Patients with myelodysplastic syndromes (MDS) have high frequencies of cytogenetic abnormalities and evidence is accumulating of associations between exposure history and primary MDS. The objective of this article is to examine the relationship between histories of occupational or environmental exposure and presence of cytogenetic abnormalities. A case control study of MDS patients estimated lifetime exposure to more than 90 potential hazards in 400 age, sex, and area of residence matched patient and control pairs. A parallel cytogenetics study undertaken at time of diagnosis, independently of any knowledge of exposure history, identified 75 cytogenetically abnormal and 139 normal (186 not studied). Odds ratios of MDS patients and their matched controls were compared for 3 groups: cytogenetically abnormal, normal, and not known. The odds ratios for all exposures combined were possibly higher among cytogenetically abnormal 2.0 (95% confidence interval 0.8-5.9) than among normal 1.0 (0.6-1.8). This pattern was observed for exposure to semimetals, abnormal 4.0 (0.4-195.1) and normal 0.5 (0.1-1.0) and inorganic dusts, 1.6 (0. 6-3.8) and 0.4 (0.1-1.4) respectively. The pattern was principally in abnormalities in chromosomes 5 and 7. For organic chemicals and radiation, the odds ratios for both cytogenetically abnormal and normal were marginally raised: organic 1.8 (0.6-6.0) and 1.3 (0.6-2.9), respectively, and radiation 1.7 (0.5-5.6) and 1.3 (0.4-4.7) respectively. For radiation, abnormalities were mostly in chromosome 8. This study of association between exposures and cytogenetics in primary MDS complements those previously reported in secondary MDS and may provide some insight into pathogenetic mechanisms that lead to development of MDS.

Cytogenetic abnormalities are identified at diagnosis in 30% to 70% patients with de novo myelodysplastic syndrome (MDS); the frequency increasing with higher risk disease.1 2Chromosome translocations in MDS are rare and the most common karyotypic lesions involve chromosomes 8 (gain), 5 (loss/deletion), and 7 (loss/deletion).3 The accumulation of karyotypic abnormalities with disease progression provides some support for the multistep process of malignant transformation from MDS to acute myeloid leukemia (AML). Survival of patients with abnormalities involving chromosomes 5, 7, and 8 has been shown to be significantly reduced compared with patients with normal karyotypes.2-4

The etiologic insults leading to the development of MDS and the latency period between the initial genomic insult and disease manifestation are largely unknown. The best-defined xenobiotic insult in the development of MDS is that which follows cytotoxic chemotherapy for cancer with alkylating agents. Therapy related MDS (t-MDS) is associated with a higher frequency of karyotypic abnormalities than de novo MDS.5 6 The majority of these abnormalities involve chromosomes 5 and/or 7 suggesting that these chromosomes are particularly susceptible to genomic damage and that this leads to proliferative advantage. Furthermore, it has been shown that chemotherapy treated patients in clinical remission also harbourRAS and/or FMS oncogene mutations in peripheral blood DNA in the absence of hematologic disease and this may be a manifestation of genomic instability or damage.7-10Postchemotherapy patients do not, however, show increased chromosome aberration frequencies compared with normal subjects, although they do show qualitative differences in the type of aberrations. A higher frequency of exchanges is seen amongst patients, particularly in those who received multiple compared with single courses of therapy and the frequency of gaps is lower.11

There have been many reports of associations between histories of exposures to certain organic chemicals, notably benzene solvents, pesticides, and radiation and MDS.12-14 However, only benzene has been strongly implicated in the etiology, with an elevated relative risk identified in a large cohort study of benzene-exposed workers compared with nonexposed controls.15 Benzene exposed workers developing hematologic abnormalities also showed polymorphism in metabolic pathways, which would predispose to the accumulation of the highly genotoxic quinone benzene metabolic intermediates.13 In vitro benzene metabolites induce peripheral blood lymphocyte chromosome 5 and 7 loss and long arm deletion.14 It has also been suggested that exposure to pesticides and organic solvents are associated with aberrations in chromosomes 5 and 7 in both AML and MDS. These were consecutive patients referred to the centers for specialist treatment of their conditions, not by reason of suspected past exposures. Exclusions were only for early death (less than 1 month of diagnosis) or severe illness (too ill to be interviewed).18-22 To elucidate the role of environmental mutagens in the pathogenesis of MDS, this study investigates the relationship between a history of exposure to chemicals/hazards and cytogenetic changes in primary MDS.

Materials and methods

Case-control study of lifetime exposure

Four hundred primary myelodysplastic syndrome (MDS) patients, diagnosed in 3 specialist regional centers (Bournemouth, Cardiff, and Leeds), were referred to the study. Controls were selected from hospital outpatients and inpatients, with a broad range of diagnoses but excluding malignancy, and were matched by age, sex, and area of residence. Lifetime occupational or environmental exposures of MDS patients and matched controls were estimated by questionnaire and semistructured interview at home after discharge by trained interviewers “blind” as to the case/control status of the patients.16 Lifetime exposure histories were estimated for more than 90 chemicals or putative hazards, including radiation. Exposures at 3 subjective intensities: (1) light (for example, being in the same room as open chemicals), (2) moderate (for example, working directly with chemicals), and (3) heavy (for example, working with volatile substances in confined space with poor ventilation and/or poor protection) were estimated as hours per day, days per year, and years to give lifetime hours. In analysis, patients were classified as “exposed” at 4 “thresholds”: (1) = 10 hours at = low intensity (a practical minimum detectable memorable level), (2) = 2500 hours at = low intensity, (3) = 50 hours at = moderate intensity, and (4) = 2500 hours at = moderate intensity to each of 90 chemicals or hazards (for example, arsenic), 13 groups of chemicals or hazards (for example, semimetals), 3 major groupings (organics, inorganics, and radiation), and any potential hazard.

Cytogenetic analysis

Bone marrow karyotype analysis was carried out according to conventional cytogenetic procedures. Approximately 20 metaphases were karyotyped by G banding.3 23 Clonal abnormalities were defined as 2 or more cells with the same additional whole chromosome or chromosome rearrangement, or 3 or more cells with the same chromosome missing. In line with other studies, the most frequent clonal abnormalities were trisomy 8, loss or deletion of chromosome 5, and loss or deletion of chromosome 7. Karyotypes were defined according to the Cytogenetic Nomenclature.23 Karyotype analysis was opportunistic and depended on availability of bone marrow aspirate. There were no recognizable biases in selecting patients for karyotyping but, in case there were unrecognized biases, the study design compared the odds ratios of cytogenetics known with cytogenetics not known.

Statistical analysis

Odds ratios (ORs) for each putative exposure were calculated as the ratio of discordant pairs24 and 95% confidence intervals were based on the binominal distribution.25 The analysis compares the ORs of matched pairs (each comprising 1 MDS patient and 1 age, sex, and area of residence matched control) among 3 groups of MDS patients: cytogenetically normal, cytogenetically abnormal, and cytogenetics not known. The last group was added for comparison, because not all patients included in the case control study were karyotyped. Because overall ORs of MDS patients exceeded 1.0 (averaged 1.2) and for several hazards significantly exceed 1.0,16 an association between exposure and cytogenetic abnormality is indicated not by an absolute OR but by comparison of ORs among cytogenetically abnormal with ORs among cytogenetically normal. The comparison thus seeks OR (abnormal) > OR (normal). Furthermore, because the cytogenetics not known group includes normals and abnormals, the OR of this group would be expected to lie between the above 2. The analysis starts with all potential hazards combined and focuses progressively through 3 major groupings, 13 groups to 90+ individual chemicals. The OR differences were assessed also for 3 most common chromosome abnormalities 5,7, and 8. Because numbers of patients with specific chromosome abnormalities were small, their “unmatched” ORs were also compared with those of cytogenetically normal patients. These unmatched ratios should be interpreted with caution, because of possible differences between groups in MDS diagnosis, age, sex, and area of residence. The principal results are summarized in ORs for exposures at ≥ 50 hours at ≥ “moderate” intensity (“threshold 3” above) with 95% confidence intervals and the ratios of OR abnormal/OR normal. The tables show those ORs that were significant at P < .05.


Patient characteristics

Cytogenetic analysis was completed in 214 MDS patients. There was no significant difference in patient characteristics, age, sex, and clinical diagnosis between these and patients for whom cytogenetic status was not known. Seventy-five (35%) had abnormal cytogenetics; the more common chromosomal abnormalities were in chromosome 8 (21,18 trisomy 8), chromosome 5 (14,7 monosomy 5) and chromosome 7 (9,6 monosomy 7). Lifetime exposure histories were obtained for a further 186 MDS patients, for whom cytogenetics were not known. Patients with cytogenetic abnormalities were possibly older than those who were cytogenetically normal (χ2 = 7.94, df = 4,P < .15), possibly included more men (χ2 = 3.2, df = 1, P < .10) and were also possibly more likely to be diagnosed with the poor prognostic FAB subtypes RAEB and RAEB t (χ2 = 6.6, df = 3,P < .10), but none of these differences were statistically significant.

Exposure and cytogenetic abnormality

All exposures combined and organic, inorganic, and radiation.

The OR for MDS patients with a history of any exposure (comparing patients with their age, sex, and area of residence matched controls) was higher for patients with abnormal cytogenetics than for those with normal cytogenetics: at all exposure thresholds: at threshold 3 (≥ 50 hours at ≥ moderate intensity) OR of abnormals 2.0 (95% confidence interval 0.8-5.9) and normals 1.0 (0.6-1.8). The ratio OR abnormal OR normal was 2.0 (1.0-3.9). For comparison, the OR of 186 MDS patients without known cytogenetics was 1.3 (0.8-2.2), lying between those for normal and abnormal. Similar trends were observed at other thresholds; both lower intensity and greater total hours. When putative hazards were separated into 3 broad groups, the pattern was possibly more marked for exposures to inorganics than for organics or radiation: the ORs for inorganic exposures for cytogenetically abnormal, normal, and not known were 2.4 (1.0-6.9), 0.8 (0.5-1.4), and 1.9 (1.1-3.2), respectively, and the ratio of abnormal to normal was 3.0 (1.5-5.6).

Exposures to 13 generic groups of chemicals/hazards (Table1).

Odds ratios for cytogenetically abnormal, normal, and not known and the ratios of the abnormal to normal ORs for exposures to 13 generic groups of chemicals (or hazards) at “threshold 3” (≥ 50 hours and ≥ moderate intensity) are shown in Table 1. In 2 (of 13) chemical groups, semimetals and inorganic dusts, the ORs for cytogenetically abnormal were twice or more than those for cytogenetically normal and the lower confidence intervals of these 2 ratios of ORs exceed 1.0. There were no groups for which ORs of cytogenetically abnormal were half or less than those for cytogenetically normal. Similar trends were observed for exposures above other thresholds, for example, at “threshold 4” (≥ 2500 hours at ≥ moderate intensity), they were abnormal 2.0 (0.1-118.0), normal 0.3 (0.0-2.5), and not known 0.5 (0.0-9.6) for semimetals and 1.2 (0.4-3.4), 0.4 (0.1-1.1), and 1.1 (0.5-2.3), respectively, for inorganic dusts. The OR for cytogenetically abnormal was ≥ 2 times that for cytogenetically normal for radiation at “threshold 4”: abnormal 6.0 (0.7-276.8), normal 1.5 (0.4-7.2), and not known 2.1 (0.8-6.2).

Table 1.

Odds ratios for occupational and environmental exposure and cytogenetic classification

Exposure to individual chemicals (Table2).

The ORs for individual chemicals or hazards in the metal, semimetal, and inorganic dust groups are summarized in Table 2. Arsenic showed consistent OR difference between abnormals and normals more than or equal to 2 × at all thresholds of exposure. Asbestos, silica, and formica dusts also showed OR differences, more than or equal to 2 × at several thresholds of exposure. The lower confidence intervals of the ratios of ORs were above 1.0 for copper, arsenic, and silica.

Table 2.

Odds ratios for occupational and environmental exposure and cytogenetic classification

Chromosome specificity (Tables 3, 4, 5)

ORs for 3 main chromosome abnormalities, 5, 7, and 8 are shown for exposure to 13 generic groups in Tables 3, 4, and 5, respectively. For these comparisons, the lowest threshold of exposure was chosen (≥ 10 hours and ≥ low intensity), because numbers of MDS patients in each abnormality were small. The tables also include the ORs in unpaired comparisons (the specific chromosome abnormality compared with the cytogenetically normal). Chromosome 5 abnormalities showed an elevated OR for inorganic gases and fumes (which included exhaust gases, ammonia fumes, hydrogen peroxide and mineral acids) (Table3): both the matched pairs 8.0 (1.1-356.1) and the unmatched comparison 4.3 (1.3-13.6) were statistically significant (at P < .05). Chromosome 7 abnormalities showed elevated ORs for 3 groups: organics, inorganic dusts, and inorganic gases, all significant in the unmatched comparison (Table4). Chromosome 8 abnormalities showed 3 possible associations, with organics, metals, and radiation but none achieved statistical significance (Table5). Because numbers in each chromosome abnormality group were small and each was tested for 13 exposures, the 4 statistically significant findings should be interpreted with caution. However, they were all in the same direction (abnormal OR ≥ normal OR) and, as in Tables 1 and 2, the trends were this direction (ratio ≥ 2 in 14 and < ½ in 2).

Table 3.

Chromosome 5: Abnormalities and exposure odds ratios

Table 4.

Chromosome 7: Abnormalities and exposure odds ratios

Table 5.

Chromosome 8: Abnormalities and exposure odds ratios


Several previous reports of association between occupational exposure and cytogenetic abnormality in MDS have involved small numbers of patients26 27 and have been based on internal comparisons between cytogenetically normal and abnormal patients. In such studies, age, sex, and diagnosis are potential confounders, because each may be associated with cytogenetic classification. A recent larger case-control study identified a relationship between exposure to pesticides or solvents and clonal abnormalities of chromosomes 5, 7, and 8,28 a relationship also previously suggested by 2 smaller studies of AML.21 22 It is possible that the differences in karyotype between the exposed and nonexposed groups of patients could have been influenced by the greater age and male/female ratio in the exposed versus the nonexposed group. The present study includes more patients than most previous reports and examines ORs in case and control pairs, matched for age, sex and area of residence, which helps to reduce the possible effect of confounders.

Chemicals found to be associated with cytogenetic abnormality in this study were semimetals, metals, and inorganic dusts. Within these groups, the individual chemicals most associated were arsenic (semimetals), copper, nickel, tin, and steel (metals), and asbestos, silica, and formica (inorganic dusts). Arsenic and asbestos are known human carcinogens,29 although not confirmed leukemogens. It has been shown that arsenite potentiates chromosomal aberrations, induced by a DNA cross-linking agent and potential mechanism for this may be an inhibition of DNA repair.30 Arsenic trioxide induces apoptosis in transformed myeloid and lymphoid cells 31 32and may contribute to the initiation of excessive apoptosis and ineffective hematopoiesis in MDS.33 An excess tissue concentration of crocidolite asbestos has been found in patients with acute myeloid leukemia34 and this form of asbestos induces oxidative DNA damage in HL-60 cells.35 Similarly, copper augments oxygen free radical mediated DNA damage, induced by reactive benzene intermediates, which suggests an additive role for copper exposure in benzene-induced mutagenesis.36

The association with radiation was modest: the OR for abnormals was more than twice that for normals only for exposures above the highest threshold investigated. The difference may be underestimated, since the OR for the “cytogenetics not known” group was higher still (3.0). Any explanation for an apparently weak association between radiation and cytogenetics should not be confused with associations between radiation and MDS generally: the OR for all 400 patients was 2.1 and statistically significant atP < .05.16 Whether or not radiotherapy adds to the risk of therapy-related leukemia above that of chemotherapy alone is controversial.6 DNA repair mechanisms may be impaired in individuals exposed to mutagens, rendering cells apparently radiosensitive.37 It is thought that DNA repair genes may be compromised in cancer due, for example, to mutations of check point genes, such as the tumor suppressor gene, p53, which arrest the cell cycle and allow DNA repair to occur. The time for DNA damage repair before activation of apoptosis is a critical determinant of radiosensitivity. The effects of radiation may be subtle and in this study correlated strongly with increased cytogenetic abnormalities only at the higher thresholds of exposure. Radiation induces DNA strand breaks. Therefore, point mutations of critical target genes, rather than gross cytogenetically detectable lesions, may contribute to the association between radiation and MDS in the study cohort as a whole.16

Individual chromosome abnormalities appeared to show different patterns of exposure, although numbers were small and therefore these findings should be interpreted with caution. Abnormalities of chromosome 5 and 7 were associated with exposure to inorganic gases/fumes, whereas abnormalities of chromosome 8 were associated with exposure to metals, organics, and radiation. Preferential chromosome damage has been noted after in vitro incubation of benzene metabolites with human lymphocytes14 and we postulate that other genotoxins may also show such specificity. Paradoxically, workers occupationally exposed to benzene showed higher gene duplication variants than deletional variants in the glycophorin A mutational assay.38 Thus in vivo, chromosome gain may predominate after benzene exposure demonstrating the complexity of extrapolating in vitro observations to the in vivo situation.

The specificity of well-established leukaemogens (alkylating cytotoxic drugs) and environmental xenobiotics (such as inorganic gases/fumes) for chromosome 5 and 7 damage in MDS patients provides support for the presence of tumor-suppressor genes on these chromosomes. These genes could prove to be critical regulators of hematopoiesis, loss of function of which is implicated in the pathogenesis of MDS/AML. The gain of an entire chromosome in trisomy 8 suggests an abnormality in the regulators of cell division. This study has identified potential environmental factors that may contribute to the 3 most common chromosomal abnormalities of MDS and may provide some insight into the pathogenetic mechanisms leading to the development of MDS and AML.


We are grateful to hematologists (Professors A. Burnett, T. Hamlin, A. Jacobs, Drs P. Bentley, G. Bynoe, D. Oscier, and A. Smith) for referring patients, interviewers, (Mrs J. Carter, S. Chell, K. Dunlop, K. Alias, A. Hawkes, S. Middleton, and S. Morris), Messrs A. Russon, M. Tooley for computing assistance, Mrs J. Knight for typing the manuscript, and last but not least to the many patients themselves, without whose cooperation the study would not have been possible.


  • D.T.B. is currently at Department Molecular and Cellular Pathology, University of Dundee. R.A.P. is currently at GW Hooper Research Laboratories, University of California, San Francisco, CA.

  • Submitted April 15, 1999; accepted October 29, 1999.

  • Supported by grants from the Medical Research Council and Leukemia Research Foundation.

  • Reprints: Robert West, University of Wales College of Medicine, Epidemiology, Heath Park, Cardiff, Wales CF4 4XN, United Kingdom; e-mail: westrr{at}

  • The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 U.S.C. section 1734.


View Abstract