Validation of micronuclei frequency in peripheral blood lymphocytes as early cancer risk biomarker in a nested case–control study

Available online at www.sciencedirect.com

Mutation Research 639 (2008) 27–34

Validation of micronuclei frequency in peripheral blood lymphocytes as early cancer risk biomarker in a nested case–control study Elena Murgia a,∗ , Michela Ballardin a , Stefano Bonassi b , Anna Maria Rossi a , Roberto Barale a a

b

Department of Biology, Pisa University, via Derna 2, Pisa, Italy Unit of Molecular Epidemiology, National Cancer Research Institute, Genoa, Italy

Received 6 July 2007; received in revised form 14 October 2007; accepted 26 October 2007 Available online 13 November 2007

Abstract Aim of this work was to assess the predictive value of micronuclei (MN) frequency in peripheral blood lymphocytes (PBL) for the risk of cancer death in disease-free individuals. Blood samples from 1650 subjects selected from the general population of Pisa, Italy, were collected between June 1991 and November 1993. The follow-up until January 2005 recorded a total of 111 deaths (52 for cancer). MN frequency was assessed for 49 cancer cases and 101 matched controls. A significantly higher MN frequency was found in cancer cases (4.7 ± 3.4 MN/1000 BN cells) versus controls (1.5 ± 1.7; p < 0.0001). Donors were stratified in two classes and multivariate logistic regression analysis confirmed that individuals with high MN frequency (>2.5 MN/1000 BN cells) had a significantly increased risk of cancer death (OR = 10.7; 95% CI = 4.6–24.9; p < 0.0001) when compared to individuals with low MN frequency (≤2.5 MN/1000 BN cells). Ageing was associated with a 6% increased risk per year (p = 0.03). No influence of other potential confounders (gender, occupation, smoking and drinking habits) was observed. Finally, subjects with a higher MN frequency showed a higher MR for CVD (Logrank test, p = 0.001). These findings provide strong evidence that MN frequency assessed in PBL of disease-free subjects is a good predictor of cancer death risk, evaluated by a nested case–control study performed 14 years after the original recruitment. © 2007 Elsevier B.V. All rights reserved. Keywords: Biological markers; Micronucleus test; Cancer risk; Case–control study; Validation

1. Introduction Cancer represents one of the most common causes of death in developed countries, second only to cardiovascular diseases [1,2]. Among the most promising strategies for cancer prevention, there is the development and validation of biomarkers, which can anticipate the

∗

Corresponding author. E-mail address: [email protected] (E. Murgia).

0027-5107/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.mrfmmm.2007.10.010

clinical diagnosis and address cancer prevention interventions in populations at risk, because of exposure to carcinogens or for their genetic susceptibility. Among the candidate cancer risk biomarkers, classic cytogenetic endpoints, measured since decades in peripheral blood lymphocytes (PBL), such as chromosomal aberrations (CAs), sister chromatid exchanges (SCE) and micronuclei (MN), have attracted great interest. CAs is the first biomarker of chromosome damage that has been consistently associated with the overall cancer risk [3–5]. Despite the positive evidence linking CAs to cancer risk,

28

E. Murgia et al. / Mutation Research 639 (2008) 27–34

the assay is labour and time consuming, and therefore the use of simpler and quicker risk biomarkers is more suitable. The MN assay is one of the best candidates for wide usage in public health strategies and potentially in individual risk assessment. MN have the appearance of a small nucleus, formed by condensation of chromosome fragments (originated by chromosome breakage) or by whole chromosomes that are left behind during anaphase movements (originated by spindle disturbance) due to either lack of a centromere in the case of acentric chromosome fragments or defects in centromeres or kinetochores or the spindle in the case of whole chromosomes. Thus, MN may reveal several genomic instability events which are associated with malignant cell transformation [6]. The use of MN as a measure of cytogenetic damage in PBL was first proposed by Countryman and Heddle [7]. Subsequently, the assay was improved with the development of the cytokinesis-block micronucleus method (CBMN), allowing micronuclei to be scored specifically in cells that had completed a nuclear division which eliminates confounding effects of variability in cell division kinetics [8]. The hypothesis of an association between the MN level in target or surrogate tissues and cancer development is supported by a number of indirect findings [9]: a clear increase in the frequency of MN has been observed in cancer patients [10]; subjects affected by genetic diseases, such as Bloom syndrome, have both abnormally high MN frequencies and increased cancer risk [11]; in clinical chemoprevention trials of oral premalignancies, MN in oral mucosa have been used as a surrogate cancer endpoint [12]; a suggestive correlation exists between carcinogenicity and genotoxicity, e.g. Class I carcinogens such as ionising radiation, benzene and tobacco smoke which induce increased MN frequencies in humans and in animals [13]. In addition, increased MN levels were found in fibroblasts and PBL of patients affected by familial cutaneous malignant melanoma (CMM) [14] and in PBL of smokers with lung cancer, as compared to healthy subjects [15]. A correlation between progressive stages of cervical uterine cancer and MN frequency in cervical epithelium and in PBL suggested an association between MN frequency and the cervical cancer pathogenesis [16]. More recently, a significantly increased level of MN was observed in PBL of patients affected by pleural malignant mesotelioma, as compared to patients with lung cancer, benign respiratory diseases and to healthy controls [17]. Moreover, the results of El-Zein et al. [18] showed the significant association of MN frequency in

lymphocytes with lung cancer risk and also that the other biomarkers in CBMN assay (i.e. nucleoplasmic bridges and nuclear buds) were strongly associated with cancer risk. However, despite this extensive evidence, it is difficult to establish a causal relationship between increased MN frequency and cancer risk, because a high MN level in cancer patients could be consequence of their disease status or reflect their individual susceptibility to genomic instability events. The most suitable approach to assess the role of MN in PBL as predictor of cancer risk would be the evaluation of MN levels in cancer-free donors, following them up in a longitudinal cohort study. In this way, the role of possible confounders can be taken into account, as demonstrated by collaborative studies which quantified the influence of demographic and lifestyle factors, such as age, gender, and tobacco smoking, on MN frequency [9,19]. A recent study assessed the prospective association between MN frequency and cancer risk in a large international cohort of disease-free subjects [20]. This multinational collaborative study showed a significantly higher prospective risk of cancer (RR 1.67, p = 0.002) in those diagnosed with MN frequency in the medium or high tertile compared to those in the low MN frequency tertile. It is possible that the observed association between MN frequency and cancer risk in this study may have been weakened by the large heterogeneity in MN frequency between cohorts that may have been caused by differences in lifestyle, diet and other environmental factors as well as differences in assay protocol and slide scoring and differences in accuracy of cancer registry data between cohorts that may not have been accounted for completely by the study design. The aim of the present nested case–control study is to validate the measurement of MN frequency in PBL as early biomarker of cancer risk in a single population under strictly controlled technical conditions. The longitudinal design of this study, which allowed assessing MN frequency several years before the subjects developed the disease, seems to be the most appropriate. 2. Materials and methods 2.1. Studied population The subjects selected for this nested case–control study were extracted from a cohort of 1650 disease-free individuals participating to a cytogenetic survey of the general population living in three areas with different levels of urban air pollution in the Province of Pisa, Italy. More detailed description of the original population can be found in Barale et al. [21]. Blood samples were collected by trained personnel from June 1991 to November 1993 in heparinized vacutainers. A

E. Murgia et al. / Mutation Research 639 (2008) 27–34

standardized questionnaire was administered to gather information about age, occupation, demographics, and lifestyle factors (smoking and drinking habits) at time of blood sampling. The vital status of all subjects was monitored through the registry of the municipality of residence from the date of blood sampling to January 2005. Death causes were retrieved from the regional registry of death causes. At the end of the follow-up, 111 deaths were recorded (52 for cancer, 59 for other causes). Only 49 cancer cases (30 males and 19 females) could be analysed for the present study, because the quality of slides of three subjects was poor and these were not further included in the analysis. One hundred and one cancer-free controls (66 males and 35 females) were matched with their corresponding cases by age (±5 years) and gender. 2.2. Lymphocyte cultures and MN analysis Peripheral blood samples obtained by venipuncture were immediately processed and cultured according to standard procedure [21]. A heparinized whole-blood sample (0.3 mL) was added to 4.7 mL of culture medium composed of 4.025 mL Ham’s F10 medium (ICN, Irvin, CA, USA) supplemented with 0.5 mL (10%), foetal calf serum (ICN), 0.075 mL (1.5%) phytohemagglutinin (PHA, Wellcome, Pomezia, Italy) and antibiotics (100 IU penicillin and 100 ␮g/mL streptomycin; Sigma, St. Louis, MO, USA). The cytokinesis-block micronucleus (CBMN) method [8] was used to assess MN frequency. Cells were blocked with cytochalasin B (3 ␮g/mL; Sigma) at

29

44 h and harvested at 72 h according to the CBMN standard protocol. The ratio binucleated cell/total cells (BN/Tot) was also assessed as cell proliferation index after 72 h of culturing. From the time of the collection, blood samples were identified only by sample codes, and all laboratory analyses were carried out blind to the status of the subjects. The original slides of the survey, prepared in early 1990s, were stored at room temperature ever since and retrieved for the current analysis. Criteria for cells and MN scoring were as described by Fenech and Morley [8]. Briefly, the cells should be binucleated with an intact nuclear membrane and as should be situated within the same citoplasmic boundary. MN should be morphologically identical to but smaller than nuclei, their diameter usually varied between 1/6th and 1/3rd of the mean diameter of the main nuclei. 2000 binucleated cells/donor were scored by a single scorer [22]. 2.3. Statistical analysis The distributions of MN levels and age were highly skewed and could not be normalised by standard procedures. Therefore significance of the difference between cancer cases and controls was analysed by the nonparametric Kruskall–Wallis test. A ROC analysis was performed to assess the potential of MN frequencies to discriminate between cases and controls. This procedure eliminates the problem of choosing a cut-off

Table 1 Demographic and lifestyle characteristics of the examined subjects Data at time of blood sampling

Cancer cases (n = 49)

Healthy controls (n = 101)

p-Value

Age Mean ± S.D. Range (years)

62.9 ± 7.8 41–76

60.1 ± 8.3 39–78

0.052a

Gender Males Females

30 (61.2%) 19 (38.8%)

66 (65.3%) 35 (34.7%)

0.64b

Residence site Cascina-Navacchio Pisa

22 (44.9%) 27 (55.1%)

31 (30.7%) 70 (69.3%)

0.132

Smoking status Smoker Never smoker

15 (30.6%) 34 (69.4%)

23 (22.8%) 78 (77.2%)

0.45b

Drinking status Drinker Abstainer

34 (69.4%) 15 (30.6%)

63 (62.4%) 38 (37.6%)

0.46b

Occupation Housewives Retired White collars Blue collars Not known

27 (55.1%) 4 (8.1%) 4 (8.1%) 8 (16.3%) 6 (12.2%)

37 (36.6%) 17 (16.8%) 11 (10.9%) 12 (11.9%) 24 (23.8%)

0.24b

a b

Kruskall–Wallis test. Chi-squared test with Yates’ correction.

30

E. Murgia et al. / Mutation Research 639 (2008) 27–34

Table 2 Demographic and lifestyle characteristics of the 52 cancer cases Agea

Genderb

Residencec

Cancer sited

Year death

Year of MN test

62 65 45 73 59 70 69 69 68 53 58 61 57 67 74 60 66 64 57 59 70 67 67 65 60 61 70 57 65 73 69 58 67 70 46 70 67 64 65 48 55 64 61 70 71 66 69 59 56 76 41 49

2 1 1 1 1 1 1 1 1 1 1 2 1 2 2 1 1 2 2 1 1 2 1 1 1 1 1 2 2 2 1 1 2 1 1 2 2 2 1 2 1 1 1 1 2 1 2 2 1 1 2 2

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

1 1 1 2 1 3 5 2 1 2 1 6 2 3 1 3 1 6 3 2 1 4 3 2 1 1 2 4 7 1 2 1 4 5 5 2 1 2 1 5 2 2 3 7 2 2 1 1 3 1 4 4

2003 2000 2003 1997 1996 1998 2004 1995 1994 2003 2002 1993 1999 2001 1999 1997 2001 2003 2000 2003 2000 1999 1994 1994 2001 1999 1999 1995 2005 2000 1996 2002 2002 2004 2001 1998 2003 1998 2000 1996 2005 2000 1998 1997 1998 1998 1998 2001 2001 1997 1997 2000

1991 1991 1991 1991 1991 1992 1991 1992 1992 1991 1992 1992 1992 1992 1992 1992 1991 1991 1992 1993 1992 1991 1992 1992 1993 1993 1992 1993 1991 1993 1993 1993 1993 1991 1993 1991 1993 1993 1993 1993 1991 1993 1993 1992 1993 1992 1993 1993 1992 1993 1992 1993

a b c d e f g

MN/1000 BN

6.5 3.5 2 2.5 1 6.5 5.5 0 1 6.5 3.5 5.5 2 3 1 1.5 9.5 2.5 9.5 1.5 1 9.5 5.5 6 1 11.5 9.5 2 6 11.5 8 0.5 7 6 10 4 10 10 7.5 2 1 1.5 4.5 4 6 4.5 1 0 5

Occupatione

Smoking statusf

No. of cigarettes

Drinking statusg

0 4 4 1 5 1 0 1 1 5 1 1 2 1 1 1 1 0 1 5 1 1 1 1 4 4 1 3 0 0 1 4 1 0 4 3 1 1 1 4 0 1 1 0 3 1 1 3 5 1 0 4

0 0 1 0 1 1 1 1 0 1 1 0 1 0 0 1 1 0 0 1 0 0 0 0 0 1 0 1 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0

4 20 40 60 6 10 30 30 55 20 20 0 20 0 0 10 10 10 0 25 20 20 12 25 20 20 20 13 13 0 20 0 0 0 0 0 0 20 10 10 30 20 20 25 0 35 0 0 15 40 0 0

1 1 0 1 1 1 1 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 1 0 0 1 0 0 1 1 0 0 1 1 1 0

Age at the time of sampling. 1: male and 2: female. 1: Cascina-Navacchio and 2: Pisa. 1: digestive, 2: respiratory, 3: genito-urinary, 4: breast, 5: lymphoma, 6: encephalon and 7: oral cavity. 0: not known, 1: housewives, 2: student, 3: retired, 4: blue collars and 5: white collars. 0: never smoker or ex-smoker >10 years and 1: smoker. 0: abstainer and 1: drinker.

E. Murgia et al. / Mutation Research 639 (2008) 27–34

31

Table 4 MN frequency (mean ± S.D.) in cancer cases by tumour sites and comparison to the distribution of cancer reported by the Tuscany Cancer Registry (standardized mortality rates × 100,000 inhabitants; 2003)

point and to allow multivariate analysis of results, subjects were stratified in two classes, i.e. low (≤2.5 MN/1000 BN cells) and high (>2.5 MN/1000 BN cells). The role of covariates as confounders or effect modifiers was assessed by a unconditional logistic regression analysis. Odds ratio (OR) and 95% confidence intervals (95% CI) were adjusted by age (continuous variable), gender (male/female), occupation (housewives, retired, white and blue collars), smoking (smoker/never smoker), and drinking habits (drinker/abstainer). The Kaplan–Meier (Logrank test) analysis was applied to evaluate the survival probability of each class based on MN frequency. Statgraphics plus 5.1 and STATA packages were used for statistical analysis.

Tumour site

MN/1000 BN (mean ± S.D.)

Digestive + oral cavity Respiratory Genito-urinary Breast Lymphoma Encephalon

5.0 5.1 3.8 5.2 5.3 4.0

± ± ± ± ± ±

4.5 3.5 3.0 4.7 6.7 3.5

N

%

Tuscany (%)

19 14 7 5 2 2

38.78 28.57 14.29 10.20 4.08 4.08

45.45 26.03 15.33 7.41 3.29 2.48

cer cases was from 0 MN/1000 BN to 11.5 MN/1000 BN. The distribution of cancer cases by tumour site is similar to the regional pattern reported by the regional Tuscany Cancer Registry (p = 0.86) [23], and the distribution of mean MN frequency did not differ significantly by cancer site (p = 0.91), as shown in Table 4. Table 5 shows sensibility and specificity for the various cut-off values of MN frequency. For subsequent analyses, cases and controls were stratified in two classes according to ROC analysis and unconditional logistic regression analysis was performed considering as potential confounders age, gender, occupation, smoking and drinking habits. The model was highly significant (p < 0.0001) and individuals with high MN level had a significantly increased cancer death risk (OR = 10.7; 95% CI = 4.6–24.9; p < 0.0001) as compared to individuals with low MN level. Ageing was associated with a 6% increased risk per year (OR = 1.06; 95% CI = 1.01–1.12,

3. Results The follow-up of the previously studied population [21] identified 52 subjects deceased for cancer and 59 for other causes, 39 of them for cardiovascular diseases [22]. One hundred and one age- and gender-matched controls were selected within the same cohort. Demographic and lifestyle characteristics of cases and controls were not statistically different regarding age, gender, occupation, smoking and drinking habits (Table 1); Table 2 shows the demographic and lifestyle characteristics of 52 cancer cases. In Table 3 we report MN mean values (range) for the different sub-groups of cases population examined. Univariate analysis showed a significantly higher MN frequency in the cases in comparison with the controls (4.7 ± 3.4 MN/1000 BN cells ± S.D. vs. 1.5 ± 1.7; p < 0.0001). The range of MN frequency observed in canTable 3 MN mean value (range) for the sub-groups examined Data at time of blood sampling

MN mean value (range) Cancer cases

Healthy controls

Gender Males Females

4.1 ± 3.1 (0–11.5) 5.6 ± 3.5 (0–11.5)

1.4 ± 1.8 (0–10) 1.7 ± 1.7 (0–6.5)

Smoking status Smoker Never smoker

4.4 ± 3.9 (0–11.5) 4.8 ± 3.3 (0–11.5)

1.6 ± 1.5 (0–5.5) 1.5 ± 1.8 (0–10)

Drinking status Drinker Abstainer

4.6 ± 3.6 (0–11.5) 4.8 ± 3.0 (1–10)

1.6 ± 1.6 (0–10) 1.3 ± 1.9 (0–7.5)

Occupation Housewives Retired White collars Blue collars Not known

4.6 2.4 6.5 6.1 4.0

± ± ± ± ±

3.1 (1–10) 1.8 (0–4.5) 4.1 (1.5–11.5) 3.5 (6.5–11.5) 4.5 (0–10)

0.9 2.3 1.8 1.6 1.5

± ± ± ± ±

0.8 (0–3.5) 2.4 (0–7.5) 1.2 (0–4.5) 2.1 (0–6.5) 1.9 (0–10)

32

E. Murgia et al. / Mutation Research 639 (2008) 27–34

Table 5 Sensibility and specificity for the various cut-off values of MN frequency Area under the ROC curve Standard error 95% confidence interval Significance level p (area = 0.5)

0.800 0.042 0.727–0.861 0.0001

Criterion

Sensitivity

95% CI

Specificity

95% CI

+LR

≥0 >0 >0.5 >1 >1.5 >2 >2.5* >3 >3.5 >4 >4.5 >5 >5.5 >6 >6.5 >7 >7.5 >8 >9.5 >10 >11.5

100.00 95.92 93.88 79.59 73.47 65.31 61.22 59.18 55.10 51.02 46.94 44.90 38.78 30.61 24.49 22.45 20.41 18.37 10.20 4.08 0.00

92.7–100.0 86.0–99.4 83.1–98.6 65.7–89.7 58.9–85.0 50.4–78.3 46.2–74.8 44.2–73.0 40.2–69.3 36.3–65.6 32.5–61.7 30.7–59.8 25.2–53.8 18.3–45.4 13.4–38.9 11.8–36.6 10.3–34.3 8.8–32.0 3.4–22.2 0.6–14.0 0.0–7.3

0.00 23.76 37.62 56.44 69.31 79.21 86.14 88.12 92.08 93.07 95.05 95.05 96.04 96.04 98.02 98.02 99.01 99.01 99.01 100.00 100.00

0.0–3.6 15.9–33.3 28.2–47.8 46.2–66.3 59.3–78.1 70.0–86.6 77.8–92.2 80.2–93.7 85.0–96.5 86.2–97.2 88.8–98.4 88.8–98.4 90.2–98.9 90.2–98.9 93.0–99.7 93.0–99.7 94.6–99.8 94.6–99.8 94.6–99.8 96.4–100.0 96.4–100.0

1.00 1.26 1.51 1.83 2.39 3.14 4.42 4.98 6.96 7.36 9.48 9.07 9.79 7.73 12.37 11.34 20.61 18.55 10.31

−LR 0.17 0.16 0.36 0.38 0.44 0.45 0.46 0.49 0.53 0.56 0.58 0.64 0.72 0.77 0.79 0.80 0.82 0.91 0.96 1.00

+LR: positive likelihood ratio and −LR: negative likelihood ratio. The area under the curve (AUC) is a measure for the appropriateness of a test. An area of 0.8 (p < 0.001) means that a randomly selected individual from the cancer group has a test value (MN frequency) larger than that for a randomly chosen individual from the control group in a 80% of the time. * The criterion value indicates the highest accuracy (minimal false negative and false positive results).

p = 0.03). No influence of other potential confounders was observed. The Kaplan–Meier curve shows the survival probability for subjects, grouped according to MN classes observed at the time of the recruitment (Fig. 1). At the time of the recruitment, all subjects resulted as clinically disease free. However, we cannot exclude that some donors could be affected by some not yet diagnosed neoplastic disease that may cause the produc-

tion of clastogenic factors and consequently an increased level of MN. In order to explore this hypothesis, cases were distributed in tertiles according to the time elapsed between blood sampling and cancer deaths (0–5, 6–8 and 9–14 years) and the distribution of cases and controls in function of the MN classes is shown in Table 6. No statistically significant difference of MN frequencies was observed between the three groups (p = 0.90), as shown in Fig. 2, suggesting that the high MN level of cancer cases is independent from the time of disease onset.

Table 6 Distribution of cases and controls in two classes of MN frequency stratified by follow-up length MN/1000 BN

Fig. 1. Kaplan–Meier curve showing cumulative rates of survival in cancer cases and healthy controls, stratified into two groups according to the level of MN/1000 BN.

Low High

Cancer cases (n)

Controls (n)

0–5

6–8

9–14

0–14

6 9

7 11

9 10

87 14

0–5, 6–8, 9–14 and 0–14 denote years of follow-up.

E. Murgia et al. / Mutation Research 639 (2008) 27–34

Fig. 2. Box and whisker plot presenting MN frequency in cases (grouped in tertiles according to the years elapsed from donors recruitment) and in controls. Horizontal line within the box: median, circle: mean and squared symbols: outliers.

4. Discussion In the present paper, a significantly higher MN frequency in PBL was found in disease-free individuals who developed cancer in the course of 14 years after blood sampling (cases) as compared with those who were still cancer free at the end of the follow-up period (controls) (4.7 ± 3.4 MN/1000 BN cells vs. 1.5 ± 1.7 MN/1000 BN cells). Therefore, this nested case–control provides strong evidence of the predictive value of MN on cancer death risk for the first time in a large sample of disease-free subjects, drawn from the general population. The MN frequency resulted as strongly predictive with a sensitivity as high as 80% (95% CI 69–91%) which likely relies on the high homogeneity of this study for many technical aspects, regarding subjects selection, blood collection and cell culturing, slide processing and MN scoring. The association between higher MN levels and cancer death risk was not confined to sitespecific cancers: actually MN increase ranged between 2.5 and 3.5 times, over the control level, for all cancer sites. In general terms, the use of a biomarker as predictor of disease risk, in a population at risk for exposure to genotoxic agents or for genetic susceptibility, is supported by the relationship observed between risk factor and biomarker, which will translate into a similar relationship between risk factor and related disease [24]. Thus, the conceptual basis for the use of MN as biomarker relies on the fact that several cytogenetic abnormalities are found in cancer cells, supporting the hypothesis that chromosome damage is directly involved in cancer etiology [25]. Confirmation of this assumption is crucial for the validation of MN as predictive of cancer [9]. Although a large amount of data has been published on the association between CA and neoplastic diseases,

33

a specific association between MN frequency and cancer risk could only be inferred from studies on cancer patients and on professionally exposed workers [9]. The MN increase might be in the former ones consequent to the disease onset, whilst in the latter ones causally related mainly to genotoxins present in the work place. In our study the time elapsed between blood sampling and cancer death did not influence the increased risk in individuals with high MN level, therefore the hypothesis that some undiagnosed neoplastic disease could have caused a MN increase even 14 years before can be ruled out. On the other hand, such increase does not seem to be due specific exposure to genotoxic agents on the working place, as it can be expected since high standards of health protection in working places have been introduced in Tuscany since 1960–1970s. The increase of MN observed in individuals extracted from the general population more likely reflects constitutive genome instability and/or exposure to endogenous or widely spread genotoxins rather than the consequence of specific lifestyle factors or professional exposure. An alternative possibility is that increased MN frequency may be caused to dietary deficiency of key micronutrients required for genome maintenance which include folate and vitamin B12 [26–28]. Actually we have considered the potential confounding or modifying effect on MN frequencies of several factors, including age, gender and lifestyle, such as occupation, smoking and drinking habits, which were described in many studies [29–30]. In our sample, no significant effect on cancer death risk was observed for any factor except for ageing. The increased risk associated to age (6% per year) is very relevant, since a doubling of cancer risk can be expected in some 15–18 years. In conclusion, this nested case–control study based on a longitudinal cohort study allows for the first time supporting the predictive value of MN frequency as biomarker for cancer risk in general population. Nonetheless, these results will be further validated by continuing this follow-up in the forecoming years considering not only cancer mortality but also morbidity as well as susceptibility to other degenerative diseases such as cardiovascular and Alzheimer disease both of which are associated with increased MN frequency in lymphocytes [31,32]. Acknowledgments This research was supported by grants funded by Associazione Italiana per la Ricerca sul Cancro (AIRC), Italy, and by the EU (Cancer Risk Biomarkers).

34

E. Murgia et al. / Mutation Research 639 (2008) 27–34

References [1] R.N. Anderson, B.L. Smith, Deaths: leading causes for 2002, Natl. Vital Stat. Rep. 53 (17) (2005) 1–89. [2] World Health Organization (WHO), Mortality Database Table 4, 2002, http://www.who.int/. [3] L. Hagmar, S. Bonassi, U. Stromberg, et al., Cancer predictive value of cytogenetic markers used in occupational health surveillance programs: a report from an ongoing study by the European Study Group on Cytogenetic Biomarkers and Health, Mutat. Res. (1998) 171–178. [4] P. Rossner, P. Boffetta, M. Ceppi, et al., Chromosomal aberration in lymphocytes of healthy subjects and risk of cancer, Environ. Health Perspect. (2005) 517–520. [5] P. Boffetta, O. van der Hel, H. Norppa, et al., Chromosomal aberrations and cancer risk: results of a cohort study from Central Europe, Am. J. Epidemiol. (2007) 36–43. [6] M. Fenech, Cytokinesis-block micronucleus assay evolves into a “cytome” assay of chromosomal instability, mitotic dysfunction and cell death, Mutat. Res. (2006) 58–66. [7] P.I. Countryman, J.A. Heddle, The production of micronuclei from chromosome aberrations in irradiated cultures of human lymphocytes, Mutat. Res. (1976) 321–332. [8] M. Fenech, A.A. Morley, Measurement of micronuclei in lymphocytes, Mutat. Res. (1985) 29–36. [9] M. Fenech, N. Holland, W.P. Chang, E. Zeiger, S. Bonassi, The HUman MicroNucleus Project: an international collaborative study on the use of the micronucleus technique for measuring DNA damage in humans, Mutat. Res. (1999) 271–283. [10] F. Duffaud, T. Orsiere, P. Villani, et al., Comparison between micronucleated lymphocyte rates observed in healthy subjects and cancer patients, Mutagenesis (1997) 227–231. [11] M.P. Rosin, J. German, Evidence for chromosome instability in vivo in Bloom syndrome: increased numbers of micronuclei in exfoliated cells, Hum. Genet. (1985) 187–191. [12] S.E. Benner, S.M. Lippman, M.J. Wargovich, Micronuclei, a biomarker for chemoprevention trials: results of a randomized study in oral pre-malignancy, Int. J. Cancer (1994) 457–459. [13] M. Sorsa, J. Wilbourn, H. Vainio, Human cytogenetic damage as a predictor of cancer risk, in: H. Vainio, P.N. Magee, D.B. McGregor, A.J. McMichael (Eds.), Mechanisms of Carcinogenesis in Risk Identification, IARC Scientific Publications, 1992, pp. 543–554. [14] M. Weichental, M. Roser, U. Ehlert, S. Frenzer, E.W. Breitbart, H.W. Rudiger, Increased numbers of spontaneous micronuclei in blood lymphocytes and cultured fibroblasts of individuals with familial cutaneous malignant melanoma, J. Cancer Res. Clin. Oncol. (1989) 264–268. [15] T. Cheng, D.C. Christiani, X. Xiping, J.C. Wain, J.K. Wiencke, K.T. Kelsey, Increased micronucleus frequency in lymphocytes from smokers with lung cancer, Mutat. Res. (1996) 43–50. [16] C.H. Leal-Garza, R.M. Cerda-Flores, E. Leal-Elizondo, E.I. Cortes-Gutierrez, Micronuclei in cervical smears and peripheral blood lymphocytes from women with and without cervical uterine cancer, Mutat. Res. (2002) 57–62. [17] C. Bolognesi, F. Martini, M. Tognin, et al., A molecular epidemiology case–control study on pleural malignant mesothelioma, Cancer Epidemiol. Biomarkers Prev. (2005) 1741–1746.

[18] R.A. El-Zein, et al., Cytokinesis-blocked micronucleus assay as a novel biomarker for lung cancer risk, Cancer Res. 66 (2006) 6449–6456. [19] C. Bolognesi, A. Abbondandolo, R. Barale, et al., Age-related increase of chromosome aberrations, sister chromatid exchanges and micronuclei in human lymphocytes, Cancer Epidemiol. Biomarkers Prev. (1997) 249–256. [20] S. Bonassi, A. Znaor, M. Ceppi, et al., An increased micronucleus frequency in peripheral blood lymphocytes predicts the risk of cancer in humans, Carcinogenesis (2007) 625–631. [21] R. Barale, L. Chelotti, T. Davini, et al., Sister chromatid exchange and micronucleus frequency in human lymphocytes of 1650 subjects in an Italian population. II. Contribution of sex, age, and lifestyle, Environ. Mol. Mutat. (1998) 228–242. [22] E. Murgia, V. Maggini, R. Barale, A.M. Rossi, Micronuclei, genetic polymorphisms and cardiovascular disease mortality in a nested case–control study in Italy, Mutat. Res. (2007) 113–118. [23] Tuscany Cancer Registry, 2003, http://www.cspo.it/registry. [24] A. Schatzkin, L.S. Freedman, M.H. Schiffman, S.M. Dawsey, Validation of intermediate end points in cancer research, J. Natl. Cancer Inst. (1990) 1746–1752. [25] F. Mitelman, F. Mertona, B. Johansson, A breakpoint map of recurrent chromosomal rearrangements in human neoplasia, Nat. Genet. (1997) 417–474. [26] J.T. MacGregor, C.M. Wehr, R.A. Hiatt, B. Peters, J.D. Tucker, R.G. Langlois, R.A. Jacob, R.H. Jensen, J.W. Yager, M.K. Shigenaga, B. Frei, B.P. Eynon, B.N. Ames, ‘Spontaneous’ genetic damage in man: evaluation of interindividual variability,relationship among markers of damage, and influence of nutritional status, Mutat. Res. (1997) 125–135. [27] M. Fenech, P. Baghurst, W. Luderer, J. Turner, S. Record, M. Ceppi, S. Bonassi, Low intake of calcium, folate, nicotinic acid, vitamin E, retinol, beta-carotene and high intake of pantothenic acid, biotin and riboflavin are significantly associated with increased genome instability—results from a dietary intake and micronucleus index survey in South Australia, Carcinogenesis (2005) 991–999. [28] N. Botto, M.G. Andreassi, S. Manfredi, S. Masetti, F. Cocci, M.G. Colombo, S. Storti, A. Rizza, A. Biagini, Genetic polymorphisms in folate and homocysteine metabolism as risk factors for DNA damage, Eur. J. Hum. Genet. (2003) 671–678. [29] Nordic Study Group on the Health Risk of Chromosome Damage, A Nordic database on somatic chromosome damage in humans, Mutat. Res. (1990) 325–337. [30] S. Bonassi, C. Bolognesi, A. Abbondandolo, et al., Influence of sex on cytogenetic endpoints: evidence from a large human sample and review of the literature, Cancer Epidemiol. Biomarkers Prev. (1995) 671–679. [31] M.G. Andreassi, N. Botto, DNA damage as a new emerging risk factor in atherosclerosis, Trends Cardiovasc. Med. (2003) 270–275. [32] L. Migliore, A. Testa, R. Scarpato, N. Pavese, L. Petrozzi, U. Bonuccelli, Spontaneous and induced aneuploidy in peripheral blood lymphocytes of patients with Alzheimer’s disease, Hum. Genet. (1997) 299–305.