Conceived and designed the experiments: VUB CH GPR. Performed the experiments: VUB OH. Analyzed the data: VUB OH GWD ERB GPR CH. Contributed reagents/materials/analysis tools: VUB OH GWD MM GPR CH. Wrote the paper: VUB GWD ERB GPR CH. Revisions and approval of final manuscript: VUB OH GWD ERB MM GPR CH.
The authors have declared that no competing interests exist.
The identification of a blood-based diagnostic marker is a goal in many areas of medicine, including the early diagnosis of prostate cancer. We describe the use of averaged differential display as an efficient mechanism for biomarker discovery in whole blood RNA. The process of averaging reduces the problem of clinical heterogeneity while simultaneously minimizing sample handling.
RNA was isolated from the blood of prostate cancer patients and healthy controls. Samples were pooled and subjected to the averaged differential display process. Transcripts present at different levels between patients and controls were purified and sequenced for identification. Transcript levels in the blood of prostate cancer patients and controls were verified by quantitative RT-PCR. Means were compared using a t-test and a receiver-operating curve was generated. The Ring finger protein 19A (RNF19A) transcript was identified as having higher levels in prostate cancer patients compared to healthy men through the averaged differential display process. Quantitative RT-PCR analysis confirmed a more than 2-fold higher level of RNF19A mRNA levels in the blood of patients with prostate cancer than in healthy controls (p = 0.0066). The accuracy of distinguishing cancer patients from healthy men using RNF19A mRNA levels in blood as determined by the area under the receiving operator curve was 0.727.
Averaged differential display offers a simplified approach for the comprehensive screening of body fluids, such as blood, to identify biomarkers in patients with prostate cancer. Furthermore, this proof-of-concept study warrants further analysis of RNF19A as a clinically relevant biomarker for prostate cancer detection.
While metastatic prostate cancer remains an incurable disease
We describe here the usefulness of an averaged differential expression (ADE) approach to identify biomarkers in blood samples of men with prostate cancer. Differential display methodology does not require prior knowledge of RNA transcripts. Differential display thus offers an “open” system in which both known and unknown RNA transcripts associated with a disease state can be identified, a fact that distinguishes it from microarray technology
ADE retains the advantages of the differential display technique while facilitating the analysis of heterogeneous samples. The pooling of heterogeneous samples efficiently identifies transcripts that are differentially expressed in a high proportion of the samples that comprise two groups
Although ADE has typically been used to evaluate transcripts that are differentially expressed in tissue (hence, the terminology “averaged differential expression”), we adapted this technique to identify RNA transcripts that are present at different levels in whole blood RNA from two groups of patients. These transcripts are not necessarily differentially expressed in cells that are present in the blood, or even in tumor compared to normal tissue, but their presence in the blood can be used to differentiate between these two groups. In this study, we describe the use of ADE to identify an RNA transcript, RNF19A (Gene ID: 25897), that is present at higher levels in the blood of prostate cancer patients than in healthy controls. These results establish proof-of-principle that averaged differential expression (ADE) methodology can be used for the discovery of RNA markers in body fluids of men with prostate cancer and support the study of RNF19A as a potential prostate cancer biomarker.
All research involving human participants was approved by the Henry Ford Health System (HFHS) Institutional Review Board. Written informed consent was obtained for all participants and all clinical investigation was conducted according to the principles expressed in the Declaration of Helsinki.
Prostate cancer patients were identified through the Department of Urology at the Henry Ford Health System (HFHS). Healthy controls were recruited from the general population and HFHS Internal Medicine clinics. Whole blood (2.5 ml) was collected in PAXgene Blood RNA tubes (Qiagen, Valencia, CA), allowing for transport at room temperature without RNA degradation. Samples were frozen at −80°C until extraction of RNA was performed.
Blood in PAXgene Blood RNA tubes was centrifuged. The pellet was washed with RNase-free water and resuspended in Trizol reagent (Invitrogen, Carlsbad, CA). RNA was extracted according to the manufacturer’s protocols. The isolated RNA was treated with RNase-free DNase (Invitrogen, Carlsbad, CA) and the concentration of total RNA was determined using a Qubit Fluorometer (Invitrogen, Carlsbad, CA).
Total RNA from the blood of 10 patients and 10 healthy men were pooled separately, reverse transcribed, and subjected to ADE as described by Bai et al
RNA was reverse transcribed using random hexamer primers and Transcriptor Reverse Transcriptase (Roche Applied Science, Indianapolis, IN) according to the manufacturer’s protocol. After reverse transcription, RNA was subjected to qPCR on an Applied Biosystems 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA) using sequence-specific primers for RNF19A (Assay Id: Hs00968447_m1) and 18S RNA (Assay Id: Hs03928985_g1). The cycle number at which the reaction crossed a threshold fluorescence (CT) was determined for each transcript, and the level of each test gene relative to 18S RNA (reference transcript) was determined using the equation 2−ΔCT where ΔCT = CT,test – CT,ref
Means were compared using a t-test. The equality of the group variances was tested, and the Satterthwaite p-value was reported for distributions with unequal variances. The Pearson correlation coefficient was used to evaluate linear correlation. A receiver operator curve (ROC) was generated by plotting the sensitivity against the false positive rate (1 − specificity) as the discrimination of the test was varied. Area under the curve (AUC) was calculated. All statistical computations were done in SAS.
Blood samples were obtained from healthy male controls and men with localized prostate cancer prior to definitive therapy. After RNA extraction, 20 ng total RNA from each of the 10 prostate cancer patients or from each of the 10 healthy men were combined to form two separate pools of RNA (prostate cancer vs. healthy) that were then analyzed by ADE as described by Bai et al
As shown in
ADE analysis identified two RNA transcripts with levels significantly higher in whole blood RNA from prostate cancer patients (Ca) than in healthy men (H). Nucleotide sequence analysis revealed that these two transcripts shared a common sequence. This sequence had 100% homology to the nucleotide sequence in the E3-ubiquitin ligase Ring Finger Protein 19a (RNF19A) transcript and is located in the 3′ untranslated region (UTR).
After the identification of the RNF19A transcript as a potential biomarker distinguishing the blood of prostate cancer patients from the blood of healthy controls, qRT-PCR was used to measure the level of this transcript, relative to that of 18S RNA, in whole blood RNA samples from individual patients. Relative RNF19A levels were measured in 33 prostate cancer patients and 19 healthy male controls. These cohorts were independent of the patient groups used for the ADE discovery process. The control samples (median age 40 yrs, range 31–50 yrs) were taken from the general population, and as such, were not age-matched to prostate cancer cases. Baseline clinical data for the prostate cancer cohort are shown in
Median | Range | ||
Age (yrs) | 66 | 53–82 | |
PSA (ng/mL) | 6.4 | 1.3–33.4 | |
Number | Percent | ||
Gleason | |||
6 | 8 | 24% | |
7 | 18 | 55% | |
8–10 | 7 | 21% | |
T-stage | |||
cT1a | 1 | 3% | |
cT1c | 30 | 91% | |
cT2a | 1 | 3% | |
cT3a | 1 | 3% |
Characteristics of the 33 patients included in the prostate cancer validation cohort are shown. Age and baseline PSA were abstracted from the time of diagnosis. Gleason score is as recorded by the clinical genitourinary pathologist. Clinical staging of the primary tumor is per the 7th edition of the American Joint Committee on Cancer (AJCC) staging manual.
Quantitative RT-PCR was performed on whole blood RNA samples from patients with localized prostate cancer (n = 33) as well as healthy male controls (n = 19). Levels of RNF19A transcript were normalized to 18S RNA (reference gene). Normalized results are presented in box plot format, with boxes representing the 25th, 50th, and 75th percentiles and whiskers representing the 10th and 90th percentiles of the data. Outliers are also displayed. The difference between the means was statistically significant (p = 0.0066).
A receiver-operating curve for RNF19A was generated as an overall estimate of the sensitivity and specificity of this blood biomarker for distinguishing prostate cancer cases from controls (
A receiver-operating curve (ROC) was generated as a preliminary estimate of the accuracy of relative levels of RNF19A in classifying patients with cancer or healthy controls in our cohort of patients. The true positive rate (sensitivity) was plotted against the false positive rate (1-specificity). The area under the curve was calculated as 0.7273.
Although this finding is encouraging, it should be noted that using PSA to differentiate patients with prostate cancer from controls in our cohorts may have also resulted in an improved AUC given the relatively younger age of our healthy cohort. With a median age of 40, the expected PSA in our healthy cohort would be less than 1.0 ng/mL, substantially lower than the median PSA value of 6.4 ng/mL seen in our prostate cancer cohort. However, RNF19A levels did not correlate with age (Pearson coefficient of 0.09), making it unlikely that our findings are simply due to an age-related effect. Clearly, further validation needs to be performed in populations that more closely resemble the real-world setting of patients who are being referred for prostate biopsy. In addition, it remains to be determined whether RNF19A or other such biomarkers will add additional value to clinically established methods to establish prostate cancer risk such as PSA, age, race and family history. This study was not designed to address the critical question of increased detection of potentially lethal cancer while avoiding the overdiagnosis of indolent cancer. Further efforts at biomarker discovery to identify these lethal subtypes are an essential goal for the future.
In conclusion, we describe here the use of averaged differential expression to identify RNF19A as an RNA transcript that is present at significantly higher levels in the blood of prostate cancer patients compared to healthy controls. In our sample of patients, this transcript was able to distinguish prostate cancer patients from controls with an AUC of the receiver-operating curve of 0.7273. The results discussed here establish proof-of-principle that ADE methodology can be used for the discovery of RNA markers in the blood of men with prostate cancer and warrant further analysis of RNF19A as a clinically relevant biomarker for prostate cancer early detection.
(PPT)
(PPT)
Raw CT data are presented for the validation cohorts of prostate cancer patients (PrCa pts) and healthy controls. Mean CT values were lower in PrCa pts compared to controls (28.25 vs. 29.50, p = 0.02). Mean CT values for the 18S reference gene were not statistically different between patients and controls (16.62 vs 16.67, p = 0.89). Means were compared using a t-test.
(DOC)
The authors would like to thank the patients and their families who made this work possible.