovic et al, 1998; Stewart et al, 2001) or a low level (c.8494C>T) mutant ATM protein (Barone et al, 2009), but with clear ATM kinase activity (Supplementary Table S4, Supplementary Figure S4); therefore, the mutant ATM could have either low (e.g., p.Val2424Gly) or high (e.g., p.Arg2832Cys) specific activity.

In total, 50 out of 185 A-T patients without tumours showed the presence of some ATM kinase activity and 14 out of 66 A-T patients with tumours (in one tumour patient the ATM kinase activity could not be determined – Supplementary Table S5). The ATM kinase activity in these individuals, therefore, came either from normal ATM as a result of a leaky splice site mutation or from mutant ATM expressed by an allele with a missense mutation.

The smaller proportion of tumours in UK compared with Dutch A-T patients was due in part to the presence of the IVS40 splice site mutation in the British and Irish population and also the higher proportion of older Dutch patients with more years at risk.

Cell culture

We derived a lymphoblastoid cell line from each A-T patient's blood by separating the lymphocytes, infecting the peripheral blood mononuclear cells with Epstein–Barr virus by adsorption for 4 h, followed by addition of cyclosporin A (0.1 μg ml−1) to kill any remaining T cells and allowing 14 days for the transformed B lymphocytes to emerge. The lymphoblastoid cell line was subsequently grown in RPMI 1640 medium (Sigma-Aldrich, Irvine, UK) supplemented with 10% fetal calf serum (PAA Laboratories, Pasching, Austria). We derived lymphblastoid cell lines from normal controls in the same manner.

Immunoblotting for ATM expression and ATM activity assays

For ATM protein expression assays, patient-derived lymphoblastoid cells were harvested and cell pellets were resuspended in UTB buffer (9 M urea, 50 mM Tris, pH 7.5, 150 mM β-mercaptoethanol) and lysed on ice by sonication. For ATM activity assays, patient-derived lymphoblastoid cells were either mock-irradiated or exposed to 5 Gy ionising radiation and harvested after 30 min before the cell pellets were resuspended in UTB buffer and lysed on ice by sonication. Whole-cell lysate (50 μg) was separated by SDS–polyacrylamide gel electrophoresis, and the proteins transferred to nitrocellulose membrane (Pierce, Rockford, IL, USA). Nitrocellulose strips were subjected to immunoblotting and the protein bands visualised using the enhanced chemiluminescence (ECL) system (Amersham, Little Chalfont, UK) and exposing to Hyperfilm MP autoradiography film (GE Healthcare, Little Chalfont, UK). Antibodies used for immunoblotting were ATM (custom-made mouse monoclonal antibody clone 11G12 previously described), pATM Ser1981 #AF1655 (R&D Systems, Abingdon, UK), Smc1 #A300-055A, pSmc1 Ser966 #A300-050A, Nbs1 #Ab398, pNbs1 Ser343 #Ab47272 (Abcam, Cambridge, UK), p53 Ser15 #9284, pChk2 Thr68 #2661, Chk2 and p53 (gift from Dr G Stewart).

Statistical methods

The statistical analyses were based on data from 296 A-T patients from the UK (n=244) and the Netherlands (n=52), having excluded 13 A-T patients for whom date of birth was unknown. For the main analysis, the at-risk person-years were defined as beginning at birth and continuing until the earliest of death, first cancer diagnosis, loss of contact or database completion (end of 2009 for the Netherlands; March 2010 for UK). Cancer incidence rate ratios for patients with ATM kinase activity relative to a baseline of no kinase activity were estimated using Poisson regression. Survival until cancer diagnosis for patients with and without kinase activity was compared using a Kaplan–Meier plot and the log-rank test. Differences in the proportions of lymphoid and non-lymphoid tumours between age-groups and between kinase-absent and -present patients were compared using Fisher's exact test. We were not able to obtain kinase activity information from all patients, but this did not introduce bias, as all comparisons were between patients with known status.

The breast-cancer-specific analysis, differed in that follow-up, continued until the earliest of breast cancer diagnosis, death, loss of contact or database completion, and was left-truncated at 1960, as population rates are unreliable before this date.

Results

Distribution of tumours in the patient cohort

Overall, 66 first cancer diagnoses were made including several rare types of tumour; 47 tumours were lymphoid and 19 non-lymphoid; 42 were diagnosed ⩽16 years of age and 24 were diagnosed >16 years (Table 2). The proportion of tumours, which were lymphoid, was significantly greater in the younger age-group (P=0.002).

Table 2 Tumour types observed in ataxia telangiectasia patients by age and presence/absence of ATM kinase activity
Full size table

Of the 44 families with multiple cases of A-T, 7 families contained two A-T patients with a cancer diagnosis. For five of the families, the two cases were of the same tumour type (breast cancer, mixed-cellularity Hodgkin's lymphoma, Hodgkin's lymphoma, T-cell lymphoma and T-ALL, respectively). The other two families contained a T-PLL and a pancreatic cancer, and a myeloma and a myeloid leukaemia, respectively.

Relationship of the presence of ATM kinase to tumour

Our hypothesis was that the majority of tumours at ⩽16 years were related to absence of ATM kinase activity. We sought to identify the ATM mutations in as many patients as possible, and to determine whether their cells expressed any ATM protein with residual ATM kinase activity (Supplementary Tables S1–S5 and Supplementary Figures S1–S4). We were able to determine whether there was any ATM kinase activity for 251 patients; 187 showed no kinase activity and 64 did. Overall survival until first cancer diagnosis was significantly better among A-T patients with some ATM kinase activity (log-rank test P=0.0001) (Figure 1). The proportion of tumours that were lymphoid as opposed to non-lymphoid was significantly lower in those with some ATM kinase activity (P=0.018) (Table 2).

Figure 1
figure 1

Kaplan–Meier survival to cancer diagnosis by ATM kinase activity. Shaded areas are 95% CIs. The solid line is for A-T patients with no ATM kinase activity, and the dashed line is for patients with some kinase activity. The values under the graph refer to the number of patients in each group at risk at the corresponding analysis time.

Full size image

Table 3 shows the incidence rates of lymphoid and non-lymphoid tumours, according to the absence or presence of ATM kinase activity. The incidence rate for lymphoid tumours was significantly lower in those with kinase activity than in those without (incidence rate ratio (IRR)=0.22, 95% confidence interval (CI)=0.10–0.49, P<0.001), but there was no difference in the rates of non-lymphoid tumours between the two groups (P=0.91). The presence of kinase activity was also associated with a lower risk of lymphoid tumours when restricted to tumours diagnosed ⩽16 years of age (IRR=0.15, 95% CI=0.03–0.63, P=0.01 for lymphoid tumours) (Table 4). All six of the non-lymphoid tumours diagnosed at ⩽16 years, including three brain tumours, were in patients with no kinase activity, hence it was not possible to estimate a non-lymphoid IRR. In fact, only two cancers of any type were diagnosed before the age of 16 years in A-T patients with kinase activity; a T-cell lymphoma at an age of 2 years in a patient with two missense mutations and T-ALL at 9 years in a patient with both a frameshift and leaky splice site mutation (Supplementary Table S4).

Table 3 Cancer incidence rates according to presence or absence of ATM kinase activity, for lymphoid and non-lymphoid tumours
Full size table
Table 4 Cancer incidence rates according to presence or absence of ATM kinase activity, for lymphoid and non-lymphoid tumours, by age at diagnosis
Full size table

The IRRs associated with kinase activity were less extreme in the over-16 age-group, and none of the differences were significant, possibly due to the smaller numbers of tumours. Importantly, the proportion of A-T patients with kinase activity was significantly higher in the older age-group; of patients with known kinase status, 64 out of 251=25% of patients in the ⩽16 age-group had kinase activity, compared with 57 out of 138=41% of patients who survived beyond the age of 16 years (P=0.002). This likely reflects the generally milder phenotype typically associated with kinase-active mutations, as well as the lower cancer risk.

In the 24 A-T adults with a tumour, 11 were of lymphoid origin, 7 with the same spectrum as the paediatric lymphoid tumours (Table 2). The remaining five included four T-PLLs (diagnosed at ages 27, 32, 35 and 43 years) and a myeloma (48 years), all normally associated with old age in the general population. The 13 non-lymphoid adult tumours consisted of one pancreatic, two thyroid, one ectopic pituitary, a testicular seminoma, a myeloid tumour and seven breast cancers (Table 2). The breast cancers were diagnosed at ages 27, 32, 37, 43, 44, 44 and 50 years and were all diagnosed in women.

Given the existing evidence for an increased risk of breast cancer in female heterozygous ATM mutation carriers (Swift et al, 1991; Olsen et al, 2001; Thompson et al, 2005; Geoffroy-Perez et al, 2001; Renwick et al, 2006), we performed a separate analysis of breast cancer risk in A-T patients (i.e., with biallelic ATM mutations) Using breast cancer diagnosis as the exit criteria, rather than diagnosis with any cancer, the cohort contained an eighth eligible breast cancer, diagnosed at the age of 29 years following a T-ALL at the age of 15 years. Overall, 0.26 breast cancers were expected in these 141 women (2963 person years) according to population rates (UK and the Netherlands, as appropriate, by 5-year age-group and ∼5-year calendar period) (see Supplementary references for cancer in five continents). This gives a standardised mortality ratio of 31.1 (with a wide 95% CI=14.4–74.7) equivalent to a cumulative risk of breast cancer diagnosis by the age of 50 years of 45% (95% CI=24–76%), or 10-year risks of 13% (95% CI=6–28%) between ages 30–39 years and 36% (95% CI=19–66%) between ages 40–49 years.

Discussion

We show here that development of childhood tumours (lymphoid and brain) in A-T patients is associated almost exclusively with absence of ATM kinase activity. Conversely our findings suggest that expression of some residual ATM kinase activity has a strongly protective effect against tumour development in A-T in childhood. The source of the residual ATM kinase activity was a low level of either normal ATM or mutant ATM. The origin of the retained kinase activity in 25 patients (median age, 29 years) was the presence of the IVS40-1050A>G splice site mutation (McConville et al, 1996; Stewart et al, 2001; Sutton et al, 2004) expressing a low level (∼5%) of normal ATM. Despite the protection against tumour development provided by retained ATM kinase activity, two developed tumours, although significantly only in adulthood (one patient had a T-ALL at 17 years of age; the other patient had a testicular seminoma at an age of 27 years and a cerebral diffuse large B-cell lymphoma at an age of 34 years). This apparent contradiction may be reconciled if the level of ATM kinase activity in these IVS40-1050A>G-carrying patients is close to a threshold of ATM kinase activity that would normally be tumour suppressive.

Ataxia telangiectasia patients also clearly show a very substantially increased risk of breast cancer at young ages; the estimated risk by the age of 50 years is 45%, which is higher than the equivalent risks, in unselected carriers of BRCA1 or BRCA2 mutations (31% and 20%, respectively) (Antoniou et al, 2008). In contrast, the breast cancer risk, by the age of 50 years, associated with a heterozygous ATM mutation, is ∼9% (Thompson et al, 2005). Given the small number of adult female A-T patients, the 95% CI for this estimate is wide (24–76%), but even the lower bound would represent a very important level of risk. The eight cases included a pair of sisters who are homozygous for the c.7271T>G (p.Val2424Gly) mutation, and who have previously been reported as having breast cancer (Stankovic et al, 1998). This mutation has also been reported as being a breast cancer risk allele in its heterozygous form (Chenevix-Trench et al, 2002). Excluding these two sisters (the only 7271T>G carriers in this study), the estimated breast cancer risk by the age of 50 years was 39% (95% CI=18–75%), hence, the high risk among A-T patients is not restricted to this mutation. This observation has important implications for the clinical management of women with AT.

Two unrelated breast cancer A-T patients were compound heterozygous for the p.Val2716Ala protein, which retains some ATM kinase activity, although greatly reduced compared with normal (Verhagen et al, 2009). Interestingly, development of breast cancer, in A-T patients, may occur either in the total absence of ATM kinase activity (Supplementary Tables S1–S3) or in the presence of a low level of retained ATM activity (Supplementary Table S4), and it is the longer survival of A-T patients with some residual ATM kinase that allows this predisposition for breast cancer to manifest.

Overall, the protective effect of residual ATM kinase does not appear to extend to non-lymphoid tumours. The difference may be because the mechanism for development of these solid tumours is different to the translocation-driven lymphoid tumour.