Abstract
Background
Ataxia telangiectasia (A-T) is a DNA repair disorder with cancer predisposition.
Objective
We sought to characterize the prevalence and outcomes of hematologic and solid cancers and treatment-associated toxicities in individuals with A-T.
Methods
Data were retrospectively analyzed from the Johns Hopkins Ataxia Telangiectasia Clinical Center cohort. Cumulative incidence and standardized incidence ratios of cancer, survival probability after cancer diagnosis, and standardized mortality ratios were calculated. Cox regression estimated risk of death on the basis of chemotherapy (standard vs reduced) dosing, and multivariable logistic regression evaluated cancer risk associations with ataxia telangiectasia mutated (ATM) exons and variants.
Results
Eighty-four (16.5%) of 508 individuals were diagnosed with a primary cancer, of whom 62 (74%) were hematologic in origin and 22 (26%) were solid-organ cancers. The cumulative incidence of cancer was 29% by age 35 years. Non–Hodgkin lymphoma occurred most frequently (n = 39), whereas solid cancers disproportionately affected those 18 years and older (n = 22). The standardized mortality ratio was 24.6 (95% CI, 21.1–28.4) overall and 232.9 (95% CI,178.1–299.2) among individuals with cancer. Risk of death was higher when treated with standard/unknown versus modified chemotherapy (hazard ratio, 2.2; 95% CI, 1.1–4.4; P = .024). Chemotherapy- associated toxicities developed in 58% of individuals, predominantly neurologic (n = 14) and gastrointestinal (n = 10) systems. Three exons were enriched for cancer-associated variants.
Conclusions
Individuals with A-T experience a wide array of blood and solid-organ malignancies, high mortality rates, and treatment-related toxicities, highlighting need for targeted therapies to mitigate toxicity and optimize survival.
Keywords: 1. Ataxia telangiectasia, 2. ATM, 3. pathogenic, 4. mutation, 5. non–Hodgkin lymphoma, 6. solid-organ cancers, 7. chemotherapy, 8. toxicity, 9. mortality
Graphical Abstract
Capsule summary: Individuals with ataxia telangiectasia have higher than previously reported prevalence of solid-organ cancers and incur therapy-related toxicities independent of therapy attenuation, with neurological and gastrointestinal complications being most prevalent.
Ataxia telangiectasia (A-T) is a DNA repair disorder secondary to biallelic mutations in the ataxia telangiectasia mutated (ATM) gene,1 which encodes a Ser/Thr protein kinase that orchestrates double-strand DNA break repair and cellular responses to other damage such as oxidative stress.2 Individuals with defective ATM protein develop a multisystem syndrome with clinical features including progressive cerebellar degeneration, ocular and mucocutaneous telangiectasias, immunodeficiency, recurrent sinopulmonary infections, radiation sensitivity, and cancer predisposition.3 Cancer-related morbidity and mortality remain high in individuals with A-T because of increased organ toxicities to antineoplastic agents in the setting of A-T–associated comorbidities.4–11
Decreased genome surveillance in the setting of ATM deficiency in individuals with A-T results in a 10% to 25% prevalence of cancer, with non–Hodgkin lymphoma (NHL) being the most common malignancy.4,6,7,10,12 In addition, ATM carriers have a severalfold higher predisposition to breast, ovarian,13–15 melanocytic,16,17 pancreatic,18 prostate,17,19 gastric,17,20 and colorectal carcinomas15,21 compared with the general population. Malignancy is a major cause of mortality among individuals with A-T. Cancer therapy in individuals with A-T is challenging because of pre-existing multiorgan dysfunction compounded with inherent hypersensitivity to chemotherapy and/or ionizing radiation, resulting in poor outcomes. Current strategies to circumvent this involve chemotherapy dose reductions and minimizing exposure to radiation therapy and radiomimetic and neurotoxic agents.3,7,9,11 Both modified and unmodified regimens have shown variable success in achieving sustainable cancer remissions.3,7,9–12 However, clinical trials seeking to balance therapy-associated toxicities while achieving anticancer effect/cure have been precluded by the rarity of A-T.
The high prevalence of cancer in individuals with A-T, coupled with the challenges associated with tolerating anticancer therapies, underscores the necessity of gathering data from large cohorts to develop treatment guidelines, prospective protocols for optimal curative therapy, and practical surveillance strategies. To this end, we performed a retrospective analysis of a longitudinal cohort of 508 individuals with A-T to characterize cancer incidence, patterns of treatment selection, survival, toxicity outcomes, and ATM genotype–cancer associations.
METHODS
Participants
Individuals with A-T were referred by physicians or self-referred to the Johns Hopkins Ataxia Telangiectasia Clinical Center (ATCC) (Baltimore, Md) because of previous A-T diagnosis, abnormal laboratory and/or physical examination results, abnormal T-cell receptor excision circles on newborn screening, or family history of A-T. Medical history, physical examinations, laboratory indices, and imaging were obtained for each patient at the initial and subsequent visits under the Natural History of Ataxia Telangiectasia Institutional Review Board–approved protocol (NA_00014314). The diagnosis of A-T was based on the presence of biallelic ATM variants, absent ATM expression via immunoblot, and/or fulfillment of established A-T clinical criteria including ataxia, telangiectasias, and/or elevated alpha-fetoprotein levels.3
To ascertain cancer-associated morbidity and mortality in the A-T cohort, we performed a retrospective analysis of data collected by the ATCC from 1996 to 2021. Data reviewed were either documented at annual visits at ATCC or reported to ATCC by patients, family members, and specialty care providers (ie, oncologists and immunologists). Reviewed data included demographic characteristics, medical history, ATM genetic testing, laboratory test results, personal or family history of malignancy, type of cancer diagnosed with associated laboratory and pathology reports relevant for cancer diagnoses, chemotherapy regimens used, therapy modifications, organ toxicities, survival, and secondary cancer diagnoses. Data collected regarding treatment regimens, therapy modifications (defined as any reduction from the dose stated in the treatment protocol, omission of therapy cycles, doses, or agents, and/or switching of chemotherapy agents), and organ toxicities were determined and reported by the treating oncologists at 56 North American institutions to ATCC. As a result, a standardized toxicity grading system was not feasible. Specific values of dose reductions were not documented for every patient with A-T who received dose-reduced chemotherapy.
Statistical Analysis
Demographic characteristics were tabulated for A-T survivors with and without cancers. The cumulative prevalence of cancer was estimated for the overall cohort and by cancer type, treating death as a competing risk. Standardized incidence ratios (SIRs) of cancers were calculated using age-, sex-, and calendar year–specific US cancer incidence rates from the Surveillance, Epidemiology, and End Results program22 to determine the expected number of events. Follow- up time started at birth and ended at cancer diagnosis or date of death.
Standardized mortality ratios (SMRs) were estimated for all patients with A-T and for patients with A-T who developed cancers and compared with the age-, sex-, and calendar year–specific rates in the US population. Overall survival analysis considered follow-up time to begin at birth, whereas survival analyses for patients with A-T and cancer began at cancer diagnosis. Follow-up time ended at the date of death or last follow-up. Survival probability after cancer diagnosis in patients with A-T was estimated, stratified by the chemotherapy reduction status. Among those who received chemotherapy, Cox regression was used to estimate the hazard ratio (HR) comparing standard dose/unknown dose reduction versus reduced-dose chemotherapy on mortality, adjusting for age-, sex-, and race-specific cancer diagnosis.
When performed, molecular diagnostic testing was done in either a clinical or a research setting via diverse methods including, but not limited to, Sanger and short-read Illumina-based sequencing. Some did not achieve full molecular diagnosis (the detection of biallelic pathogenic/likely pathogenic variants in trans) because of limitations of technology at the time of testing. To our ability, all ATM variants were harmonized to the most up-to-date reference sequences and Human Genome Variation Society nomenclature, annotated and classified according to distinct variant types, predicted/observed variant effects, and pathogenicity classes. For a few genotypes, the original notation was too difficult to decipher or map to a known variant. Pathogenicity was assigned to the best of our ability using ATM-specific ClinGen guidelines for application of the American College of Medical Genetics and Genomics classification criteria23 while considering legacy classifications and both published and unpublished functional data.
The frequencies of homozygotes and compound heterozygotes between cancer and noncancer groups were compared using the Fisher exact test. To assess the burden of predicted loss-of-function (LOF) variants (frameshift, start loss, stop loss, nonsense, and large insertions/deletions) on cancer, ATM variants were binned into 3 categories (0 = no predicted LOF alleles; 1 = 1 predicted LOF allele; and 2 = 2 predicted LOF alleles) and analyzed for associations to overall cancer, hematologic malignancy (HM), and solid tumor using logistic regression, adjusting for sex and race. In exon-level analysis, all variants were assigned to each of the 63 exons (RefSeq ID: NM_000051.4) according to the following rules: (1) point variants assigned to their corresponding exon location, (2) frameshift and large indels assigned to the first exon predicted to be interrupted, and (3) intronic changes assigned to their nearest exons. Each exon was coded on the basis of associated variants and tested for its association with cancer risk by multivariable Cox and logistic regression. In single-variant analysis, individual effect of 20 ATM variants found in at least 5 individuals with A-T on cancer risk was examined by multivariable logistic regression. A P value less than .05 from Cox or logistic regression–based likelihood ratio test (LRT) was considered statistically significant.
RESULTS
Cancer spectrum and incidence in individuals with A-T
Of the 508 study participants, 261 (51.4%) were male and most were White (71.9%; Table I). Eighty-seven percent of patients were referred from US treatment centers. Eighty-four (16.5%) individuals were diagnosed with a primary cancer (Table I) and 5 (1%) developed at least 1 subsequent cancer. Among the primary cancers, 62 (74%) were of hematologic origin and 22 (26%) were solid-organ cancers (solid tumor, Fig 1, A; Table I). Hematologic cancers included NHL (n = 39), T-cell acute lymphoid leukemia (T-ALL; n = 7), T-cell prolymphocytic leukemia (T- PLL; n = 6), Hodgkin lymphoma (HL; n = 7), mixed-phenotype acute leukemia (MPAL; n = 1), early T-cell precursor acute lymphoid leukemia (ETP-ALL; n = 1), and B-cell acute lymphoid leukemia (B-ALL; n = 1) (Fig 1, A. Solid-organ cancers included skin (n = 8), gastrointestinal (n = 6), breast (n = 3), and single cases of parotid, liver, pancreatic, lung, and ovarian cancer (Fig 1, A; see also Table E1). The most common malignancy occurring in patients younger than 18 years was NHL (n = 31 [57%]) (Fig 1, B), whereas solid cancers (n = 18) were most common in those 18 years or older (n = 18 [55%]) (Fig 1, B).
Table I.
Baseline characteristics of the ATCC cohort
| Characteristics | No cancer (n = 424) | Cancer (n = 84) | Entire cohort (N = 508) |
|---|---|---|---|
| Age (y) at death or censoring | |||
| 0–4 | 56 (13.2) | 0 (0.0) | 56 (11.0) |
| 5–9 | 94 (22.2) | 7 (8.3) | 101 (19.9) |
| 10–14 | 81 (19.1) | 17 (20.2) | 98 (19.3) |
| 15–19 | 65 (15.3) | 18 (21.4) | 83 (16.3) |
| 20–24 | 53 (12.5) | 14 (16.7) | 67 (13.2) |
| 25–29 | 41 (9.7) | 8 (9.5) | 49 (9.6) |
| ≥30 | 34 (8.0) | 20 (23.8) | 54 (10.6) |
| Sex | |||
| Male | 217 (51.2) | 44 (52.4) | 261 (51.4) |
| Female | 207 (48.8) | 40 (47.6) | 247 (48.6) |
| Race/ethnicity | |||
| White | 296 (69.8) | 69 (82.1) | 365 (71.9) |
| Black | 42 (9.9) | 1 (1.2) | 43 (8.5) |
| Hispanic/Latino | 57 (13.4) | 5 (6.0) | 62 (12.2) |
| Asian | 22 (5.2) | 5 (0.0) | 27 (5.3) |
| Other | 5 (1.2) | 2 (8.3) | 7 (1.4) |
| Unknown | 2 (0.5) | 2 (2.4) | 4 (0.8) |
| Year of birth | |||
| 1964–1969 | 5 (1.2) | 3 (3.6) | 8 (1.6) |
| 1970–1979 | 21 (5.0) | 3 (3.6) | 24 (4.7) |
| 1980–1989 | 74 (17.5) | 21 (25.0) | 95 (18.7) |
| 1990–1999 | 120 (28.3) | 27 (32.1) | 147 (28.9) |
| 2000–2009 | 119 (28.1) | 26 (31.0) | 145 (28.5) |
| 2010–2019 | 79 (18.6) | 4 (4.8) | 83 (16.3) |
| 2020–2022 | 6 (1.4) | 0 (0.0) | 6 (1.2) |
| Death | 122 (28.8) | 61 (72.6) | 183 (36.0) |
| Age (y) at death, mean ± SD | 22.2 ± 7.5 | 20.5 ± 8.7 | 21.6 ± 8.0 |
| Cancer type | |||
| Hodgkin lymphoma | NA | 7 (8.3) | |
| Leukemia | NA | 16 (19.0) | |
| NHL | NA | 39 (46.4) | |
| Solid tumor | NA | 22 (26.2) |
Data are presented as n (%), unless otherwise indicated.
NA, Not applicable.
Fig 1.
Cancer spectrum in A-T changes with age. A, Primary malignancies in individuals with A-T divided into 4 subgroups of NHL, leukemia, solid tumors, and Hodgkin lymphoma. B, Individuals with A-T and NHL, leukemia, Hodgkin lymphoma, or solid tumors stratified on the basis of those younger than 18 years vs those 18 years or older. ALCL, Anaplastic large-cell lymphoma; B-ALL, Bcell acute lymphoid leukemia; B-LBL, B-cell lymphoblastic lymphoma; DLBCL, diffuse large B-cell lymphoma; ETP-ALL, early T-cell precursor acute lymphoid leukemia; LyPlasm Ly, lymphoplasmacytic lymphoma; MALT, mucosa-associated lymphoid tissue lymphoma; MPAL, mixed-phenotype acute leukemia; T-ALL, T-cell acute lymphoid leukemia; T-LBL, T-cell lymphoblastic lymphoma.
The median age at cancer diagnosis was 14.4 (interquartile range [IQR] = 10.3–22.8) years and differed by specific cancer type (Table II). Specifically, the median age of diagnosis was 12.4 (IQR, 7.3–16.7) years for NHL, 12.6 (IQR, 11.8–14.6) years for Hodgkin lymphoma, 15.8 (IQR, 9.6–28.5) years for leukemia, and 26.3 (IQR, 18.7–29.0) years for solid tumors. The cumulative incidence of a primary cancer was 4.7% (95% CI, 2.7–6.7) at age 10 years, 17.7% (95% CI, 13.4–21.9) at age 20 years, and 26.2% (95% CI, 20.7–31.8) at age 30 years (Fig 2, A). The SIR of primary cancer was 47.1 (95% CI, 37.5–58.3) for all cancers, 355.9 (95% CI, 253.0–486.5) for NHL, 67.4 (95% CI, 27.0–138.9) for Hodgkin lymphoma, 45.6 (95% CI, 26.1–74.1) for leukemia, and 18.0 (95% CI, 11.3–27.3) for solid tumors (Table II). The mean time to diagnosis of a subsequent cancer from time of primary cancer diagnosis was 6.8 ± 5.1 years. Four patients were diagnosed with cancer before being diagnosed with A-T. Immunophenotype markers did not significantly differ between the 4 cancer subtypes among individuals with A-T diagnosed with cancer.
Table II.
SIR and SMR
| Observed | Expected | SIR (95% CI) | Age (y) at cancer diagnosis, median (IQR) | |
|---|---|---|---|---|
| All cancers | 84 | 1.8 | 47.1 (37.5–58.3) | 14.4 (10.3–22.8) |
| Non–Hodgkin lymphoma | 39 | 0.1 | 355.9 (253.0–486.5) | 12.4 (7.3–16.7) |
| Hodgkin lymphoma | 7 | 0.1 | 67.4 (27.0–138.9) | 12.6 (11.8–14.6) |
| Leukemia | 16 | 0.4 | 45.6 (26.1–74.1) | 15.8 (9.6–28.5) |
| Solid tumor | 22 | 1.2 | 18.0 (11.3–27.3) | 26.3 (18.7–29.0) |
| Observed | Expected | SMR (95% CI) | ||
| All patients with A-T | 183 | 7.4 | 24.6 (21.1–28.4) | |
| All cancers | 61 | 0.3 | 232.9 (178.1–299.2) | |
| Hodgkin lymphoma | 7 | 0 | 3829.8 (1534.3–7891.3) | |
| Non–Hodgkin lymphoma | 29 | 0.1 | 273.2 (182.9–392.3) | |
| Leukemia | 11 | 0.1 | 213.0 (106.2–381.1) | |
| Solid tumor | 14 | 0.1 | 136.9 (74.8–229.7) |
Fig 2.
Cancer incidence and therapy-related outcomes in A-T. A, Cumulative incidence of cancers in A-T. B, Survival probability for individuals over 15 years after receiving standard vs modified chemotherapy.
Table III.
Exon-outcome associations detected by Cox and logistic regression models
| Exon | n∗ | Outcome | Cox regression | Logistic regression | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| N† | MAC | Frq0 | Frq1 | HR‡ | P LRT§ | N† | MAC | Frq0 | Frq1 | OR‖ | P LRT§ | |||
| 1 | 1 | Solid tumor | 247 | 1 | 0 | 0.0455 | 11.83 | 0.0925 | 206 | 1 | 0 | 0.0455 | >100 | 0.0171 |
| 3 | 4 | Cancer | 247 | 11 | 0.0256 | 0.0096 | 0.16 | 0.0366 | 247 | 11 | 0.0256 | 0.0096 | 0.34 | 0.2564 |
| 3¶ | 4 | HM | 247 | 11 | 0.0267 | 0 | 0 | 0.0189 | 236 | 10 | 0.0256 | 0 | 0 | 0.0455 |
| 8¶ | 4 | Solid tumor | 247 | 8 | 0.0127 | 0.0909 | 11.35 | 0.0209 | 206 | 8 | 0.0154 | 0.0909 | 9.18 | 0.039 |
| 24 | 7 | Solid tumor | 247 | 9 | 0.0169 | 0.0455 | 56.05 | 0.0218 | 206 | 9 | 0.0205 | 0.0455 | 1.83 | 0.5739 |
| 37¶ | 5 | Cancer | 247 | 9 | 0.0103 | 0.0481 | 3.6 | 0.0126 | 247 | 9 | 0.0103 | 0.0481 | 3.46 | 0.0451 |
| 37¶ | 5 | HM | 247 | 9 | 0.0097 | 0.061 | 3.75 | 0.0115 | 236 | 9 | 0.0103 | 0.061 | 4.42 | 0.0203 |
| 38 | 5 | Cancer | 247 | 9 | 0.0154 | 0.0288 | 7.98 | 0.0091 | 247 | 9 | 0.0154 | 0.0288 | 1.96 | 0.3776 |
| 38 | 5 | HM | 247 | 9 | 0.0146 | 0.0366 | 8.48 | 0.0079 | 236 | 9 | 0.0154 | 0.0366 | 2.73 | 0.2026 |
| 42 | 3 | Cancer | 247 | 6 | 0.0051 | 0.0385 | 3.3 | 0.0575 | 247 | 6 | 0.0051 | 0.0385 | 8.73 | 0.0125 |
| 42 | 3 | HM | 247 | 6 | 0.0073 | 0.0366 | 2.81 | 0.1455 | 236 | 5 | 0.0051 | 0.0366 | 8.68 | 0.024 |
| 45 | 1 | Solid tumor | 247 | 2 | 0 | 0.0909 | 0 | 0.3693 | 206 | 2 | 0 | 0.0909 | >100 | 0.0181 |
| 55 | 4 | HM | 247 | 6 | 0.0146 | 0 | 0 | 0.0365 | 236 | 5 | 0.0128 | 0 | 0 | 0.1597 |
| 60 | 6 | Solid tumor | 247 | 8 | 0.0127 | 0.0909 | 6.25 | 0.0634 | 206 | 6 | 0.0103 | 0.0909 | 9.03 | 0.039 |
Frq0, Percentage of patients without cancer with A-T carrying ATM variants; Frq1, percentage of patients with A-T and cancer carrying ATM variants; MAC, cumulative count of minor
Number of variant sites within an exon
Number of patients with A-T in the analysis.
Estimated from Cox regression adjusting for covariates.
Using Cox or logistic regression.
Calculated from logistic regression adjusting for covariates.
Outcome associations significant by both Cox and logistic regressions
Mortality
One hundred eighty-three (36%) participants died during the retrospective study period, among which the age of death (mean ± SD) was 20.5 ± 8.7 and 22.2 ± 7.5 years for those with and without cancer, respectively (Table I). The SMR for the overall A-T cohort independent of cancer diagnosis was 24.6 (95% CI, 21.1–28.4), whereas a notably higher SMR of 232.9 (95% CI, 178.1– 299.2) was observed among individuals with cancer (Table II). We observed 100% mortality resulting in an SMR of 3829.8 (95% CI, 1534.3–7891.3) among those with Hodgkin lymphoma, 273.2 among those with NHL (95% CI, 182.9–392.3), 213.0 among those with leukemia (95% CI, 106.2–381.1), and 136.9 (95% CI, 74.8–229.7) among those with solid tumors (Table II).
Among the 84 individuals with A-T diagnosed with a primary cancer, 62 (73.8%) received chemotherapy, 17 (20.2%) received surgical therapy, and 6 (7.1%) received palliative care/no therapy. Of the chemotherapy recipients, 38 (61.3%) received modified treatment regimens and 24 (38.7%) either received standard therapy (n = 15) or the therapy adjustments were unknown (n = 9). The risk of death was 2-fold (HR, 2.2; 95% CI, 1.1–4.4) higher in those who received standard/unknown dosing strategies compared with those who received modified chemotherapy regimens (P = .024; Fig 2, B). We then assessed for this hazard in those with NHL, the most commonly diagnosed (Fig 1, A) and treated tumor in our cohort, which showed a similar risk of 2-fold (HR, 2.3; 95% CI, 0.9–5.6) increase in standard (n = 11 [45.8%]) versus dose-modified (n = 23 [60.5%]) regimen recipients but did not reach statistical significance because of low sample size (P = .067; see also Fig E2).
Treatment-related organ toxicities
Thirty-seven (60%) of 62 patients were reported to develop chemotherapy-associated toxicities across 8 organ systems. Specifically, neurologic (n = 14), gastrointestinal (n = 10), infectious (n = 5), musculoskeletal (n = 5), genitourinary (n = 4), hematologic (n = 3), cardiac (n = 3), pulmonary (n = 2), and unknown (n = 1) toxicities (Fig 3) were reported by treating oncologists. Eight patients had 2 or more systems involved (Fig 3, B). Neuropathy was the most common neurologic toxicity (n = 10), followed by single cases of headache, worsening ataxia, and 2 unspecified events. Of the gastrointestinal toxicities, 4 involved gastrointestinal bleeding, 2 pancreatitis, and 1 case each of eosinophilic gastroenteritis, abdominal distension, and tamoxifen-associated liver cirrhosis. One remained unspecified. Infectious complications included aspergillus pneumonia, pneumonia, pseudomonal bacteremia, bacteremia not specified, and sepsis. Musculoskeletal toxicities included avascular necrosis (n = 3), Campath-associated muscle pain (n = 1), and muscle weakness (n = 1). Genitourinary toxicity included late-onset hemorrhagic cystitis (n = 3) and ruptured bladder leading to cystectomy (n = 1). Hematologic toxicities involved coagulopathy (n = 2) and a single case of persistent anemia. Pulmonary side effects included single cases of pneumonitis and worsening of lung disease. Cardiac toxicity included cardiomyopathy (n = 2) and reduced ejection fraction (n = 1) following doxorubicin treatment. Baseline ejection fraction was unknown for all 3 cases.
Fig 3.
Chemotherapy-associated toxicities affect multiple organ systems in individuals with A-T. Columns denote single A-T cases with reported chemotherapy-related toxicities. Row labeled P/LP (pathogenic/likely pathogenic) denotes cases with 0, 1, or 2 predicted LOF ATM variants. Δ denotes therapy modifications. Organ systems coded by different colors and more than 1 organ involvement are shown in additional rows.
ATM genotype–cancer correlation
Two hundred fifty-three unique ATM variants were detected in 247 individuals with A-T with available genetic diagnosis and A-T clinical features. Compound heterozygous ATM variants occurred in 85% or more of the 247 individuals. Seventeen individuals (3 in the cancer cohort and 14 in the noncancer cohort) with clinical features, ATM immunoblot, and/or laboratory findings consistent with A-T only had 1 detected ATM allele. Variants were distributed throughout the ATM protein in 52 cancer (Fig 4) and 195 noncancer individuals with A-T. Sixty-two exons contained at least 1 variant, with a median of 4 variants per exon. No variants were observed at exon 51. Of the 186 exon-cancer comparisons, 14 associations were detected by either Cox or logistic regression and 4 were significant by both regression models (Table III). Exon 3 was associated with HMs (Cox regression: HR, 0.00; PLRT = .019; logistic regression: odds ratio [OR], 0.00; PLRT = .046). Exon 8 was associated with solid tumors (Cox regression: HR, 11.35; PLRT = .021; logistic regression: OR, 9.18; PLRT = .046), although the small sample size (n = 11) warrants cautious interpretation. Exon 37 was associated with overall cancer (Cox regression: HR, 3.60; PLRT = .012; logistic regression: OR, 3.46; PLRT = .0451) and HMs (Cox regression: HR, 3.75; PLRT = .012; logistic regression: OR, 4.42; PLRT = .020).
Fig 4.
Germline ATM variants and exons demonstrate cancer risk associations in individuals with A-T. ATM protein map denotes exon numbers. Circles represent single cases, the numbers in the circles represent cases per variant, and different colors correspond to variant type (bottom right). Red arrows denote variants with cancer association. Exons enriched for cancer association are shaded in orange (Cox), yellow (logistic), or red (both regressions). FAT, N-terminal FRAP-ATM-TRAP domain; FRAP, FKBP12-rapamycin-associated protein; TRAP, target of rapamycin; FATC, FAT C-terminal domain; PIKKc domain, phosphoinositide 3-kinase-related kinase catalytic domain; TAN, telomere-associated protein N-terminal domain.
Next, we evaluated associations between single ATM variants and cancer risk. There were no significant associations between predicted LOF variants and cancer risk using logistical regression analysis. Twenty unique ATM variants present in at least 5 individuals with A-T were analyzed for their associations with cancer and HMs, excluding participants with solid tumors because of the small sample size. Individuals with cancer had greater odds of harboring c.6095G>A (OR, 7.223; PLRT = .038), whereas those with HMs had greater odds of harboring c.3245_3247delinsTGAT (OR, 7.806; PLRT = .030). c.170G>A and c.5763–1050A>G variants were found only in individuals without cancer with A-T.
DISCUSSION
We report clinical outcomes for a large national cohort of individuals with A-T and cancer, which reflects population-level risk for this rare, underreported yet medically vulnerable group of individuals. Our study reveals previously unreported observations, including an increased prevalence of solid-organ cancers among individuals with A-T who were older than 18 years and a higher number of secondary malignant neoplasms than previously reported.4,6,10 Furthermore, our data suggest a higher mortality risk in patients with A-T who received standard chemotherapy compared with modified treatment protocols. Importantly, we observed a wide number of chemotherapy-associated organ toxicities that remain inadequately characterized in current literature and critically important for informing new treatment approaches. Our cohort demonstrates that individuals with A-T and cancer experience significant mortality. We also detected a high cumulative incidence (29%) of cancer by age 35 years in our cohort and observed hematological malignancies to disproportionately affect individuals with A-T.
ATM is critical to B- and T-cell development because these cells must generate and repair multiple double-strand DNA breaks to form mature antigen receptors.24−26 In its absence, low B- and T-cell subsets cause reduced cellular immunity and surveillance,27,28 which could contribute to oncogenesis in individuals with A-T. In fact, selective IgA deficiency,29 IgG2 deficiency, and hyper- IgM phenotype30 have been associated with decreased survival among individuals with A-T. We observed the highest occurrence of mature B-cell lymphomas, which is consistent with published studies.4,10 Hodgkin lymphoma was associated with the highest mortality risk, which is in agreement with the study by Elitzur et al31 who showed the lowest 4-year event-free survival among A-T patients with Hodgkin lymphoma compared to other hematologic malignancies. Consistent with previous studies,4,10 we noted a high occurrence of T-cell leukemias spanning a wide age spectrum. This is primarily attributed to T-PLL, a T-cell malignancy typically diagnosed in non-A-T adults in their sixties, being common in our cohort, with a median diagnosis age of 30 years. This contrasts with the low incidence of T-PLL reported among 69 patients with A-T in the French A-T registry.10 We did not observe myeloid malignancies, supporting the rarity of it reported in other cohorts with A-T.4,6,10,31
Carriers of pathogenic/likely pathogenic ATM variants experience a severalfold higher risk for solid cancers.13−21 Data remain limited regarding solid-organ malignancies experienced by individuals with A-T, reflecting the rarity of A-T and the truncated life expectancy.32 Our study represents the largest number of single-cohort A-T–associated solid tumor cases (n = 22) to date. This is likely reflective of our 25-year longitudinal follow-up and thus our ability to capture a greater number of patients with A-T who survived into adulthood because of improved care of other A-T– associated comorbidities. Among the solid tumors, we observed a high incidence of traditionally adult-onset cancers occurring at a much younger age (median age, 26.3 years), with skin and gastrointestinal carcinomas being the most common. Skin cancers have not been systematically reported in patients with A-T. Melanomas accounted for half of skin cancers, which further suggests ATM as a melanoma predisposition gene.16,33 Although breast cancer is the most associated malignancy in ATM carriers, it was rare in our cohort. Previously reported central nervous system malignancies in individuals with A-T were not observed in our cohort.3,4,6,10,12
More than 3000 ATM variants have been classified in ClinVar as pathogenic/likely pathogenic. Presumed null variants have been associated with significant immunodeficiency,3,34 shorter overall survival, and decreased cancer-free survival.10,35,36 Most recently, among 110 patients with A-T who were treated for hematologic malignancies, Elitzur et al31 demonstrated poorer event- free survival and higher treatment-related mortality rates in individuals with A-T with absent ATM kinase activity compared with those with residual kinase activity. Our analysis did not reveal a significant difference in the impact of predicted null variants with absent ATM kinase activity on cancer survival, which may be due to our small sample size. Correlations between ATM genotypes and cancer risk have been assessed through cohort studies and cellular models. A limited number of ATM variants are associated with either a higher (c.1A>G, c.6679C>T, c.7271T>G, and c.8494C>T)37−39 or a lower (c.3576G>A, c.5762–1050A>G, and c.8147T>C)30,40,41 susceptibility to cancer. Through exon enrichment analysis, we demonstrated exons 3, 8, and 37 to be enriched for cancers. Moreover, we found an indel and a missense ATM variant to portend cancer predisposition. Conversely, 1 nonsense and 1 intronic ATM variant were found only in noncancer individuals with A-T, suggesting a potential cancer-protective effect.
Treatment of A-T–associated malignancies is challenging because of the lack of protocol-based cancer therapy regimens and treatment-related organ toxicities. Recently, the largest international cohort (n = 202) of patients with A-T treated for hematologic malignancies demonstrated poor event-free and overall survival because of high treatment-related mortality.31 We present a wide spectrum of chemotherapy-related organ toxicities in patients with A-T highlighting neuropathy and gastropathy as predominant, contrasting with hematologic, infectious, bladder, and pulmonary toxicities reported in studies and case series.4–11 Lack of a standardized grading system spanning the duration of the study limited us to reporting only acute toxicities noted by treating oncologists by organ systems. This, in combination with differences in reporting practices, nomenclature, and malignancy/treatment heterogeneity, precluded cohort-wide comparisons to toxicity rates in patients without A-T who were treated on clinical trials. However, specific examples such as gastrointestinal toxicities in NHL and neuropathy in T-cell leukemia/lymphomas appear to have occurred at a notably higher frequency in individuals with A-T treated with chemotherapy compared with patients without A-T treated on corresponding clinical trials.42,43 The low number of hematologic toxicities (ie, delayed count recovery) in our cohort was likely due to underreporting of this well-recognized and frequently managed side effect by treating oncologists, which is a limitation of our study. We also demonstrated standard dosing strategies to result in twice the risk of death compared with modified therapy regimens among individuals with A-T treated for all cancers, as well as NHL. The reason for this observation remains unclear; however, it is likely due to a combination of factors, including treatment-related mortality, disease progression, and severity of A-T–associated comorbidities. Other small, retrospective studies have observed reductions in toxic side effects,5 insignificant differences in remission rates (T-cell acute lymphoid leukemia),9 and inferior median survival rates with modified chemotherapy regimens in A-T.8 A recent study showed advanced respiratory and neurologic disease in A-T to be correlated with poorer event-free survival and advanced respiratory disease to increase treatment-related mortality.31 Large-scale, prospective, and collaborative efforts are required to determine the impact of attenuated therapy on cancer recurrence.
Our findings should be interpreted in the context of the following limitations. Annual follow-up was hindered by factors such as distance to travel, physical constraints, and loss to follow-up. The study relied on cancer diagnoses reported to the ATCC clinic by oncology providers or families, potentially leading to underreporting of age-related malignancies. Because the study spanned over 25 years and involved observations from 56 different US institutions, it lacked a toxicity grading system. This study duration also coincided with substantial advancements in supportive care approaches, particularly with respect to pulmonary disease and nutrition, which may have influenced the median age of death observed in our cohort. Moreover, our retrospective analysis is constrained by the relatively small number of cancer diagnoses within the cohort of individuals with A-T, a rare disorder. Consequently, the observations and conclusions drawn require further validation. These limitations emphasize the need for additional prospective studies to address knowledge gaps of treating malignancies in DNA repair disorders.
Our study illuminates diverse blood and solid-organ malignancies in A-T along with the complexities associated with cancer treatment. It is crucial to heighten oncologists’ awareness regarding the cancer spectrum, especially solid-organ cancers, and the increased likelihood of causing treatment-related multiorgan toxicities in individuals with A-T. This calls for an international effort to form consensus therapy protocols for the treatment of common cancer types in A-T to improve outcomes. Future therapeutic strategies should prioritize using targeted cancer therapy, thereby minimizing adverse effects on healthy cells and maximizing cancer-free survival. Finally, strategies for cancer surveillance leading to earlier diagnosis need to be implemented, which might be particularly relevant for the outcome of solid adult-type cancers in individuals with A-T.
Supplementary Material
CLINICAL IMPLICATIONS.
Patients with A-T and cancer face elevated mortality rates, underscoring the urgency for tailored therapies to minimize toxicity and improve survival outcomes.
ABBREVIATIONS
- A-T
Ataxia telangiectasia
- ATCC
Johns Hopkins Ataxia Telangiectasia Clinical Center
- ATM
Ataxia telangiectasia mutated
- HM
Hematologic malignancy
- HR
Hazard ratio
- IQR
Interquartile range
- LOF
Loss of function
- LRT
Likelihood ratio test
- NHL
Non–Hodgkin lymphoma
- OR
Odds ratio
- SIR
Standardized incidence ratio
- SMR
Standardized mortality ratio
- T-PLL
T-cell prolymphocytic leukemia
Footnotes
DISCLOSURE STATEMENT
This work was funded by the National Institutes of Health (grant no. RR000052 to H.L., T.O.C., J.W., M.A.L.G., and S.M.M. and grant no. R21 TR003534 to H.L., T.O.C., J.W., M.A.L.G., S.M.M., and V.N.), the Food and Drug Administration (grant no. R01 FD007605 to H.L., T.O.C., J.W., M.A.L.G., S.M.M., and V.N.), and the A-T Children’s Project (to H.L., T.O.C., J.W., M.A.L.G., S.M.M., and V.N.).
Disclosure of potential conflict of interest: The authors declare that they have no relevant conflicts of interest.
REFERENCES
- 1.Gatti RA, Berkel I, Boder E, Braedt G, Charmley P, Concannon P, et al. Localization of an ataxia-telangiectasia gene to chromosome 11q22–23. Nature 1988;336:577–80. [DOI] [PubMed] [Google Scholar]
- 2.Lee JH, Paull TT. Cellular functions of the protein kinase ATM and their relevance to human disease. Nat Rev Mol Cell Biol 2021;22:796–814. [DOI] [PubMed] [Google Scholar]
- 3.Veenhuis S, van Os N, Weemaes C, Kamsteeg EJ, Willemsen M. Ataxiatelangiectasia. In: Adam MP, Feldman J, Mirzaa GM, Pagon RA, Wallace SE, Bean LJH, et al. , editors. GeneReviews. Seattle: University of Washington; 1993. [Google Scholar]
- 4.Bakhtiar S, Salzmann-Manrique E, Donath H, Woelke S, Duecker RP, Fritzemeyer S, et al. The incidence and type of cancer in patients with ataxia-telangiectasia via a retrospective single-centre study. Br J Haematol 2021;194:879–87. [DOI] [PubMed] [Google Scholar]
- 5.Bienemann K, Burkhardt B, Modlich S, Meyer U, Moricke A, Bienemann K, et al. Promising therapy results for lymphoid malignancies in children with chromosomal breakage syndromes (ataxia teleangiectasia or Nijmegen-breakage syndrome): a retrospective survey. Br J Haematol 2011;155:468–76. [DOI] [PubMed] [Google Scholar]
- 6.Dutzmann CM, Spix C, Popp I, Kaiser M, Erdmann F, Erlacher M, et al. Cancer in children with fanconi anemia and ataxia-telangiectasia—a nationwide register based cohort study in Germany. J Clin Oncol 2022;40:32–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sandlund JT, Hudson MM, Kennedy W, Onciu M, Kastan MB. Pilot study of modified LMB-based therapy for children with ataxia-telangiectasia and advanced stage high grade mature B-cell malignancies. Pediatr Blood Cancer 2014;61:360–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sandoval C, Swift M. Treatment of lymphoid malignancies in patients with ataxiatelangiectasia. Med Pediatr Oncol 1998;31:491–7. [DOI] [PubMed] [Google Scholar]
- 9.Schoenaker MH, Suarez F, Szczepanski T, Mahlaoui N, Loeffen JL. Treatment of acute leukemia in children with ataxia telangiectasia (A-T). Eur J Med Genet 2016;59:641–6. [DOI] [PubMed] [Google Scholar]
- 10.Suarez F, Mahlaoui N, Canioni D, Andriamanga C, Dubois d’Enghien C, Brousse N, et al. Incidence, presentation, and prognosis of malignancies in ataxiatelangiectasia: a report from the French national registry of primary immune deficiencies. J Clin Oncol 2015;33:202–8. [DOI] [PubMed] [Google Scholar]
- 11.Upadhyaya SA, Mody R, Walkovich K, Hutchinson RJ, Sandlund JT, Connelly JA. Ataxia telangiectasia and cancer predisposition: challenges in management. J Pediatr Hematol Oncol 2018;40:483–6. [DOI] [PubMed] [Google Scholar]
- 12.Morrell D, Cromartie E, Swift M. Mortality and cancer incidence in 263 patients with ataxia-telangiectasia. J Natl Cancer Inst 1986;77:89–92. [PubMed] [Google Scholar]
- 13.Renwick A, Thompson D, Seal S, Kelly P, Chagtai T, Ahmed M, et al. ATM mutations that cause ataxia-telangiectasia are breast cancer susceptibility alleles. Nat Genet 2006;38:873–5. [DOI] [PubMed] [Google Scholar]
- 14.Kurian AW, Hughes E, Handorf EA, Gutin A, Allen B, Hartman AR, et al. Breast and ovarian cancer penetrance estimates derived from germline multiple-gene sequencing results in women. JCO Precis Oncol 2017;1:1–12. [DOI] [PubMed] [Google Scholar]
- 15.Hall MJ, Bernhisel R, Hughes E, Larson K, Rosenthal ET, Singh NA, et al. Germline pathogenic variants in the ataxia telangiectasia mutated (ATM) gene are associated with high and moderate risks for multiple cancers. Cancer Prev Res (Phila) 2021;14:433–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Goldstein AM, Xiao Y, Sampson J, Zhu B, Rotunno M, Bennett H, et al. Rare germline variants in known melanoma susceptibility genes in familial melanoma. Hum Mol Genet 2017;26:4886–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Dombernowsky SL, Weischer M, Allin KH, Bojesen SE, Tybjaerg-Hansen A, Nordestgaard BG. Risk of cancer by ATM missense mutations in the general population. J Clin Oncol 2008;26:3057–62. [DOI] [PubMed] [Google Scholar]
- 18.Shindo K, Yu J, Suenaga M, Fesharakizadeh S, Cho C, Macgregor-Das A, et al. Deleterious germline mutations in patients with apparently sporadic pancreatic adenocarcinoma. J Clin Oncol 2017;35:3382–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Pritchard CC, Mateo J, Walsh MF, De Sarkar N, Abida W, Beltran H, et al. Inherited DNA- repair gene mutations in men with metastatic prostate cancer. N Engl J Med 2016;375:443–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.