Heterozygosity for mutations in the ataxia telangiectasia gene is not a major cause of radiotherapy complications in breast cancer patients.

Of patients being treated by radiotherapy for cancer, a small proportion develop marked long-term radiation damage. It is believed that this is due, at least in part, to intrinsic individual differences in radiosensitivity, but the underlying mechanism is unknown. Individuals affected by the recessive disease ataxia telangiectasia (AT) exhibit extreme sensitivity to ionizing radiation. Cells from such individuals are also radiosensitive in in vitro assays, and cells from AT heterozygotes are reported to show in vitro radiosensitivity at an intermediate level between homozygotes and control subjects. In order to examine the possibility that a defect in the ATM gene may account for a proportion of radiotherapy complications, 41 breast cancer patients developing marked changes in breast appearance after radiotherapy and 39 control subjects who showed no clinically detectable reaction after radiotherapy were screened for mutations in the ATM gene. One out of 41 cases showing adverse reactions was heterozygous for a mutation (insertion A at NT 898) that is predicted to generate a truncated protein of 251 amino acids. No truncating mutations were detected in the control subjects. On the basis of this result, the estimated percentage (95% confidence interval) of AT heterozygous patients in radiosensitive cases was 2.4% (0.1-12.9%) and in control subjects (0-9.0%). We conclude that ATM gene defects are not the major cause of radiotherapy complications in women with breast cancer.

Summary Of patients being treated by radiotherapy for cancer, a small proportion develop marked long-term radiation damage. It is believed that this is due. at least in part, to intrinsic individual differences in radiosensitivity, but the underlying mechanism is unknown. Individuals affected by the recessive disease ataxia telangiectasia (AT) exhibit extreme sensitivity to ionizing radiation. Cells from such individuals are also radiosensitive in in vitro assays, and cells from AT heterozygotes are reported to show in vitro radiosensitivity at an intermediate level between homozygotes and control subjects. In order to examine the possibility that a defect in the ATM gene may account for a proportion of radiotherapy complications, 41 breast cancer patients developing marked changes in breast appearance after radiotherapy and 39 control subjects who showed no clinicalty detectable reaction after radiotherapy were screened for mutations in the ATM gene. One out of 41 cases showing adverse reactions was heterozygous for a mutation (insertion A at NT 898) that is predicted to generate a truncated protein of 251 amino acids. No truncating mutations were detected in the control subjects. On the basis of this result, the estimated percentage (950o confidence interval) of AT heterozygous patients in radiosensitive cases was 2.4% (0.1-12.9%) and in control subjects (0-9.0%). We conclude that ATM gene defects are not the major cause of radiotherapy complications in women with breast cancer.
Keywords: ataxia telangiectasia: ATM, radiation sensitivity; breast cancer For most solid tumours. curatixve radiotherapy inx olves deliverinn a dose schedule at the limits of normal tissue tolerance. Most sideeffects lead to moderate functional impairment. but occasionallv these are severe and even life-threatening (Maher Committee. 1995). The severitx of normal tissue reactions after a given course of radiotherapy Xaries x idely from one patient to another. Severe reactions can often in part be explained by radiotherapv technique or by predisposing factors such as prior surgern. chemotherapy or diabetes. Nexvertheless. exen after allowing for known factors. considerable xariation still exists. The clearest evidence for this is the xxork of Turesson et al (1989. 1990). Thev measured earls and late manifestations of radiation skin damage under well-controlled conditions in breast cancer patients. some of whom hax-e been folloxwed up for oxer 10 y-ears. A standard treatment protocol was found to produce xvery different degrees of telangiectasia. ranging from a barely detectable response to a sexere reaction. Analysis of these clinical data by Tucker et al (1992) has suggested that -ariation in tolerance between patients is determined by differences in Indixvidual intrinsic radiosensitivitv. exen among patients wxho show no clinical symptoms of recognized radiosensitix e sy-ndromes. An understanding of the basis of these interpatient differences could lead to significant improvement in treatment by the individualization of the radiotherapy prescription. Ataxia telangiectasia (AT) is an autosomal recessive disorder that is characterized by cerebellar ataxia. oculocutaneous telangiectasia and a predisposition to cancer (Boder and Sedgxxick. 1958) Clinicallx. AT homozv5otes exhibit marked hypersensitixitv to ionizing radiation. and fibroblasts or ly mphocytes from AT homozygotes are highly radiosensitive in various in vitro assays (Gotoff et al. 1967: Taylor et al. 1975: Weeks et al. 1991: Jorgensen and Shiloh. 1996. Although AT itself is a rare disease. it is estimated that approximately 1 %c of individuals in the general population are AT heterozvgotes (Easton. 1994: Nagasaw-a et al. 1987. A number of in vitro studies have suggested that cells from AT heterozvgotes may exhibit an intermediate level of radiosensitivfitv between AT homozvyotes and controls (West et al. 1995). Moreover. cells from patients show ing adxerse normal tissue damage after radiotherapy hax e been shown to exhibit a dearee of cellular radiosensitivitv similar to that of AT heterozgootes (Johansen et al. 1996). Taken tooether these findings have led to the hypothesis that heterozvgosity for AT may account for some of the radiation complications obserx ed in clinical practice.
The AT gene (ATM) has recently been isolated (Saxvitsky et al. 1995). It is a large gene spanning approximately 200 kb of genomic DNA x-ith a transcript size of approximately 10 kb encoding a predicted protein of 3056 amino acids. The mutations thus far discoxvered are highly heteroceneous. and are distributed throughout the entire extent of the gene. The majority are null mutations resulting in premature termination of translation (Byrd et al. 1996: Gilad et al. 1996. In this study. xxe examined the Heterozygosity in ATM and RT cornplications 923

MATERIALS AND METHODS Study population
Between January 1986 and July 1994. 915 patients were entered into a randomized trial comparing three fractionation regimens after breast-preserving surgery for early-stage operable breast cancer. All patients attended the Royal Marsden Hospital. Sutton. or the Gloucestershire Oncology Centre. Cheltenham. A total of 835/915 (91%) patients had baseline post-operative photographs of the breast. against which later radiation-induced changes scored from photographs were compared on an annual basis. The clinical and treatment characteristics of these 835 patients are summarized in Table 1. At the time of assessment. 735 of these had at least one follow-up photograph and made up the study sample.

Radiotherapy
The duration of whole-breast radiotherapy was 5 w-eeks in all arms. involving five treatments a fortnight for patients randomized to 13 fractions (3.0 Gy or 3.3 Gy per fraction) and five treatments per w-eek for patients in the third arm (2.0 Gy per fraction). Patients were treated in a supine position and most patients were treated with 6-MV X-rays. The breast was encompassed by opposed tangential fields using 15-30' wedges as tissue compensators. Radiotherapy to the lymphatic pathways was included at the discretion of the clinician depending on disease stage and axillary surgery. An electron boost to the tumour bed of 14 Gy to the 90% isodose in seven daily fractions was given to all patients with cancer cells at the microscopic margins of resection. In patients with complete microscopic resection of the primary tumour. an option to randomize the boost (boost vs no boost) was offered with patient consent. A boost was otherwise given routinely.

Definition and assessment of end points
The pnrmary end point of the trial. which was used in this analysis. relates to normal tissue responses in the breast as assessed by serial photographs. Frontal photographs of both breasts were taken after primary surgery and repeated annually for 5 years. All photographs were reviewed by three independent observers (two clinicians and one senior nurse) blind to patient identity. fractionation allocation and year of follow-up. Inclusion of the contralateral breast at each time point made it possible to distinguish radiotherapy effects from other time-related changes. e.g. w-eight gain. Changes in breast appearance caused by radiotherapy were scored on a three-point graded scale (none/minimal. 0: moderate. 1: marked. 2) based on change in breast size and/or shape. usually shrinkage. Interand intra-observer v-ariability were monitored by comparing, scores between observers. All discrepancies between observers were re-evaluated. Intra-observer variabilitv was evaluated by assessing the reproducibility of scores for each obser er by reassessing a random sample of photographs. Degree of agreement between scores was assessed using a weighted kappa statistic.

Case-control selection
Cases were defined as all individuals developing marked changes (grade 2) at any time between 1 and 5 years post radiotherapy or moderate changes (grade 1) scored for at least 3 years as assessed by clinical photographs. We identified 56 patients in these categories. 41 of whom were available for study. Control subjects were defined as individuals with no tissue reaction (grade 0) at the same time since radiotherapy as the case experienced a reaction. We identified 39 control patients. matched as closely as possible for the factors listed in Table 2 Table 4) wxere obserxed in the course of the mutational screen of ATM. Of these. only one

Sexeral sequence vaniants summarized in
x as predicted to generate a truncated protein. This mutation w as an insertion of A at nucleotide (nt) 898 in exon 8 and was heterozygous. The predicted consequence is the production of a truncated protein including the N-terminal 251 amino acids. a product only 8%7, of the normal size. This xariant wxas in a case xwith marked (grade 2) radiotherapy changes in breast appearance. No truncating mutations were detected in any of the 39 control subjects.
An additional radiosensitixe case was heterozvgous for a G-sA transition at nt 4108 leading to substitution of Arg for GIv at amino acid 1306. This sequence xariant was not found in the breast cancer control subjects or in 147 healthys women. Gly 1306 is conserxed in mouse ATV (mouse ATM is 95%7 identical to human ATM) but is not within the kinase domain that shows substantial similaritx to other members of this gene family. As 80-90%c of ATTM mutations result in truncated proteins. at present it is difficult to determine whether this is a rare innocuous polI morphism or a mutation deleterious to ATM function.
Three other sequence xariants xere obserxed in a single case but not in anx of the 39 control subjects. Two of these are intronic. insertion T at nt 160-5 and G -* A at nt 2438 + 80. Neither of these change consensual splice sequences and therefore are likely to be rare polI-morphisms. The third alteration is a non-coding change. C -*T at nt 7710 (Ala -* Ala).
Sexen sequence x-ariants xere detected in a sinale control but not in any of the 41 cases. These include: twxo intronic chan,ces.
G -s T at nt 2088-39 and ins A at nt 3027 + 28: a xariant in the 3' untranslated region. C -* G at nt 9389: txo non-coding changes. T -* C at nt 5982 (Ala -* Ala) and G A at nt 7251 (Ala Ala): and two missense coding xariants G A at nt 7572 (Arg His) and C -s T at nt 8683 (gAr -His).
The remainder of sequence x ariants w-as observed in both cases and control subjects and no substantial differences in heterozygote frequency (as ascertained from CSGE gels) betu-een cases and control subjects w-ere obserxved.
From these results the only sequence x-anant that is confidently predicted to alter ATM function is the heterozygous insertion of A at nucleotide 898 in exon 8.

DISCUSSION
A total of 80 patients 141 cases and 39 control subjects) selected from 735 exaluable wromen with earlv breast cancer randomized into a radiotherapy fractionation study were screened for mutations in ATM. One out of 41 cases showved a typical mutation that w-as predicted to generate a truncated protein (insertion A at nucleotide 8981. This case had no other predisposing factors for radiation damage and dexeloped marked breast shrinkage w-ith moderate cutaneous telangiectasia follow ing 39 Gv-in 13 fractions (approximately equix alent to 46 Gy in 23 fractions of 2.0 Gyl. No truncating mutations w ere detected in any of the 39 control subjects. It is likelx that the mutational screening technique used will miss a minoritv of mutations. particularlof single base substitutions and large genomic rearrangements. and therefore the numbers reported may be underestimates. Nevertheless. the results suggest that ATM mutations are unlikely to account for a substantial proportion of patients with dose-limiting complications of radiotherapy (although a small contribution cannot be excluded). These results are consistent w-ith prexious reports of three AT heterozvootes who had radiotherapy for breast cancer without unusual reactions (Ramsav et al. 1996: Fitzcerald et al. 1997) and 16 breast cancer cases show-ing radiotherapy complications in whom ATM mutations wxere not detected (Applebv et al. 1997).
From studies of relatix-es of AT patients. there is exvidence that AT heterozvgositv may be associated with an increased frequency of certain types of cancer. particularly breast carcinoma (Swift et al. 1987. 1991: Pippard et al. 1988. Additional evidence supporting this hypothesis has recently been obtained by genetic link-age analyses of families of AT cases using mark-ers in the xicinitv of ATM on chromosome 1 l q (Athma et al. 1996). How ever. direct examination bv mutational screening of the ATM gene rexealed mutations in 2/401 wAomen w-ith breast cancer compared w ith 2/202 control subjects (Fitzgerald et al. 1997). W'hereas these data do not exclude a role for ATM as a low--penetrance breast cancer susceptibility gene (Bishop and Hopper. 1997). they do not lend strong support either. Although the present study is not a formal test of this hypothesis because there is no matched control group and the numbers are small. detection of a single AT heterozvoote in 80 breast cancer cases does not add further w-eight to the notion that ATM is a low-penetrance breast cancer susceptibility gene.
Radiotherapy-induced breast shrinkage and distortion changres in a proportion of women after radiotherapy are progressix e. permanent and of clinical relexvance to the patient. They are also clearly related to radiotherapy dose. In the clinical trial from w-hich these patients are drawn. a 10%k difference in randomized dose (42.9 Gv in 13 fractions vs 39 Gv in 13 fractions) w-as associated with roughly a tw-ofold difference in the chance of breast shrinkage (Owen et al. 1994). It has been shown in this study that testinc for AT heterozy gosity does not appear to offer a worthwhile approach for the identification of the radiosensitive subgroup of breast cancer patients and the search for the genetic loci responsible should continue.