The A-T syndrome varies little from family to family in its late stages.1-6 Its primary features include (a) progressive gait and truncal ataxia with onset from 1 to 3 years of age; (b) progressively slurred speech; (c) oculomotor apraxia, i.e., an inability to follow an object across the visual fields; (d) oculocutaneous telangiectasia, usually by 6 years of age; (e) elevated serum α-fetoprotein; (f) frequent infections, with accompanying serum and cellular immunodeficiencies; (g) susceptibility to cancer, usually leukemia or lymphoma; (h) hypersensitivity to ionizing radiation, contraindicating the use of conventional doses of radiation therapy for cancer; and (i) reciprocal translocations that involve chromosomes 7 and 14 almost exclusively. Other features include premature aging and endocrine abnormalities. Fig. 29-1 highlights some of the major features of this complex syndrome.
Neurology and Neuropathology
The most obvious and disabling characteristic of the A-T syndrome is the progressive cerebellar ataxia. Shortly after learning to walk, A-T children begin to stagger. By 10 years of age, they are confined to a wheelchair for the remainder of their lives. The ataxia begins as purely truncal but within several years also involves peripheral coordination. Deep tendon reflexes are decreased or absent in older patients; plantar reflexes are upgoing or absent. Slurred speech and oculomotor apraxia are noted early. Both horizontal and vertical saccadic eye movements are affected. Writing is affected by 7 or 8 years of age. Choreoathetosis is found in almost all these patients. Myoclonic jerking and intention tremors are present in about 25 percent. Drooling is a frequent complaint. All teenage A-T patients need help dressing, eating, washing, and using the toilet. The neurologic status of some patients appears to improve between 3 and 7 years of age, and then begins to progress again; this is probably due to the rapid neurologic learning curve of young individuals. Muscle power is normal at first, but wanes with disuse, especially in the legs. Arm strength generally remains. Contractures in fingers and toes are common in older patients, but may be prevented through rigorous exercise.
The typical patient with A-T is of normal intelligence, although slow responses make it difficult to support this by timed IQ testing. Many American and British patients have finished high school with good grades; some have finished college or university. A few seem to be minimally retarded. Most patients have excellent memories.
The most obvious lesion in the central nervous system at postmortem examination is the paucity of Purkinje cells (PCs) in the cerebellum. About 10 years ago, my laboratory wished to determine whether these cells are absent from birth or degenerate afterward. Knowing that basket cells form only around preexisting PCs, we sought to visualize basket cells by Bielschowsky silver-staining. We showed that normal or nearly normal numbers of basket cells were present (Fig. 29-2). We therefore concluded that PC numbers must also have been normal or near normal at birth and degenerated after birth.7,8
Neuropathology. Photomicrographs of the cerebellum. M—molecular layer; G—granular cell layer. A, Bielschovsky stain of normal cerebellum (10×) showing basket cell fibers surrounding Purkinje cells (arrow). B, Bielschovsky stain of cerebellum from an A-T patient (10×) showing empty basket cells (arrow) in the Purkinje cell layer. C, In situ hybridization with ATM cDNA on normal infant cerebellum (40×) showing a Purkinje cell (arrow) with a positive peroxidase stain for ATM mRNA. D, In situ hybridization with ATM cDNA on cerebellum from an A-T patient (20×) showing ectopic (overmigrated) Purkinje cells (arrow) with a positive peroxidase stain for ATM mRNA. (Courtesy of S. Becker-Catania.)
We also found evidence suggesting that PC migration and arborization are not completely normal.8 A significant number of ectopic PCs can be found in cerebellar sections from A-T patients from both undermigration and overmigration (Fig. 29-2). PCs make their last cell division at about 13 weeks of gestation and then begin to migrate toward the pial surface. Following the tracts of climbing fibers, they arrive at the single-cell PC layer during the fifth to seventh month of gestation. Thus, this lesion of ectopic PCs would most likely be expressed by the last trimester of pregnancy. This is the earliest known manifestation of the A-T defect.
It remains possible that PCs are not the primary A-T lesion and that the observed PC defects are due to other factors, such as an absence or abnormalities of supporting cells such as basket cells, mossy fibers, parallel fibers, climbing fibers, or glial cells. The frequent presence of choreoathetosis suggests that the basal ganglia, not the cerebellum, are the primary site of neuropathology in A-T. Anterograde or retrograde degeneration of PCs would then occur, with the underlying lesion being either afferent or efferent to the PCs. Becker-Catania, in our laboratory, was able to demonstrate the presence of ATM message in PCs, as well as in cells of the internal and external granular layers and in neurons of the dentate nucleus. ATM mRNA is seen in both healthy and affected tissues (Fig. 29-2). However, because most mutated ATM proteins would be truncated and unstable, the primary site of pathogenesis in the cerebellum remains unclear. This has important implications when considering where to target gene or cell therapy.
Changes also have been noted in the dentate and olivary nuclei. The medulla shows neuroaxonal dystrophy.9 Degenerative changes are seen in the substantia nigra. Diffuse demyelination in the posterior columns of the spinal cord was noted in some of the original autopsies1 and is a progressive change. For further details of postmortem changes in A-T patients, consult the two lengthy reviews authored by Sedgwick and Boder.2,5
Nucleomegaly is a universal finding throughout the organs of A-T patients.2,5 Nucleomegaly is best seen in organs where nuclear morphology is very regular, such as the hepatic cords and renal tubules. Here it is obvious that the size of the nucleus is extremely variable in A-T tissues, as compared to normal. Some nuclei are very large, hyperchromatic (dark staining), and irregular in shape. Nucleomegaly also is seen in association with normal aging and in viral lesions. Numerous studies, however, have failed to demonstrate virus or viral particles in A-T cells, including Gadjusek's attempt to inoculate primates with brain-tissue extracts from two A-T patients.5 It is entirely possible that the nucleomegaly seen in A-T tissues results from defective cell-cycle checkpoints that lead to mitotic division without cell replication in random cells. Naeim et al. 10 demonstrated polyploidy (4n and 8n) in lymphoblastoid cell lines (LCLs) from A-T patients by flow cytometric cell-cycle analysis.
It was expected that by developing mouse strains in which the atm gene was made nonfunctional by one means or another, animal models would become available for dissecting the neuropathology of A-T. Unfortunately, at least five independent atm knockouts (atm-/-) have failed to show the severe progressive ataxia seen in A-T patients.16,169,203,222,223,254,255 This observation limits the application of neuropathology findings in atm-/- mice to patients. Kuljis and Baltimore254 described changes in the cerebellar cortex when the tissues were examined by electron microscopy. More recently, Borghesani et al. 203 described an atmy/y strain that survives beyond 1 year. The mice do not show a progressive ataxia; however, they do show significant changes in Pcs and they manifest motor learning deficits compatible with perturbed cerebellar function. Based on in vitro irradiation of yet another atm-/- strain, Herzog et al. 256 suggested that inappropriate cell death and apoptosis may underlie the abnormal neurologic development. All of these knockout strains show changes in radiosensitivity, immunologic development, and marked cancer susceptibility.
Fig. 29-3 shows a typical pattern of telangiectasia in a 12-year-old patient. Telangiectasias aid in diagnosis. Telangiectasias can be seen on the conjunctiva, as well as on the ears, over the bridge of the nose, in the antecubital fossae, and behind the knees in some patients. Occasional patients have them all over their bodies. Telangiectasias usually do not appear until about 4 to 6 years of age, and although they are a hallmark of the disorder, they sometimes do not become obvious for several years after the onset of ataxia. Elderly individuals without A-T occasionally have similar telangiectasias in many of the same places. Boder felt that telangiectasias appear in response to ultraviolet light exposure; however, that would not explain finding them behind the knee in some patients. About 5 percent of A-T patients never develop prominent telangiectasias. These tend to be patients with milder symptoms. Cafe-au-lait spots are found in almost all A-T patients but are not pathognomonic for just A-T.
Characteristic telangiectasias over the conjunctiva of a 12-year-old A-T patient.
Telangiectasias are composed of dilated capillaries. The pattern of these capillaries does not resemble a response to angiogenic factors; instead, it appears to be a response of endothelial cells to a dilatory stimulus. On the other hand, it is still possible that the propensity of A-T patients to develop tumors and form telangiectasias reflects a defective balance between activation of the p53 pathway, apoptosis, angiogenesis, and oxidative stress.317,325 Van Meir et al. have shown that at least one pathway for inhibition of angiogenesis is through p53, 11 and that p53 expression and phosphorylation are reduced in A-T cells (see “Cell-Cycle Aberrations,” below). Recent studies of gene expression arrays in several laboratories find that unstimulated A-T cells are already in an elevated state of oxidative stress. With rare exceptions, 1,12-15 telangiectasias are not found internally at surgery or at postmortem examination, nor do atm knockout mice manifest internal telangiectasias.16 Amromin et al. 15 noted widely distributed gliovascular nodules in the cerebral white matter and, to a lesser extent, in the brain stem and spinal cord postmortem in a 32-year-old patient. These nodules consisted of dilated capillary loops, many with fibrin thrombi, with perivascular hemorrhages and hemosiderosis, surrounded by demyelinated white matter, reactive gliosis, and numerous atypical astrocytes. These nodules were not seen in the cerebellum and have not been observed in other postmortem examinations of younger patients.
Telangiectasia occasionally develop within the fields of prior radiation therapy, not only in A-T patients or carriers, but in apparently normal persons as well.289 Telangiectasia are sometimes observed in the parents and sibs of A-T patients and may be a subtle manifestation of heterozygosity and radiosensitivity.
Radiosensitivity in vivo and in vitro
Over the past 30 years, radiation therapists have observed that when A-T patients with cancer are treated with conventional doses of ionizing radiation, they develop life-threatening sequelae characteristic of much higher doses.17-21 This radiosensitivity can also be demonstrated in vitro using fibroblasts or lymphoblasts from A-T homozygotes, which are sensitive to ionizing radiation and to a variety of radiomimetic and free-radical-producing agents.22,23,111 (Also see the discussion of colony survival assay under “Differential Diagnosis” below.)
Early radiosensitivity assays for A-T measured colony formation efficiency of fibroblasts. Cells are irradiated and then cultured, and the number of colonies that grow are scored after a measured period of incubation. Fibroblasts from A-T heterozygotes form colonies with an efficiency that is intermediate between A-T homozygosity and normal;22 the same observation can be made using neocarzinostatin.24 Despite this, colony-forming efficiency is not a reliable way to detect individual heterozygotes.23-26 Many other methods have been tried to identify A-T heterozygotes, but none are reliable because the normal and heterozygous data sets overlap.10,27-31 Recently, a “comet” assay has been described for identifying ATM heterozygotes by measuring DNA repair in peripheral blood lymphocytes (BPL) following 3 Gy of irradiation.315 .
Now that genetic testing in families with prior affected patients can identify A-T heterozygotes (carriers) more easily, physicians are confronted with the dilemma of having to advise carriers about the risks of radiation exposure. Unfortunately, there are as yet no clinical studies on which to base such advice. For example, data are lacking on whether the in vitro radiosensitivity of heterozygotes has any clinical correlate; i.e., unusual reactions to standard radiologic procedures or increased cancer risk. One can only make the general recommendation that exposure to all types of ionizing radiation be minimized in persons suspected of being A-T carriers, e.g., both parents, remembering as well that two-thirds of the sibs of A-T patients are likely to be carriers. Whether routine dental x-rays should be recommended in A-T patients is also an unresolved issue.
ATM knockout mice that are heterozygous (atm +/−) do not show an abnormal response to total-body irradiation.16 Some caveats: (a) Only one specific site in the ATM gene was disrupted in each of the knockout mice strains, and (b) knockout mouse models seldom mimic a human disease in all facets of the syndrome. Swift's epidemiologic data suggest that exposure of heterozygotes to myelograms and other diagnostic x-rays may increase their cancer risk.32 There is also cause for concern about mammograms in female A-T carriers, who appear to be at an increased risk of breast cancer (see below). The recommendation at this writing is that mammograms be continued on the same age-dependent schedule used for noncarriers but that the most up-to-date mammography machines be employed to minimize exposure to only a few rads. The added risk of cancer to such women is only slightly increased (from 1.5 in 100 from annual mammography screening doses in noncarriers to perhaps 2 in 100 in A-T carriers), and this should be compared to a 1 in 9 natural lifetime risk and the 30 percent reduction in mortality from annual mammography screening in women over age 40.33-35
Epidemiologic and radiosensitivity studies of A-T family members further suggest that many cancer patients may be receiving the wrong doses of radiotherapy—too much for A-T heterozygotes and too little for noncarriers.33,34,36 Considering that 5 percent of cancer patients under 46 years of age may be A-T carriers and intermediate in their radiosensitivity, 37 this issue could involve many thousands of patients annually (see “Cancer Susceptibility,” below). Some x-ray dosage regimens were first tested empirically on cadres of cancer patients, and if those cadres were to have contained up to 5 percent A-T carriers, one can easily imagine how radiation sequelae might be more apt to appear in the A-T carriers, thereby lowering the “safe” doses defined for everyone else. Once A-T carrier testing can be implemented on a wide scale, these issues may be resolved.
Although radiation damage to DNA has been used for many years as a laboratory tool for characterizing the phenotype of A-T cells, it should be remembered that A-T cells normally are not exposed to irradiation in vivo. Thus, the radiosensitivity of A-T cells is largely a laboratory artifact because irradiation damage mimics double-strand DNA breaks and tests the cells' abilities to rejoin these breaks in an orderly fashion. Because the complex A-T syndrome develops quite uniformly in most affected patients, the major substrate for ATM protein must be a naturally-occurring molecule(s) generated by an equally common event that causes double-strand DNA breaks. p53 is a good candidate for the major substrate, 198,199 while oxidative stress and the normal production of DNA-damaging free-radical products of metabolism are good candidates for the commonly-recurring and inciting event.38,328,329 Although many attempts have been made to demonstrate defective DNA repair in A-T cells, this has not been convincingly defined.39 Recent observations suggest that it is the inability to sense this damage that is defective in A-T cells, not the actual mechanisms of repair.38,40
When DNA synthesis of irradiated fibroblasts is measured, A-T homozygotes show a characteristic dose-response curve (Fig. 29-4) that is diagnostic of the disease.82 This phenomenon is called radioresistant DNA synthesis (RDS) because, unlike normal cells which temporarily halt the synthesis of new DNA after irradiation damage, A-T cells simply continue into S phase of the cell cycle. Later experiments that attempted to complement RDS and other radiomimetic features of A-T cells by transfection found, however, that these phenomena often were dissociated.6,83 Thus, RDS most likely reflects a cell-cycle checkpoint failure at S phase that is independent of other radiosensitivity phenotypes of A-T cells. Using RDS, heterozygous cells cannot be distinguished from normals.
Graph of radioresistant DNA synthesis (RDS) depicting theoretical targets of radiation at increasing doses of radiation. The major defect in A-T (black squares) is seen at very low radiation doses, affecting replicon initiation, while chain elongation appears normal in A-T, as depicted by the normal slope of that portion of the curve.
The shape of the RDS curve (Fig. 29-4) for A-T cells has been insightfully interpreted by Painter85,86 as having two components, one reflecting replicon initiation and the other reflecting chain elongation. The slope of the curve above 20 Gy determines the second component, and this slope does not differ in normal and A-T fibroblasts. The early component, however, is essentially missing from the A-T curves, suggesting that replicon initiation is quite abnormal. Painter further suggested that while in unirradiated normal cells initiation occurs synchronously at the origins of a cluster of replicons and that in irradiated normal cells damage to one replicon inhibits the entire cluster, in irradiated A-T cells damage to one replicon inhibits only that replicon; i.e., the damage is not sensed or translated to the rest of the cluster, and chain elongation near the growing forks is not curbed. Thus, radiation to A-T cells blocks initiation of individual replicons rather than blocking the initiation of clusters of replicons.39,85,86 Hand and Gautschi87 provided evidence that one single-strand break may inactivate the initiation of as many as 100 replicons. Although Painter's interpretation of the RDS curve was put forth over 10 years ago, it remains a very attractive hypothesis for explaining one facet of the pathogenesis of A-T. Atomic force microscopy analyses suggest that ATM protein can exist either as a monomer or as a tetramer in the repair of double-strand DNA breaks.134,200-202,318
During their shortened lifetimes, 38 percent of A-T homozygotes develop a malignancy.41 This represents a 61- and a 184-fold increase in European-American and African-American patients, respectively. Roughly 85 percent of these malignancies are either leukemia or lymphoma. In younger patients, an acute lymphocytic leukemia is most often of T-cell origin, 42,43 although the pre-B origin common to ALL of childhood (HLADR1, CD101, CD51, and CD191) has also been seen in A-T patients. When leukemia develops in older A-T homozygotes, it is usually an aggressive T-cell leukemia with a morphology similar to that of chronic lymphoblastic leukemia, hence the old name T-CLL;42-45 T-cell prolymphocytic leukemia (T-PLL) is the equivalent modern nomenclature.45,46 The leukemic cells often contain a translocation and/or inversion involving the T-cell receptor α-chain gene complex at 14q11-1247-49 (see discussion of non-leukemic clonal expansions in A-T patients under “Chromosomal Instability,” below). Myeloid leukemia is very uncommon in A-T patients. Lymphomas in A-T homozygotes, in contrast to leukemias, are common and are usually B-cell types, although T-cell lymphomas also have been observed. As A-T patients are living longer, more nonlymphoid cancers are being observed. Several of our older patients have developed breast cancer and melanoma. Cancers of the stomach and ovary have been reported.5,42 When fibroids or leiomyomas are found in A-T females, a special effort should be made by the pathologist to quantitate high-power fields for mitotic figures, because leiomyosarcomas of precocious onset have been reported.50 For specific lists of other tumor types and frequencies, see references41,42, and 44 .
A-T heterozygotes are also believed to be cancer-prone.37 They are not clinically distinguishable from normal individuals. An increased incidence of breast cancer in female A-T heterozygotes has been reported in the United States, England, and Norway.32,51-53 In the U.S. study, the risk of breast cancer was found to be fivefold higher among the mothers of A-T homozygotes than in a comparable female population.32,52 Based on this observation, Swift et al. estimated that between 8 and 18 percent of all breast cancer patients may be A-T heterozygotes.37 This would imply that ATM is the most common cancer susceptibility gene in the general population. While the issue is far from settled, many recent genetic studies of breast cancer cohorts have failed to find an increased frequency of mutations in the ATM gene, 54,55,257,258,306,331 with the exceptions of a report of an allelic variant at the ATM locus being associated through a rare HRAS1 allele in a stratified study of 66 sib pairs affected with breast cancer260 and recent findings be Teraoka et al.305 of 11 ATM mutations in 142 breast cancer patients versus only 1 in 80 controls. They used primarily denaturing high-power liquid chromatography (dHPLC); all mutations detected were of the missense type. Vorechovsky et al. 54 looked for ATM mutations in tumor tissue from 38 sporadic breast cancers. They first screened this material by single-stranded conformational polymorphism (SSCP) gels, and then sequenced any regions suspected of harboring mutations. They found no significant mutations. Spurr and coworkers56 looked for a linkage to BRCA1 and BRCA2 in 63 early onset breast cancer families; 55 percent linked to BRCA1 and 45 percent linked to BRCA2. This implied that none linked to 11q22-23, as Wooster et al. 58 and Cortessis et al. 57 had reported earlier. However, because the epidemiologic data from A-T families suggested that the breast cancer seen among A-T mothers (who are obligate heterozygotes) peaks in the age group of 45 to 54, 52 and is not an early onset pattern, other studies have screened late-onset or sporadic breast cancers for ATM mutations. Using PTT, FitzGerald et al. 55 detected ATM mutations in 2 of 401 women with sporadic early onset breast cancer. They also found mutations in 2 of 202 control samples. Recently, Swift and coworkers59 again noted an increased incidence of breast cancer in A-T heterozygotes who were identified by haplotyping members of extended A-T families.
The apparent paradox of (a) not finding many ATM mutations in breast cancer cohorts and (b) not finding A-T patients in the families of breast cancer patients with ATM mutations may find resolution in the hypothesis that two types of A-T carriers may exist within the general population, those with nonsense (truncating) mutations and those with missense mutations, and each type of mutation may have a different Phenotype—both in the heterozygous and in the homozygous state. Heterozygous missense mutations might create a dominant negative situation, 330 whereby the defective copy of the ATM protein binds to sustrates but cannot phosphorylate them, or vice versa. This would create a far more serious situation than a nonsense mutation that would not produce any stable defective protein. Most A-T patients have two truncating mutations. Homozygous missense mutations are extremely uncommon in A-T patients and some could even be lethal. Thus, it is possible that ATM missense carriers are more likely to be cancer prone than ATM nonsense carriers. (This hypothesis is developed further in reference306 .)
Several independent studies of ATM knockout mice have confirmed the extreme cancer susceptibility of atm−/− homozygotes; all atm−/− strains develop massive, widely metastasized malignant thymic lymphomas.16,169,203,222,223,254,255 Heterozygous animals have not shown tumors, nor has breast cancer been observed in either homozygous or heterozygous animals.
Carter et al. reported significant loss of heterozygosity in sporadic breast cancers across chromosome 11q22-2360 as did others.267-270,297,298 Although this region includes the ATM gene (at 11q22.3-23.1) and the CHK1 gene (at 11q23.3), 271 it measures 35 cM and probably includes more than 1000 other genes as well. Thus, the contribution of ATM mutations to familial breast cancer appears to be low, with ATM perhaps playing a more important role in a sporadic, low-penetrance form of breast cancer.
The most frequently reported cancers in American A-T heterozygotes are breast, trachea/bronchus/lung, stomach, prostate, melanoma, and gallbladder.32 In Italy and Costa Rica, gastric cancer has been especially noteworthy. Among 64 A-T parents and grandparents in Costa Rica, half of the 12 cancers reported were gastric cancer.61 (Costa Rica ranks among the top three countries in the world for stomach cancer in the general population, making it difficult to interpret these observations.) In Italian families, 7 of 20 cancers in grandparents were gastric cancer.62 Stomach cancer has been reported in homozygotes as well, 42,44 including two families in which both affected sibs developed stomach cancer. Despite this, Morrell et al. 41 did not note any increase in stomach cancers in A-T homozygotes.
Other recent observations move the cancer susceptibility association with A-T in new directions. Most individuals who develop T-PLL have ATM mutations in one or both alleles. While it was first thought that these patients represented constitutional A-T heterozygotes, the data are lacking to establish this.259 Thus, these patients may be at no increased prior risk of cancer but simply develop an ATM mutation somatically. The ATM mutations found in T-PLL cells are mainly missense mutations, 84 unlike the predominantly nonsense mutations found in A-T patients237 . However, if A-T heterozygotes acquire T-PLL by a second “hit”84,129 on the ATM gene, this would then suggest that ATM can function as a tumor-suppressor gene, in addition to its growing list of other roles.264
Four reports link loss of ATM integrity to an aggressive subgroup of B-cell chronic lymphocytic leukemia (B-CLL) patients.301-304 Perphas 40 percent of B-CLL patients have 11q deletions or fail to express ATM protein. Here again, where ATM mutations have been identified, they have been mainly missense types. Unfortunately, missense mutations are still very difficult to detect in such a large gene (see below).
A gain of chromosome 3q is present in many cancers, including cervical carcinomas, small cell lung carcinomas, head and neck squamous cell carcinomas, and embryonal rhabdomyosarcomas.261 When microcell hybrids were transferred into a differentiation-competent myoblast cell line C2C12, the cells exhibited a nondifferentiating phenotype.262 Selecting 3q candidate genes, ATR (AT- and rad3-related/FRAP-related protein 1) was tested. ATR is a protein kinase with strong homology to ATM.71,72,233 It was found that forced expression of ATR resulted in a phenocopy of the 3q-containing microcell hybrids. ATR apparently inhibits MyoD, which is a marker for classifying sarcomas as rhabdosarcomas. ATR is thought to share functional overlap with ATM in cell-cycle progression139,263 and may be phosphorylated by ATM. Like ATM, ATR phosphorylates the Serine 15 position on p53, albeit at a much-reduced level and at a slower rate.199 Furthermore, overexpression of ATR corrects the defective radiation-induced S-phase checkpoint in A-T cells.263 (ATR's relationship to ATM is further described in “Chromosomal Instability” below).
Important checkpoints monitor the progress of the cell cycle and prevent mutagenic damage to DNA from becoming fixed into future cell generations. The G1 checkpoint prevents replication of a damaged DNA template; the G2 checkpoint prevents segregation of damaged chromosomes.64 Kastan et al. 65,66 showed that A-T cells have a delayed radiation-induced increase in p53, compared to normal cells. p53, dubbed the “guardian of the genome,” acts to suppress normal cell-cycle progression at G1 until DNA repairs have been completed. (For reviews pertinent to p53 in A-T cells, see references38,40, and 67 .) It accomplishes this by binding DNA at sequence-specific sites, thereby transcriptionally activating a signal-transduction cascade. In so doing, p53 functions as a tumor suppressor. Cells from p53 knockout mice, lacking both normal alleles of p53 (p53−/−), fail to observe the G1 checkpoint; they do not experience G1 arrest after irradiation nor do they show neurologic abnormalities, immune defects, or problems with sterility, 40,67,68 as in A-T. The strong association of multiple types of cancer with p53 deficiency67,74-77 further suggests that involvement of this pathway in apoptosis and in differentiation may help explain the increased frequency of cancer in A-T.
Interestingly, p53−/− mutants are not radiosensitive.76,78 Thus, yet another mechanism must account for the radiosensitivity of A-T cells. Much work still needs to be done before the role of ATM proteins in intracellular signaling can be fully appreciated. For example, despite much evidence of the inefficiency of G1/S, S, and G2/M checkpoints in A-T cells, holding A-T cells in G0 for up to 7 days does not improve their postirradiation survival, 40 which suggests that even these checkpoint defects may not be the crucial common denominator underlying A-T pathogenesis. Evidence presented by Jung and et al. 79,80 using SV40-transformed fibroblasts, implicates the NF-kB and IkB-α proteins in ATM function; the ATM protein appears to phosphorylate IkB-α, thereby activating the transcription factor NF-kB.81 These findings would also support observations of increased radiation-induced apoptosis in cell lines derived from A-T patients.40,76,77 However, a recent report by Ashburner et al. 324 suggests that the constitutive activation of NFκB reported by Jung et al. may be due more to SV40-transformation than to the A-T phenotype. The phosphorylation of replication factor A (RPA) is also delayed in A-T cells after irradiation.69,265 Three groups have independently shown that the Serine-15 position of p53 is selectively phosphorylated by ATM.198,199,300 Further, Shafman et al. 70 have reported that c-abl binds to an SH3 domain on the ATM molecule.
At postmortem examination, virtually every A-T homozygote has a small embryonic-like thymus.12 In the late 1960s, and again in the early 1980s, many attempts were made to characterize the immunodeficiencies of A-T patients.88 No single, consistent abnormality could be identified in all A-T patients; affected sibs often differ in the degree and profile of their immunodeficiencies. In a review of British patients, Woods and Taylor89 noted normal immunologic function in 27 of 70 patients. Only 10 percent had severe immunodeficiencies.
When the genomic order of the IGH V, D, J, and H gene subfamilies was first described, we noted that the immunoglobulin (Ig) classes that were most frequently decreased in A-T patients were those with the greatest genomic distances between the variable (V) genes and the respective heavy-chain genes;90 60 to 80 percent of A-T homozygotes manifest an IgA, IgE, and/or IgG2 deficiency, 12,88,89-95,121,122 whereas serum levels of IgM, IgG1, and IgG3 are usually normal. This suggests that B cells from these patients have a maturational problem with Ig class switching, perhaps based on a recombinase-related deficiency. On a similar note, an increased proportion of T gamma/delta cells noted in one early study suggested a maturational delay in T cells; however, this was before normal T gamma/delta cell ranges had been clearly defined and has not been generally confirmed. IgM levels are occasionally extremely high (see below), which could be based on a similar defective maturational mechanism that arrests some B cells at the IgM-producing stage. However, when V(D)J recombination was examined in A-T cells, both signal and coding joint formation were normal.96-98 Approximately half of A-T patients with immunodeficiencies have T-cell deficiencies. CD41/CD45RA1 (naive) T cells are decreased in some patients.99 Responses to antigens are poor, especially allogeneic antigens.12,88,100-105 T-cell cytotoxicity to influenza-infected target cells is reduced.106 T lymphocytes show abnormally fast capping of FITC-labeled concanavalin A.88 Markedly elevated cyclic AMP levels have been observed in T cells from A-T patients.88 Neutrophil chemotaxis was reported to be decreased in some studies and normal in others. Similarly, NK cell activity and NK cell levels have been described as normal, decreased, or increased in various studies.88,107,108,266 Some of these discrepancies no doubt reflect the transient immune status of patients with active infections. Although 91 percent of Costa Rican A-T patients had diminished PHA responses, 65 percent of them had the same mutation; thus, this sample would be skewed against some features and would favor others, and probably has only minimal bearing on patients around the world with other mutations. Further immunologic analyses of this cohort are under way. Knockout ATM mice have many of these same immune defects as A-T patients; T and B cell precursors in thymus and bone marrow, respectively, are present in normal numbers.169
Sanal et al. 275 have recently described a new form of immunodeficiency in A-T patients. IgG antibody responses to pneumococcal polysaccharide vaccine (serotypes 3, 6A, 7F, 14, 19F, and 23F) were studied in 29 classic A-T patients; in 22 patients (76 percent), no responses were observed. The remaining patients had responses to 1, 2, or 4 serotypes. Zeilin et al. 327 support this finding.
Hyper-IgM with Ataxia-Telangiectasia
Elevated serum IgM levels are fairly common in A-T patients, 100,277 arising perhaps as compensation for low IgA, IgE, and IgG2 levels. However, occasional A-T patients with classic symptoms have an extended syndrome that may include very high serum IgM levels, splenomegaly, lymphoadenopathy, neutropenia, thrombocytopenia, hypertension, renal anomalies, and congestive heart failure from high blood viscosity.109,110 The latter symptoms were somewhat ameliorated by reducing blood volume, and further by splenomegaly. Steroids markedly improved three patients (unpublished, personal experience). The postirradiation colony survival assay (CSA)111 in six of these families, although not easily quantifiable, suggests a level of radiosensitivity that is intermediate between that seen in normals and that seen in other A-T patients (see “Differential Diagnosis,” below). In three families, ATM mutations have already been identified. It is of interest that in three families, the affected sibs were discordant for hyper-IgM.278-280 Another patient was atypical in that depletion of cerebellar Purkinje cells was not seen, and ATM protein levels were normal.109 In an Argentine family, the hyper-IgM followed treatment of the immunodeficiency with IVIg.281 Thus, while hyper-IgM and A-T have been observed together in a number of families, the underlying pathology remains obscure and the observation of discordant sibs in three families suggests that the hyper-IgM represents a somatic, not a genetic, variation. Recently, Rosenblatt et al. 296 have provided some evidence that this hyper-IgM may be due to an up-regulation of the CD40 ligand gene.
Although elevated serum α-fetoprotein (AFP) levels can be very useful in confirming a suspected diagnosis of A-T, 5 to 10 percent of typical A-T patients have normal AFP levels. This is independent of race, sex, or complementation group, and is usually concordant in affected sibs. AFP levels do not increase with patient age.89 Serum AFP levels still elevated from infancy sometimes can be misleading in children under 2 years of age in whom normal ranges have not been carefully defined by most clinical laboratories. Thus, it is best to avoid using AFP as a diagnostic criterion until after 2 years of age. Other causes of elevated AFP, such as liver disease, familial hyper-AFP309,310 and the presence of a teratoma, are not likely to confound a diagnosis of A-T. Ishiguro et al. 115 showed that the lectin-binding profile of elevated AFPs from A-T patients was most likely of hepatic origin, and although no evidence of liver disease is present at postmortem examination, other liver proteins, such as serum glutamic-pyruvic transaminase (SGPT), serum glutamic-oxalacetic transaminase (SGOT), alkaline phosphatase, and carcinoembryonic antigen, are often increased as well.90,112
AFP is thought to have a suppressor effect on the developing immune system and on immune function.112-114,311,312 The mechanisms by which AFP is elevated in sera of most A-T patients remains unclear but may involve the NFκB/IκBα complex and/or p53, both of which are phosphorylated by ATM.79,80,198,199,300,313
With the routine monitoring of AFP in amniotic fluid now in vogue, the question is occasionally asked whether amniotic AFP levels are elevated when the fetus has A-T. AFP levels are very high in all fetuses, peaking at about 13 weeks of gestation.112 In two cases who had been diagnosed by prenatal testing, and in which a decision had been made by the parents not to terminate the pregnancies, amniotic AFP levels were measured and were within normal ranges. A cord blood AFP was elevated in one of these patients, and remained so over the next 3 years. (In the other patient, cord blood was not tested.) Thus, although the serum AFP level of a fetus is high, there appears to be no extravasation or secretion into the amniotic fluid of A-T-affected fetuses, as occurs in open neural tube defects and Down syndrome.
A-T homozygotes show nonrandom chromosomal aberrations in lymphocytes, such as translocations and inversions, which preferentially involve chromosomal breakpoints at 14q11, 14q32, 7q35, 7p14, 2p11, and 22q11.43,117 These aberrations appeared to correlate generally with the regions of the T-cell receptor (TCR-α, β, and γ) and B-cell receptor (IGH, IGK, and IGL) gene complexes. Because these six sites contain the only gene complexes in the genome that are presently known to require site-specific gene rearrangement/recombination before expressing a mature protein, it was logical to examine V(D)J recombination mechanisms in A-T cells. As was noted above, signal and coding joint formation are both normal.96-98 When we examined the chromosomes of fibroblasts from eight A-T homozygotes, all with typical 7:14 translocations in their lymphocytes, the fibroblast aberrations were totally random.117,118 Hecht and Hecht studied almost 50,000 amniotic fluid cell metaphases; of 37 translocations in that non-A-T sample, none involved chromosomes 7 and 14.119 This is intriguing when one considers that, like lymphocytes and lymphoblasts, fibroblasts and amniotic cells express the radiosensitivity defect, suggesting that the radiosensitivity is intrinsic to A-T cells, whereas the chromosome aberrations are secondary to chromosome movement and telomere clustering in the nucleus.134 Heterozygotes show t(7;14) translocations in lymphocytes, but only in 1 to 2 percent of metaphases.119
In some patients, cell clones with the above breakpoints expand, 45,120 sometimes accounting for 100 percent of the lymphocytes that are karyotyped. Despite this, lymphocyte counts remain within the normal range for years thereafter. Some of these clones have been followed for 10 to 20 years by us and others.48,118 These clones tend to evolve, with subclones adding new rearrangements, such as inv(14;14)(q11;q32), i(8q), and 6q-, in addition to many other smaller clones. Eventually, most such patients develop T-PLL, previously referred to as T-CLL (T-cell chronic lymphoblastic leukemia).45,46 Affected sibs usually are not concordant for developing such clones, thus again implicating somatic influences superimposed on an A-T genotype.
These clonal expansions have allowed the breakpoint sites to be analyzed by molecular techniques. Three types of patients have been studied: (a) A-T patients with nonleukemic clones, (b) A-T patients with leukemic clones, and (c) non-A-T patients with similar cytogenetic translocations and T-PLL. Thanks to many years of perseverance by Taylor and coworkers124 in trying to pinpoint the breakpoints of these translocations or inversions, a fascinating story is now emerging that is quite similar to that of myc in Burkitt lymphoma. The A-T expanded clones always juxtapose one of the TCR genes, usually TCRα, with another family of genes located proximal to, but not actually within, the B-cell receptor-gene complexes. The most common and best-studied translocations are those involving 14q11 (TCRα) and a breakpoint cluster region 10 Mb proximal to the IGH locus at 14q32. Within 400 kb, at least 8 such breakpoints have been identified in A-T patients with and without leukemia, and in several non-A-T patients with T-PLL. This region centers on the TCL-1 (T-cell leukemia-1) gene, 123 the 1.3-kb transcript of which is preferentially expressed in immature (and leukemic) B and T cells. Circulating mature T cells do not express this gene. Leukemia cells without the t(14;14) or inv(14;14) clones typically do not express TCL-1.223
An occasional A-T patient has a large t(X;14)(q28;q11) clone, including at least two that have developed T-CLL/T-PLL and one without leukemia when last studied.45 The breakpoints at Xq28 cluster to within a few kilobases in a region of 70 kb proximal to the factor VIII gene. This region contains the genes c6.1A and c6.1B. (The latter gene is believed to be the crucial one in these translocations, because two of the breakpoints fall within the first exon of c6.1B, also known as MTCP1, “mature T-cell proliferation-1.”124-126) Most interesting, c6.1B has homology with TCL-1 (40 percent identity, 60 percent similarity) and is a mitochondrial protein.127 TCL-1 and MTCP-1 also share three-dimensional structure. TCL-1 prevents apoptosis and is p53-independent. Because TCRα/TCL-1 translocations do not by themselves cause leukemia, another factor must interact with the protein product or products that result from the translocations. Based on the recent finding that most non-AT patients with T-PLL have ATM mutations in one or both alleles, the ATM protein is a likely candidate for this role.84 Despite this, leukemia cells from an occasional A-T/T-PLL patient do not show abnormal TCL-1 expression, suggesting that yet other genes are involved in this pathway from clonal expansion to leukemia.
Inherited cytogenetic defects involving translocations or deletions at 11q22-23 have not been observed in A-T homozygotes, even though karyotypes of >500 patients have been examined worldwide. Many cytogenetic reports on children with suspected A-T return with the statement “insufficient metaphases for analysis.” This problem occurs because the necessary lymphocyte response to mitogens, such as phytohemagglutinin (PHA), is often weak or delayed in A-T patients, and when cell cultures are harvested routinely at 48 h, few cells are dividing. Harvest results can be improved by using a double-dose of mitogen and harvesting at 72 h or at several time points.
Telomeric fusions are observed frequently in A-T patients, which is a provocative finding considering the strong homology between ATM and the yeast Tel-1 mutant gene.71 Tumor cells and senescent cells of normal persons can also show such fusions.133 Pandita et al. 130 showed that although the telomeres of A-T cell lines are shorter than normal cells, telomerase activity was normal. Metcalfe et al. 135 demonstrated significant telomere shortening in A-T peripheral blood lymphocytes (PBLs). PBLs from 20 A-T patients showed an average loss of 95 ± 23 bp (base pairs) per year of age, compared to a loss of 35 ± 9 bp per year in normals. The preleukemic T-cell clones described above showed an even greater loss of 158 ± 9 bp per year and are especially prone to show telomeric fusions. Recently, as the biochemistry of telomere maintenance is being unravelled, 134 it appears that the Ku70/85 heterodimeric complex is physically bound to telomeres in yeast. Ku protects telomeres from nucleases and recombinases. Cells without Ku do not repair double-strand breaks or perform gene rearrangements for T or B cell maturation; Ku-deficient mutants display telomere shortening. In mammalian cells, Ku is the DNA-binding subunit of a large enzyme, DNA-dependent protein kinase (DNA-PKcs), which is a member of the large-molecular-weight protein kinase family that also includes ATM.71,72 The Ku complex interacts with the Rad50/Mre11/Xrs2/Brca1 complex for nonhomologous end-joining.134 Xrs2 (yeast) was recently identified as the human Nijmegen Breakage syndrome protein, nibrin201,202 (See discussion of “Related Syndromes” below.) The Rad50/Mre11/NBS1 complex, together with Ku and Brca1, is required for the telomerase pathway of end maintenance. The Rad50/Mre11/NBS1/Brca1 complex may be the exonuclease that provides the single-strand substrate required for telomerase activity. ATM interacts with the Rad50/Mre11/NBS1/Brca1 complex by phosphorylating both Brca1299 and nibrin.307 This would explain the overlap of symptoms between A-T and NBS (Fig. 29-5). Patients lacking hMre 11 protein have recently been described and closely resemble A-T patients in that they manifest progressive ataxia, t(7;14) translocations, and radiosensitivity. ATM protein expression is normal; nibrin and Rad50 expression are diminished.308
Overlapping A-T and NBS syndromes combine to form the ATFresno syndrome.
Accelerated telomeric shortening is probably a characteristic of all rapidly dividing A-T cells. It is of further interest that telomeric shortening is associated with senescence of CD282/CD81 T cells in AIDS patients and centenarians.136 In both situations, this may account for waning T-cell immunity. A similar mechanism might explain the abnormal development and function of the immune system in A-T patients. Thus, the precocious onset of cancers such as basal cell carcinoma, 2 leiomyosarcoma, 50 and T-PLL124 may reflect the basic propensity of their cells to accelerate telomere shortening and a waning immunity due not so much to poor V(D)J joining but to telomere shortening and senescence. This would also provide a p53-independent, radiation-independent pathway to cancer susceptibility in A-T patients.
When the ATM gene was isolated and sequenced, it was noted to have its strongest homology to the yeast tel1 gene, primarily through sharing a region of PI-3 kinase homology, and secondarily through sharing weak homology with rad3.137-141 (Reference141 contains a comprehensive analysis of homologies between kinase, rad3, RH3, and FRB domains.) Absence of tel1 results in telomere shortening. Rad3 is a fission yeast gene containing helicase motifs that is required for G2 arrest after DNA damage.142,143 Of the large family of genes sharing PI-3 kinase homology with ATM, only tel1, mec1 (another yeast gene), and mei41 (of Drosophila ) also share some rad3 homology. (The rad3 homology of tel1 is admittedly weak.) A growing body of evidence suggests that tel1, mec1, and ATM perform overlapping functions. Of the three, only mec1 is an essential gene. In yeast, mec1 (mitosis entry checkpoint) is required for regulation of the S/M and G2/M checkpoints, 144 the rate of ongoing S phase in response to damage, 145 and meiotic recombination.145,146 Cells with mutations in mec1 (also called ESR1 or SAD3) proceed directly to mitosis when DNA replication is inhibited with hydroxyurea and are unable to delay the onset of mitosis (G2/M) on induction of DNA damage.147 Rad53 is also regulated by MEC1 and Tel1.147 Although tel1 mutants are not radiosensitive and mec1 mutants are, tel1/mec1 double-mutants somehow synergize to increase the sensitivity to DNA damage from ionizing radiation and radiomimetic drugs.141 The human homologue of mec1, called ATR (AT-related Rad3-related) or FRP1 (FRAP-related protein), was recently cloned and maps to chromosome 3q22-q24.141,148 It plays a reciprocal role to ATM on synapsing chromosomes during meiotic recombination, 139 localizing to the nonsynapsed portion of the chromosomes and interacting with Rad51 and BRCA1. RPA and chk1 also colocalize with ATR on late pachynema chromosomes.271,272 RPA binds to single-stranded DNA, and probably facilitates formation of recombination intermediates.273,274 (ATR is also discussed under “Cancer Susceptibility” above.)
Fusion of fibroblasts from unrelated patients will often correct or “complement” their radiosensitivity, as measured by RDS.149-151,204 Five complementation groups have been defined (Groups A, C, D, E, and V1).151 The first four groups are phenotypically identical and can be distinguished from one another only by complementation studies. It was unclear whether these complementation groups represented several distinct A-T genes, perhaps forming part of a common enzymatic pathway or coding for parts of a common multimeric molecule, or, alternatively, whether the complementation groups represented intragenic mutations of a single gene. It was also possible that complementation was a nongenetic phenomenon. In 1988, we localized the gene for A-T Group A (ATA) to chromosome 11q22-23.152 In 1991, in a collaboration with Shiloh's lab, A-T Group C (ATC) was localized to the same region, also by linkage analysis.153 Between 1990 and 1994, 26 genes were shown to complement A-T fibroblasts; none were localized to chromosome 11q23.1.6,40,154,155 No convincing evidence for genetic heterogeneity was ever found in the linkage analysis studies despite such expectations. In 1995, when Savitsky et al. 72 identified part of a single gene (ATM), mutations were found for all four major complementation groups. Most interesting is that one homozygous mutation is present in both a Group C patient and a Group E patient, suggesting either that complementation groups in A-T are somewhat artifactual or that assigning patients to complementation groups is somewhat error-prone. To date, no laboratory has confirmed whether the cells from these two patients complement each other. Most likely this reflects that complementation group assignment by fusion of A-T fibroblasts is extremely tedious and that no laboratory has performed such studies since around 1990. Varying chromosomal ploidy between fused (4N) and nonfused (2N) cells also may have accounted for what appeared to be “complementation”.
Complementation of A-T cells by gene transfections was a commonly used approach to cloning the gene. Many genes complemented various facets of the radiophenotype. These complementing genes presumably interact in some way with the ATM gene, the protein, or the signal transduction pathway. Some may bypass the ATM block in A-T cells, and they might provide exciting therapeutic opportunities for replacement therapy in A-T patients.156 Despite the lack of success in cloning the A-T gene by complementation analyses, and the existing confusion about how intragenic mutations might complement, complementation may eventually provide a useful way of identifying functional domains in the ATM molecule.
A-T is transmitted as an autosomal recessive disease.1-6 The incidence of A-T has been estimated at 1 in 40,000 to 100,000 live births, while the gene frequency is believed to be as high as 3 percent of the general population.4,163 Recent studies of breast cancer in several large populations have provided convincing data in support of an ≈1 percent carrier frequency.55,258,282 All races are affected by A-T. Despite the A-T gene's affecting so many different and apparently unrelated systems, the disease is inherited in each family as a single autosomal recessive gene defect. It is unclear why, in an autosomal recessive disorder, so many of the parents of British, Italian, and American patients are unrelated. This is borne out by the recent finding that most A-T patients worldwide are compound heterozygotes; i.e., they have different paternal and maternal mutations.72,157 In the rare instances where two patients share a common mutation, their haplotypes usually differ, indicating independent origins for the mutation. The large size of the gene certainly provides a large target for new mutations. Recent studies suggest that gametogenesis is abnormal in ATM knockout mice16 and that mitotic and meiotic recombination is increased in A-T patients.76,158 Furthermore, as was discussed above, the ATM gene shares homology with mec1 (yeast) and mei41 ( Drosophila ), 40,141 and both are meiotic-recombination defective mutants.40,159 Whether this would affect heterozygous parents in A-T families sufficiently to influence the incidence of affected fetuses remains to be clarified. Recombination fractions in A-T families (i.e., in the parents) were normal across a 40 cM range of chromosome 11q22-23.161
Claims that “A-T is not always a recessive disorder”45,160 are misleading and belie the consortium experience of having localized the ATM gene to the proper 400-kb genomic segment using a mathematical model that assumed autosomal recessive inheritance of a single gene and included 176 families. Families that do not link to 11q22-23 should be considered to carry mutations in other genes and to likely represent other syndromes. New names will have to be given to such “AT-like” disorders (see “Related Syndromes,” below).
The rate of spontaneous mutations is unknown. Of the 176 consortium families, however, all but seven linked to chromosome 11q23.1.161,162 Follow-up studies have found mutations in the ATM gene in six of these families. (A seventh family may be due to uniparental disomy.) Thus, linkage analyses of 175 families did not detect spontaneous mutations. Using the ratio of 5:176 and a gene frequency estimate of 0.01, mutation rate estimates approximate 1.5−3×10−4 percent. This is rather high even for a large gene. Of course, if the ATM gene product really affects gametogenesis, 16,139,169 it may be inappropriate to apply standard genetic algorithms, which are based on the Hardy-Weinberg equilibrium, to the existing epidemiologic data.163,164 It may also be that some young patients succumb to malignancies before a diagnosis of A-T can be recognized, further skewing the data. Furthermore, recent studies of ATM mutations in A-T patients versus cancer patients from non-AT families suggest that the frequencies of truncating versus missense ATM mutations may differ in the general population (See “Cancer Susceptibility” above). 306
Very little research has been done on endocrine defects in A-T patients. This may change considering that ATM knockout mice have problems with both spermatogenesis and ovulation.16,39,169 Gonadal streaks, absent or hypoplastic ovaries, dysgerminomas, and undeveloped fallopian tubes have been observed at postmortem examination in both mice16 and human patients.2 Laboratory tests of pituitary function reveal no consistent abnormality.
In stark contrast to the earlier statement that “female hypogonadism with sexual infantilism is found consistently [in A-T patients],”2,5 most female patients followed by the author have normal menstrual cycles, and although menstruation sometimes starts late, cycles come at regular intervals. There is no other evidence as to whether these patients ovulate normally. Anecdotally, some long-lived female patients may have entered menopause prematurely. Others report very irregular cycles. Most male patients develop normal secondary sex characteristics. Some of these patients can have erections and even ejaculate. Studies of sperm haplotyping on semen from several A-T patients have documented that some actually produce sperm. None have fathered a child. One report of a putative female A-T patient having borne a child is clouded because this woman lived beyond 50 years, which is highly atypical for A-T, and a similarly affected sib demonstrated remarkable dexterity while already in her thirties (she worked in a knitware factory). In contract, female NBS patients manifest very severe endocrine defects, most showing little or no development of secondary sex characteristics and markedly elevated (prepubertal) follicle-stimulating hormone (FSH) and luteinizing hormone (LH) levels.283
Some patients develop insulin-resistant diabetes, usually in the late teens. This is characterized by hyperglycemia without glycosuria or ketosis.165,166 Other forms of diabetes, such as juvenile diabetes mellitus and late-onset diabetes, have been frequently observed among nonaffected members of A-T families. A genetic imprinting model has been considered, but this would not explain the pattern of diabetes in these families. Telomere silencing of subtelomeric genes, such as the insulin gene, might be an alternative hypothesis.134
Many of the chromosomal instability syndromes, such as A-T, Fanconi anemia, xeroderma pigmentosum, and Bloom syndrome, show progeroid features.167 Young A-T patients often have strands of gray hairs and develop keratoses; precocious basal cell carcinoma has been reported.2,42 Some of these findings may reflect either premature menarche or the accelerated shortening of telomeres described above135 (see “Chromosomal Instability”). However, thymic dystrophy and lymphoid depletion are also characteristic of aging and may be secondary to recombination defects during T-cell maturation rather than to telomeric shortening. Autoantibody formation is also found in both aging populations and A-T patients9,168,170 (see the discussion in reference170).
Postmortem examinations of older patients show progeric changes, such as neurofibrillary tangles in large neurons of the cerebral cortex, hippocampus, basal ganglia, and spinal cord, similar to those seen in Alzheimer disease.15 Lipofuscin granules have been found in many neurons, in satellite cells of the dorsal ganglia, and in Schwann cells. Further, Marinesco bodies seen in the pigmented neurons of the substantia nigra in A-T patients are considered signs of precocious aging.171
Among Costa Rican families with classical A-T, about 40 percent of patients have clubbing of the fingertips, a finding that is usually associated with poorly oxygenated blood supply. These A-T children do not have cardiac defects. Most, but not all, live in San José, which is 3700 feet above sea level, not high enough to aggravate most cardiac or pulmonary problems. The mutations in these families have all been identified, and the clubbing does not associate with a particular mutation (see “Patient Mutations,” below). It is possible that as part of their A-T syndrome these patients also have a pulmonary abnormality that compromises the oxygenation of their blood, such as microscopic arteriovenous fistulas or an anomalous bronchial tree.197 However, this is purely speculative; at this writing, there is no explanation for the clubbing in Costa Rican A-T patients.61
Many of the Costa Rican patients also have hypertrichosis (excessive body hair).61 This has been noted in other A-T patients as well.5 Considering the diverse endocrinologic abnormalities that have been described in A-T patients (and in ATM knockout mice), hypertrichosis could reflect a mild hormonal imbalance in some patients.
Swift et al. 32 observed a fourfold increase of ischemic heart disease among female A-T carriers. Thus, while heterozygotes are at a 3.2-fold increased mortality risk, only 44 percent and 35 percent of the deaths (men and women, respectively) observed by Swift et al. were attributable to cancer; 34 percent and 35 percent of the deaths (men and women, respectively) were attributable to heart disease.
The related Nijmegen Breakage syndrome (NBS)172 and the Berlin Breakage syndrome (BBS), 173 respectively assigned to complementation groups V1 and V2, do not show ataxia and do not link to chromosome 11q23.162,174,175 These syndromes found their way into the A-T literature because cells from these patients manifest the 7;14 translocations and radioresistant DNA synthesis that are typical of A-T cells. These patients are also cancer susceptible and immunodeficient. Telangiectasias are absent, and the serum AFP level is normal. NBS patients have birdlike facies, microcephaly, and mental retardation (A-T patients typically are not mentally retarded). BBS very closely resembles NBS, and when the NBS1 gene was cloned in 1998, both BBS and NBS patients had mutations in that gene. NBS1 is on chromosome 8q11.253 The NBS1 protein, nibrin, is absent from cell lysates of both NBS and BBS patients. NBS and BBS are now considered to be a single disorder. Because new evidence suggests that ATM phosphorylates nibrin, 307 in the Rad50/Mre11/nibrin complex, it would follow that the A-T and NBS phenotypes might overlap and that these genes might complement radioresistant synthesis.151 Why NBS cells complement the radioresistant DNA synthesis of BBS is again a mystery of complementation experiments. Only a handful of BBS patients have been described in the literature. They have most of the signs and symptoms of NBS, with the possible addition to the syndrome of syndactyly, anal atresia, and hypospadias. Most of the reported NBS and BBS families have been of eastern Europe origin and carry the 657del5 mutation.176-178,283,284
ATFresno (ATF) combines the classical A-T syndrome with NBS (Fig. 29-5).179 Whenever microcephaly and mental retardation are seen in an otherwise classical A-T patient, diagnosis of ATF should be suspected. However, because ATF families link to chromosome 11q23.1 and ATM mutations have been found in four ATF families, the clinical importance of this diagnostic distinction is presently unclear. Furthermore, the same ATM mutations found in two ATF families have also been observed in classic A-T patients. If a second modifier gene were involved, it would have to link to the 11q22-23 region as well.
Many other reports describe patients who do not meet all the diagnostic criteria for A-T discussed above.180-182,189,308,314 Many of these reports describe: (1) very young patients (when the A-T syndrome would not yet be fully expressed), (2) transient ataxias (some possibly infectious), (3) probable A-T patients without telangiectasias, 162,183,184 (4) patients with normal AFP levels, or (5) those with nearly normal immunologic parameters. Recent screening for ATM mutations in such “variant” families in the international consortium (families that were categorically excluded from the linkage analyses so as to avoid contaminating the positional cloning data) suggests that most of these were A-T. In several families with classically affected patients, prominent telangiectasias have been noted in members who do not have ataxia and who do not carry the two affected ATM haplotypes.162 Some of these persons are bona fide A-T heterozygotes.
Other families have been described with intermediate radiosensitivity, a parameter that is difficult to quantitate; nonetheless, in some hands, this must be considered a quantifiable result that will probably relate in some way to the sites of ATM mutations in those families or to mutations in other genes that link to the 11q22-23 region308,318 (see “Correlating Phenotypes with Genotypes,” below). Undoubtedly, other radiosensitive individuals exist whose symptoms partially overlap the A-T syndrome. It will be interesting to learn whether these patients have leaky ATM mutations or mutations in other genes that interact with the ATM protein.
The most difficult challenge in making a diagnosis of A-T involves very young patients. The most common misdiagnosis is cerebral palsy, especially when there is a spastic component to the child's movements. With time, however, a diagnosis of A-T becomes clear when the ataxia is notably progressive, eye movements demonstrate poor tracking, and speech becomes slurred. The absence of telangiectasia at this stage should not weigh against a diagnosis of A-T. Family history may be helpful if a prior child exists with similar signs and symptoms and the parents are related. Both factors should certainly raise suspicion about a hereditary disorder, and A-T is the most common hereditary early-onset progressive ataxia. The presence or absence of cancer in the family generally is not helpful, for it can be interpreted in many ways. Laboratory studies should include serum AFP, a cytogenetic search for t(7;14) translocations, in vitro radiosensitivity (see below), and an immunologic evaluation. Recent evidence suggests that ATM protein levels in lysates of A-T cells are be very low or absent in most classical patients; these can be measured semi-quantitatively by Western blotting. In those bona fide A-T patients with ATM protein (∼20%), the function is assumed to be compromised. This need not be limited to just the p53 kinase function; defects in alternative splicing, DNA binding, or tissue specificity could also have similar phenotypic effects.
Even if some of the above tests are not informative, a diagnosis of A-T may still be valid for these following reasons: (a) The AFP remains normal throughout life in 5 percent of patients. As was discussed above, the serum AFP is occasionally elevated in normal children under 2 years of age, and thus is not a reliable test until after that age. (b) A cytogenetic search for t(7;14) translocations or clones is often unsuccessful in A-T patients because a poor mitogenic response makes it difficult to find enough good-quality metaphases for analysis (see “Chromosomal Instability,” above). Even if sufficient metaphases are found, the translocations are sometimes missed. Radiation-induced and bleomycin-induced breakage studies may be helpful, but they seldom contribute to making the diagnosis because of overlap between normal and A-T ranges.89 (c) The immunologic evaluation is normal in some A-T patients. Whether it becomes progressively more abnormal in older A-T patients is debatable (see “Immunodeficiency,” above). The response to allogeneic cells, the mixed lymphocyte response, is quite abnormal in some patients; however, this is a very laborious and costly test that is hard to quantitate without extensive controls and is therefore difficult to justify for strictly clinical purposes.
A MRI (magnetic resonance imaging) of the cerebellum will usually reveal marked dystrophy in children over 4 years of age (Fig. 29-6). Newer techniques for imaging the cerebellum are also being evaluated, such as functional MRI and PET (positron emission tomography) scanning;231,232 however, both depend heavily on patient cooperation and may not be applicable to very young children. Furthermore, PET scanning uses radioactive tracers, and although the exposure doses are very small, they could theoretically contribute to cancer risk, especially in the bladder where the radioisotope accumulates rapidly during the procedure. When risk-benefit ratios are considered for procedures using ionizing radiation, difficult judgments must be made.
Magnetic resonance imaging of a 6-year-old A-T patient showing markedly reduced size of the cerebellar shadow.
The most dangerous diagnostic situation for a young A-T patient occurs when cancer is the presenting symptom. Fortunately, this does not occur very often. Anecdotally, one child had a cerebellar astrocytoma removed at 27 months of age, but his unsteady gait actually worsened postoperatively. His clinicians were quite concerned and confused by the persistent ataxia until several years later, when the patient's younger sister began to stagger as well and a diagnosis of A-T was made on both children. Because the astrocytoma was totally resectable, no consideration was given to further therapy with chemotherapeutic agents or radiation. The patient died more than 20 years later without any sequelae of the cancer or surgery. Other children have not been so fortunate, 17-19 presenting with a malignancy and receiving conventional doses of irradiation because it was not realized that they were suffering from A-T, only to suffer iatrogenic deaths. The late Dr. Boder claimed that she could make a diagnosis of A-T in any child under 2 years of age. While this is a challenging claim, it is certainly true that most young A-T patients do have at least some suspicious neurologic findings at a very early age. On questioning, mothers sometimes volunteer that they noted head tilting or swaying in these infants (Fig. 29-7). Thus, it is prudent for pediatric oncologists and radiation oncologists to rule out the diagnosis of A-T before treating any young child with cancer, either by obtaining a complete history and performing a careful neurologic examination with this in mind, or by obtaining a neurologic consultation as part of the workup.
Head-tilting in a 6-month-old infant with A-T. Staggering was not noted until she began to walk.
While the presence of hypersensitivity to ionizing radiation is a laboratory hallmark of the disease, clinical testing for this has not been readily available, primarily because most radiosensitivity assays use fibroblasts, and establishing fibroblasts in A-T patients is painful and labor intensive. With this in mind, Huo et al.111 laboratory established the CSA, a clonogenic assay that evaluates the colony survival fraction of LCLs from patients after the cells have received 1 Gy of ionizing irradiation.111 From a single 10-ml heparinized blood sample (that should be shipped without refrigeration), cells are transformed with Epstein-Barr virus. Once a stable cell line is established, the cells are plated in two cell concentrations on 96-well tissue culture trays that are irradiated (or not irradiated) and returned to an incubator for 10 days, at which point the number of wells containing colonies larger than 32 cells is scored and compared to the colony survival fractions of normal cells. Unlike other colony survival assays, the CSA conditions were selected so that heterozygotes would score as normals, which allows for more reliable detection of A-T homozygotes (Fig. 29-8). Recently, two referred patients with normal CSA results on repeated testing were subsequently found to have the typical (GAA)n expansions of Friedreich ataxia on chromosome 9q13. Although the differential diagnosis between Friedreich ataxia (FRDA) and A-T is usually not difficult—FRDA is a later-onset ataxia (usually around puberty) and most FRDA patients have hypertrophic cardiomyopathy (by ECG testing), whereas A-T patients generally do not have cardiac problems—this experience served to underscore the value of using radiosensitivity to confirm a suspected early diagnosis of A-T. FRDA patients have normal CSA results.111,185 Patients from all complementation groups, including NBS and BBS, have the same markedly reduced CSA levels. Human Mre11 deficiency patients308 are also radiosensitive but have not yet been tested by CSA
Colony survival assay (CSA) measures radiosensitivity of LCLs from patients with A-T, A-T heterozygotes, normals, and a Bolivian family with three affected children, following 1 Gy of irradiation. Also included are results from patients with NBS (V1), BBS (V2), and ATF (V1*).
Abnormal facies other than the slowly developing smile, or masklike expression of many A-T patients, should raise suspicion about other diagnoses. Severe mental retardation and inability to speak at an appropriate age are also uncharacteristic of A-T. Mental retardation is seen more commonly in lower socioeconomic-level families and countries, perhaps because they lack the resources needed to keep A-T patients in the mainstream of family and community life, whereby they must learn to respond to various personal challenges. The absence of oculomotor apraxia by 5 years of age is also strong evidence against the diagnosis of A-T.
Ataxia is common to a variety of other hereditary disorders:186 (a) as a major feature with progressive ataxia—hMre11 deficiency308 β-lipoprotein abnormalities selective vitamin E deficiency, hexosaminidase deficiency (GM2), and cholesterolosis; (b) as a major feature with intermittent ataxia—urea-cycle defects, maple syrup urine disease, isovaleric acidosis, 2-hydroxyglutaric aciduria, Hartnup disease, pyruvate dysmetabolism, and mitochondrial disease; and (c) as a minor feature of Niemann-Pick syndrome, metachromatic leukodystrophy, multiple sulfatase deficiency, late-onset globoid cell leukodystrophy, adrenoleukodystrophy, sialidosis type 1, and ceroid lipofuscinosis. The latter can be diagnosed only by biopsy of the conjunctiva or brain. Most of the other listed disorders will show abnormalities in urinary amino acids, lysosomal hydrolases, or very long chain fatty acids. Retinitis pigmentosa, deafness, polyneuropathy, and ataxia characterize Refsum disease.187 Non-hereditary ataxia may result from an acute infection or from a posterior fossa tumor. (For further information, see reference5 .)
Determining whether a new patient's ataxia has been inherited in a dominant or a recessive manner can aid in distinguishing A-T from olivopontocerebellar atrophy and any of the spinocerebellar ataxias, all of which are dominant disorders. The familial pattern for age at onset of the ataxia is also helpful, because few other familial ataxias present in early childhood as A-T does. While an occasional case of early-onset FRDA might be mistaken for A-T on this basis, neurologic examination will reveal spinal cord ataxia with a positive Romberg sign, and, in the laboratory, homozygosity for a (GAA)n expansion in the first intron of the FRDA gene is easily diagnosed188,316 ; FRDA cells are also not radiosensitive.111
Determining whether two mutations exist within the ATM gene of a child suspected of having A-T is the most definitive way of establishing a diagnosis. At this writing such an approach is just becoming feasible (see “Patient Mutations,” below). In families of certain ethnic backgrounds, rapid DNA assays can be performed for mutations that are common in that population. By first haplotyping the DNA of a suspected A-T patient in the chromosome 11q22-23 region, previously described haplotypes carrying mutated ATM genes can be identified. However, unless the patient is homozygous for a mutation—which is very unlikely unless the parents are consanguineous—a second mutation must still be sought. This requires a great deal of effort either by mRNA/cDNA/RT-PCR-based screening assays or by a systematic genomic search of the 66 exons of the ATM gene. Even this approach is not 100 percent effective in finding all mutations, because some lie deep within introns and others require analysis of both genomic DNA and mRNA. Eventually, we hope to determine the mutations and affected haplotypes for most A-T families. This database will expedite both the diagnosis of A-T and prenatal diagnosis.
Aicardi et al. 189 described a group of 14 patients with a late-onset progressive ataxia, choreoathetosis, and oculomotor apraxia without frequent infections or telangiectasia. The AFP was normal and a search for t(7;14) translocations was negative. Aicardi suggested that these children suffered from “an unusual type of spinocerebellar degeneration,” probably not A-T. However, it would be informative to determine whether any of those patients are radiosensitive, for example, by CSA.
With the fine mapping of the ATM gene, a set of highly informative genetic markers was developed that now allows accurate haplotyping within families, with basically 100 percent reliability of the fetus either being affected or not being affected, i.e., less than 1 percent recombination between ATM and the markers used. The finding of only a single A-T gene for all complementation groups also simplifies this diagnostic approach. This is in contrast to earlier attempts to perform prenatal diagnoses by trying to quantitate spontaneous chromosome breakage89,109-193 (see the discussion in reference150), assessing radiation-induced chromosomal damage of amniocytes or fetal fibroblasts, or performing RDS, all of which were misleading at one time or another (anonymous oral communications). One hopes that these approaches have by now been abandoned.
Prenatal diagnosis by haplotyping relies on a prior affected child to (a) establish a firm diagnosis of A-T and (b) identify the two affected chromosome 11q23.1 segments (i.e., haplotypes) carrying the ATM gene.192 Figure 29-9 illustrates haplotyping and how it was possible to determine that the first cousin of an affected patient was not affected. A definitive diagnosis was possible only because a prior affected cousin existed in a consanguineous family. It is of paramount importance that the diagnosis of A-T in the prior affected member be confirmed before attempting haplotyping for prenatal diagnosis. The markers we use today are all within 1 percent recombination of the ATM gene: D11S1817, 194 D11S1819, 194 NS22, 285 D11S2179, 195 and D11S1818.194 Two of these markers are within the ATM gene itself, thereby circumventing the need for reporting the risk of recombination separating the testing markers from the actual ATM gene. Most of this testing can be performed before conception, i.e., preconceptional testing. Once a DNA sample is available from the fetus, either from growing amniocytes or from a chorionic villous biopsy, the entire haplotyping can be completed within a week. With new “molecular beacons” this will become even more streamlined.298 Because abortion guidelines vary with the country or state of residence of the mother, we ask for the referring laboratory's deadline for reporting the results of prenatal diagnosis to the family, rather than for the due date or the date of last menses. In keeping with modern guidelines for genetic testing, the results are conveyed to the family by a genetic counselor.
Prenatal testing to determine whether a fetus (3) is affected. By history, two brothers had married two sisters who were their first cousins. One couple (4 and 5) had an affected child (6), prompting the second couple (1 and 2) to seek prenatal testing. However, testing before conception would have identified the mother (2) as a noncarrier, thereby circumventing any further testing on the fetus. Haplotype [A] carries the affected ATM gene. The genetic markers that are currently being used for prenatal testing are given in the text.
No effective therapy exists for halting the progression of the ataxia.2,5,196 Clinical trials are under way to test the efficacy of myoinositol, N-acetylcysteine, and L-dopa on general symptoms. To date, preliminary data have been disappointing. (For a review of other AT-related medications, see reference196 .) Vitamin E has been prescribed by some A-T specialists for years, based on anecedotal information from Dr. Elena Boder.15 Recent studies suggesting that A-T cells may be in a constant state of increased oxidative stress make it likely that most antioxidants or free-radical scanvengers might counteract some of the progressive neurological deterioration of A-T patients.38,317,325,326,327 Thus, vitamin E continues to be recommended; α-lipoic acid and coenzyme Q10 may also slow the deterioration. Folic acid may formation further help to minimize chromosomal fragility and the formation of double-strand DNA breaks. All of the above dietary supplements are available without a prescription.
Areas of great concern to the health of A-T patients are pulmonary infections and malignancy. Pulmonary infections usually are due to the normal spectrum of microbes and are treatable by conventional approaches. Opportunistic infections do not occur in A-T patients as they do in patients with other immunodeficiency disorders (with the possible exception of mycobacterium). Malignancies must be treated with great care to avoid conventional doses of radiation therapy or radiomimetic agents. If possible, neurotoxic chemotherapeutic agents should be avoided.
Not all A-T patients manifest frequent pulmonary or sinus infections. Those with chronic bronchiectasis are best treated in the same way as patients with cystic fibrosis: routine chest percussion, postural drainage, and generally aggressive pulmonary hygiene. Periodic pulmonary function studies may assist in monitoring infection-prone patients. In older patients, pulmonary infections are the major cause of failing health and death. Increasing bulbar dysfunction may predispose to aspiration pneumonia. In addition to appropriate antibiotics, intravenous γ-globulin every 3 to 4 weeks may reduce the frequency of infections in infection-prone patients. There is some indication that the lungs of A-T patients may not be anatomically normal. Pump et al. 197 made a latexlike impression of one lung (using a substance called Vultex moulage [General Latex and Chemicals Ltd., Verdun, Quebec]) from a single A-T patient. They found bronchiectatic changes in many parts of the lung, with saccular dilatations throughout the bronchi.
Perhaps the most effective impact on care that the physician can make is to strongly encourage the parents of young A-T patients to institute an aggressive and engaging physical exercise program aimed at enhancing lung function, preventing contractures, and avoiding positional kyphoscoliosis. Almost all patients who have been denied such care develop severe contractures of the feet and hands. These become apparent in the late teens. An annual assessment by a physical therapist allows this care to be customized.
Speech therapy is also effective, not in arresting the progression of the dysarthria, but in minimizing the frustration felt by the patients when they cannot be understood by peers. Increased social interaction generally improves speech clarity.
Some of the neurologic symptoms, such as ataxia, drooling, and tremors, can be partially relieved by various agents.196 Buspirone, a serotonergic 5-hydroxytryptophan agonist, is active in some types of cerebellar ataxia.253 Amantadine improves balance and coordination and minimizes drooling in some patients. Postural tremors may be reduced by baclofen, a GABA inhibitor, or propranolol and other β blockers, while cerebellar tremors and myoclonus may respond to low doses of clonazepam or valproic acid. However, these agents sometimes increase the ataxia, drowsiness, or depression. Methyl scopolamine or propantheline hydrochloride are sometimes effective in reducing drooling. Ligating the salivary ducts can also alleviate drooling. This may also reduce the risk of aspiration pneumonia.
When radiation therapy is planned for treating a malignancy in an A-T patient, doses should be reduced by approximately 30 percent. Some chemotherapeutic agents, especially alkylating agents, probably should also be used in reduced doses. It has been suggested that topoisomerase inhibitors should be avoided. Unfortunately, there is only anecdotal literature on the important and recurring issue of oncologic treatment of A-T patients.
Occasionally, the possibility of bone marrow transplantation arises, usually because a young A-T patient has developed leukemia and has an HLA-compatible sib to serve as a potential donor. Despite several such attempts over the past 20 years, the author is unaware of a single transplant with convincing documentation of long-term engraftment. This could be for several reasons, the most compelling of which is the difficulty of establishing a safe but effective regimen for delivering the marrow-ablating irradiation or chemotherapy for the reasons given above. In general, hyperfractionation of radiation doses would seem prudent under such circumstances if this need were to arise. There is, of course, only a remote possibility that bone marrow transplantation would alter the cerebellar degeneration.286,287 Hematopoietic stem cell transplantation might reduce the need for complete ablation but probably would also reduce the chances of a full immunologic engraftment. Neural stem cell engraftment may eventually become an important therapeutic alternative.
Many A-T patients have been immunized inadvertently for smallpox, polio, and varicella, with no apparent sequelae. Nevertheless, natural varicella infections are often quite severe. Thus, contrary to the general warning that patients with immunodeficiencies not be given live vaccines, varicella immunization is advisable for patients whose immune status is satisfactory.
Most A-T patients in the United States live well beyond 20 years. Many are now in their thirties. This is a major change from just a few years ago when it was unusual for these patients to live beyond their teenage years. Unfortunately, this is still true in many countries, for reasons that are unknown; however, the improved survival in the United States may be related to better nutrition, better diagnostics, better treatment of pulmonary infections and malignancies, and more aggressive physical therapy. There is hope among A-T investigators that the young children being diagnosed today will benefit from some currently undiscovered therapy before their neurologic status becomes irreversible.