Conceived and designed the experiments: MA RMP CR MH AR JH HB AVB. Performed the experiments: MA CR AR JH. Analyzed the data: MA RMP AVB. Contributed reagents/materials/analysis tools: TIB MDT AJS KM VM. Wrote the paper: MA RMP AVB MH.
The authors herewith declare that one of the co-authors being recently employed by NOVA Research Company does not alter the authors’ adherence to all the PLoS ONE policies on sharing data and materials. The coauthor was working at Radiation Epidemiology Branch/National Cancer Institute (REB/NCI) before she changed to NOVA Research Company where she is employed now. At the time of Radiation Epidemiology Branch/National Cancer Institute (REB/NCI) she did her contribution.
The strong and consistent relationship between irradiation at a young age and subsequent thyroid cancer provides an excellent model for studying radiation carcinogenesis in humans. We thus evaluated differential gene expression in thyroid tissue in relation to iodine-131 (I-131) doses received from the Chernobyl accident. Sixty three of 104 papillary thyroid cancers diagnosed between 1998 and 2008 in the Ukrainian-American cohort with individual I-131 thyroid dose estimates had paired RNA specimens from fresh frozen tumor (T) and normal (N) tissue provided by the Chernobyl Tissue Bank and satisfied quality control criteria. We first hybridized 32 randomly allocated RNA specimen pairs (T/N) on 64 whole genome microarrays (Agilent, 4×44 K). Associations of differential gene expression (log2(T/N)) with dose were assessed using Kruskall-Wallis and trend tests in linear mixed regression models. While none of the genes withstood correction for the false discovery rate, we selected 75 genes with
One of the most important health consequences of the 1986 Chernobyl nuclear power plant accident is a dramatic increase in thyroid cancer incidence among those who were children or adolescents at the time
The study of radiation carcinogenesis in humans can take advantage of instances where relationships are strong and consistent, as in the situation where the thyroid gland is irradiated at a young age. The opportunity for molecular research of radiation-related thyroid cancer is facilitated by the resources of the Chernobyl Tissue Bank (CTB), which systematically collects biological samples from patients with Chernobyl-related thyroid pathology
To improve understanding of the molecular consequences of I-131 exposure, we evaluated for the first time differential gene expression in thyroid tissue, defined as a difference in gene expression levels between tumor and corresponding normal thyroid tissue, in relation to individual I-131 thyroid dose estimates. We hypothesized that if dose-related gene expression patterns in tumor tissue truly reflect an important event in radiation carcinogenesis rather than a long-lasting effect of radiation exposure, they should differ from patterns observed in normal tissue; hence our approach of analyzing differential dose-expression relationships in tumor relative to paired normal samples. Our study used RNA specimens from the CTB of patients who underwent thyroid surgery for thyroid cancer in the Ukrainian-American cohort study composed of approximately 13,000 Ukrainian residents <18 years at the time of the accident with individual radioactivity measurements taken within two months after the accident
In the current study, we first conducted an initial screen in half of the cases to identify promising gene candidates that were differentially expressed in tumor and normal tissue specimens in relation to I-131 dose based on whole genome RNA microarrays (phase I). We then validated the top candidates in tumor and normal tissue specimens from the second half of the cases using qRT-PCR (phase II). For the validated candidates we additionally characterized the relationship of gene expression separately in tumor and normal tissue.
As a result of four sequential screening examinations, 110 prevalent and incident thyroid carcinomas were diagnosed in the Ukrainian-American cohort between 1998 and 2008 at the Laboratory of Morphology of Endocrine System of the Institute of Endocrinology and Metabolism (IEM, Kiev, Ukraine)
Detailed operating procedures for the collection, documentation, and processing of frozen tumor and normal thyroid tissue samples are available from the CTB website
Dosimetric methods have been described elsewhere
Full details of RNA extraction can be obtained from the CTB website
Although the CTB provided 137 RNA specimens for 71 individuals with PTC, we were able to use 126 paired (tumor/normal) RNA specimens corresponding to 63 individuals (
Genome wide expression profiling was carried out using the Agilent oligo microarray (4×44 K format) combined with a one-color based hybridization protocol on 64 paired RNA specimens from 32 randomly selected individuals (half the sample set). To assure that individuals selected for microarray gene expression analysis (n = 32) and those remaining for validation analysis by qRT-PCR (n = 31) were similar, we confirmed that their distributions of sex, age, place of residence, and I-131 dose were similar.
Chemicals, kits, and software for the whole genome microarray assay were provided by Agilent Technologies, Waldbronn, Germany. 500 ng of total RNA per specimen (spiked with an internal labeling control) were introduced into an RT-IVT reaction and cDNA was then converted into labeled cRNA by in-vitro transcription step (one color Quick-Amp labeling kit). Quality of labeled non-fragmented cRNA was analyzed (RNA 6000 Nano LabChip kit), and the cRNA was cleaned and quantified (NanoDrop ND-1000 Spetralphotometer). Finally, 1.65 µg of each labeled cRNA sample was fragmented and prepared for one-color based hybridization (gene expression hybridization kit). Hybridization occurred at 65°C over 17 hours on separate microarrays consisting of 41,000 target gene-specific probes (∼20,000 genes) and thousands of control probes. After three washes with increasing stringency, fluorescent signal intensities were detected on the Agilent DNA microarray scanner and extracted from the images using Agilent extraction software. Quantile normalization was applied to the data.
We analyzed differential gene expression in tumor compared to normal tissue, obtained by subtracting log2 transformed probe signals of the normal tissue (N) from the corresponding tumor tissue (T), log2(T)-log2(N), in relation to I-131 dose estimates. This approach reduces variability in the gene expression. We used the non-parametric Kruskall-Wallis test (P kruskall) to compare differential gene expression across three dose categories (≤0.30, 0.31–1.0, >1.0 Gy) with cut-off points approximately corresponding to the tertiles of dose distribution among cases, and linear regression models with trend test (P linear) for continuous dose. Only those gene transcripts that had a call “present” in at least 50% of RNA specimens from tumor and normal tissue were included in the analysis of differential gene expression (∼15,000). We corrected for multiple comparisons using the false discovery rate (FDR)
To validate our microarray findings, we evaluated gene expression by qRT-PCR (TaqMan primer probe assays) on 62 paired RNA specimens from the remaining 31 individuals. All chemicals for qRT-PCR using TaqMan chemistry were provided by Applied Biosystems, Darmstadt, Germany. Due to technical reasons individual primer probe assays were either run on a 96-well qRT-PCR platform or using another platform called low density array (LDA). Twenty four of the 75 genes were measured in duplicates, because of available space on the platforms and in order to improve the statistic. A 0.75 µg RNA aliquot of each individual was reverse transcribed using a two-step PCR protocol (High Capacity Kit). 50 µl cDNA (equivalent to about 0.25 µg RNA) was mixed with 50 µl 2 x RT-PCR master mix and pipetted into 2 of 8 fill ports of the LDA. Cards were centrifuged twice (1,200 rpm, 1 min, Multifuge3S-R, Heraeus, Germany), sealed, and transferred into the 7900 qRT-PCR instrument. The qRT-PCR was run for two hours following the qRT-PCR protocol for 384-well LDA format. Taqman chemistry for the 96-well platform was used similarly except the volume per reaction was adjusted to 20 µl. All technical procedures for qRT-PCR were performed in accordance with standard operating procedures implemented in our laboratory in 2008 when the Bundeswehr Institute of Radiobiology became accredited according to DIN EN ISO 9001/2008.
We ran four RNA specimens in triplicate on three different LDAs to establish an upper limit of the linear-dynamic range of the threshold cycles (CT). The upper limit of the CT was 30, so we used only CT values
Both methods measured on 32 individuals included in phase I revealed comparable results in 70.6% (
To confirm phase I findings, the phase II analyses used only individuals (n = 31) not included in phase I. After a power or log transformation and/or removal of outliers for selected genes, normalized CT values of all genes were normally distributed. We first computed residuals from standard linear models fitted to gene expression values
The models were
Of 63 cases in our study, 56% were female and 54% were residents of the Chernigov oblast (
Characteristic | N or Mean ± SD | % or range |
Gender, female | 35 | 56 |
Oblast of residence |
||
Zhytomyr | 15 | 24 |
Kiev | 14 | 22 |
Chernigov | 34 | 54 |
Age at exposure, year |
7.9±4.6 | 0−<18 |
Latency, year |
16.5±2.7 | 12.5–21.6 |
I-131 thyroid dose, Gy | 1.25±1.68 | 0.008–8.6 |
0.008–0.30 | 23 | 36 |
0.31–1.0 | 23 | 36 |
1.1–8.6 | 27 | 43 |
Tumor size, mm |
16.0±7.8 | 6.0–45.0 |
Histological subtype of PTC | ||
solid | 5 | 8 |
papillary | 12 | 19 |
follicular | 16 | 25 |
mixed | 30 | 48 |
At the time of first screening examination.
On April 26, 1986.
Differences between surgery date and April 26, 1986.
Largest dimension at pathomorphology.
NB: Not all percentages sum to 100 due to rounding.
Of 19,596 gene mRNAs (41,079 transcripts) spotted on the whole genome microarray, on average 73.4% (range: 63.3%–91.0%) were distinguishable from background (expressed). The total number of gene transcripts differentially expressed in relation to I-131 dose was 2,500; of these we selected 75 gene candidates for validation by qRT-PCR (
Of 75 genes assayed, 15 developed no amplification plots, 5 yielded results in
fold-change per dose category |
|||||||
Gene | Cytoband | CNV |
N |
1 | 2 | P |
P trend |
|
2q22.3 | no | 31 | 1.1 | 0.9 | 0.001 | 0.02 |
|
1p36.32 | amplified | 20 | 1.1 | 2.4 | 0.06 | 0.03 |
|
15q22.2 | no | 31 | 0.3 | 0.3 | 0.01 | 0.02 |
|
17q12 | no | 31 | 0.7 | 0.7 | 0.01 | 0.04 |
|
16q24.3 | amplified | 31 | 0.8 | 1.3 | 0.0004 | 0.04 |
|
4q31.1 | no | 31 | 0.8 | 0.6 | 0.06 | 0.02 |
|
12p12.3 | amplified | 31 | 0.9 | 0.7 | 0.01 | 0.01 |
|
14q32.33 | amplified | 31 | 1.3 | 1.4 | 0.01 | 0.002 |
|
21q22.3 | amplified | 31 | 0.7 | 1.2 | 0.002 | 0.13 |
|
11q12.1 | amplified | 31 | 2.5 | 1.1 | 0.01 | 0.99 |
|
19p12 | no | 31 | 1.3 | 1.0 | 0.0001 | 0.74 |
CNV, Copy number variation as reported in Stein et al. with either amplified regions or not (no) is shown for each cytoband where our candidate genes are located.
Number of paired (tumor/normal tissue) observations.
Columns with subtitles 1 and 2 refer to dose categories (1 and 2) and reflect the fold change in differential gene expression for a specific dose category relative to the referent dose category (0). Fold change in expression associated with dose were computed as two to the power of the difference in the slopes, i.e. 2ˆ(dosetumor−dosenormal).
Two degree of freedom test in differential dose-response.
One degree of freedom trend test in differential dose-response.
All models of differential dose-response were adjusted for tissue type, attained age, sex, and oblast of residence.
Note: Mean of residual gene expression after removing the effects of age, oblast, and sex is plotted separately for normal tissue (left part of the graph) and tumor tissue (right part of the graph) against the mean of three I-131 dose categories (0.11, 0.57, 2.62 Gy). Circles with grey fills correspond to mean gene expression values for normal tissue and squares with black fills correspond to mean gene expression values for tumor tissue. Error bars represent 95% confidence intervals. P-values for association with dose are based on a 2 degree of freedom test and a 1 degree of freedom trend test, respectively, and given separately for normal and tumor tissue in the bottom of each panel.
Note: Mean of residual gene expression after removing the effects of age, oblast, and sex is plotted separately for normal tissue (left part of the graph) and tumor tissue (right part of the graph) against the mean of three I-131 dose categories (0.11, 0.57, 2.62 Gy). Circles with grey fills correspond to mean gene expression values for normal tissue and squares with black fills correspond to mean gene expression values for tumor tissue. Error bars represent 95% confidence intervals. P-values for association with dose are based on a 2 degree of freedom test and a 1 degree of freedom trend test, respectively, and given separately for normal and tumor tissue in the bottom of each panel.
Because the relationship between irradiation at a young age and risk of thyroid cancer is strong and strikingly consistent, this tumor provides an excellent model for studying radiation carcinogenesis in humans. Here, we employed measurement-based individual I-131 doses estimated for a cohort of Ukrainian residents who were <18 at the time of the Chernobyl accident and RNA specimens from fresh frozen thyroid tissue provided by the CTB. For the first time, we conducted analyses of dose-dependent gene expression in papillary thyroid carcinoma relative to normal thyroid tissue from the same individuals across the entire genome and confirmed findings for 11 genes by qRT-PCR in RNA specimens from a separate set of cases.
Several points should be considered when comparing our results to prior studies. Transcriptional profiling using microarrays that allows simultaneous evaluation of thousands of gene transcripts has been extensively used to gain insights into the molecular changes induced by ionizing radiation
The 11 genes that were validated in an independent case set by qRT-PCR in our study (
Several clues emerged from our study that could guide future studies of radiation carcinogenesis. Dose-dependency for differential gene expression detected years after exposure likely represents a late and/or long lasting effect of radiation. One mechanism by which radiation-induced changes could be sustained over time is through inheritance of DNA damage
Our study has several unique strengths. First, we used individual I-131 dose estimates based on radioactivity measurements taken shortly after the accident
There are also several limitations to be borne in mind when interpreting our results. We did not take into account of the impact of uncertainties in dose estimates, 95% of which are typically attributable to unknown thyroid gland mass and I-131 content in the thyroid gland in 1986
In summary, our study is among the first to provide direct human data on long term gene expression in relation to measurement-based individual I-131 doses. By studying PTCs arising after the Chernobyl accident, we identified 11 genes that exhibited evidence of dose-dependent expression in cancerous relative to normal thyroid tissue and, therefore, potentially important in radiation carcinogenesis. Our study also serves as a basis for further dose-dependent studies of gene expression, CNAs, and epigenetic changes and their role in radiation susceptibility and carcinogenesis.
(TIF)
(TIF)
(TIF)
(DOC)
(DOC)
We are very thankful for the different dosimetry teams in particular to Ilya Aronovich Likhtarev, Lina Kovgan and Andre Bouville. We appreciate the careful and very accurate technical assistance by Sven Senf in generation of qRT-PCR data. We are grateful to Dr. Geraldine Thomas (Imperial College, London, UK) for her valuable assistance and support of the study.