Noncancerous PTGS2 DNA fragments of apoptotic origin in sera of prostate cancer patients qualify as diagnostic and prognostic indicators



Our study was designed to evaluate Cell-Free DNA in sera of prostate cancer (PCA) patients as a useful biomarker. Real-time PCR was used to amplify a <200 bp PTGS2 DNA fragment that biochemically characterizes apoptosis and a larger >250 bp Reprimo DNA fragment that defines mostly other cell death entities. The apoptosis index (AI) expresses the ratio of PTGS2 to Reprimo DNA fragments. GSTP1 hypermethylation was assessed to evaluate the amount of tumor-derived DNA. We analyzed serum of 216 patients (168 PCA; 5 incidental PCA; 42 benign prostatic hyperplasia (BPH); 11 healthy individuals). Distinctly elevated concentrations of PTGS2 fragments were detected in PCA compared to BPH and healthy individuals (median: 70.2, 10.5 and 7.1 ng/ml, respectively; p < 0.0001). The AI was significantly increased in PCA vs. BPH patients and healthy individuals (6.01 vs. 1.54 and 0.84 respectively; p = 0.002–0.0001). GSTP1 hypermethylation was only present in a small percentage (mean 1.92%) of circulating DNA. Concentrations of apoptotic PTGS2 fragments discriminated sensitively (88%) and specifically (64%) between BPH and PCA, whereas the AI was more specific (82%) but less sensitive (70%). The AI correlated with histological grading (p = 0.044). Kaplan–Meier analysis for a subset of 124 patients revealed a significant correlation between apoptotic PTGS2 fragments or the AI and PSA recurrence following radical prostatectomy (p = 0.0395–0.0482). In conclusion, circulating PTGS2 fragments of apoptotic origin and the AI are promising serum biomarkers for the diagnosis and prognosis of PCA. We suggest that cancer-induced apoptosis of peripheral noncancerous tissues is relevant in many malignancies. © 2007 Wiley-Liss, Inc.

Prostate cancer (PCA) is one of the most common malignancies in the world.1 The most common diagnostic tool is digital rectal examination, but diagnostic accuracy is very limited. Testing of serum prostate specific antigen (PSA) during the last 2 decades improved the diagnosis of PCA. Nevertheless, serum PSA has several drawbacks: (i) its specificity is limited and 60–70% of PSA elevations are caused by nonmalignant disease2; (ii) a number of high-grade tumors are overlooked as patients' PSA levels are in the normal range2; (iii) many clinically insignificant PCA are overdiagnosed.3 In summary, the development of an improved cancer indicator is essential.

In the late 1970s, cell-free DNA was detected in the plasma of cancer patients.4 The development of PCR-based methods facilitated the quantification of these extremely small amounts of DNA. It was observed that cell-free DNA concentrations are increased in cancer patients compared to healthy individuals and patients with nonmalignant disease.5 More recently, it was shown that plasma DNA levels may even correlate to disease-specific survival in patients with metastasis.6 Some reports7 demonstrate that especially smaller DNA fragments (<200 bp) are present in cancer patients whereas other groups8 reported that larger DNA fragments (>400 bp) can discriminate cancer patients from healthy individuals, and, in breast cancer patients, DNA integrity (i.e. large DNA fragments) was associated with advanced tumor stage.9 These results reflect that cell death in cancer may be caused by apoptosis (fragment length usually 185–200 bp) or other cell death entities (i.e. necrosis, autophagy) presumably depending on the specific cancer entity. Also, published data shows that usually only a fraction of cell-free DNA is derived from cancer cells or dying immune cells.10

In our study, we first constructed various primer sets (PTGS2, TIG1, GSTP1, Reprimo) to test for DNA fragments in the low apoptotic (<200 bp) and high nonapoptotic (>300 bp) DNA range by quantitative real-time PCR. Two primer sets for PTGS2 and Reprimo (124 and 271 bp, respectively) proved to be most efficient in assessing apoptotic and nonapoptotic DNA fragments. We show that the total amount of small PTGS2 fragments increases considerately in PCA patients whereas the longer Reprimo fragment levels—representing mostly nonapoptotic DNA fragmentation—remain nearly unchanged in PCA vs. BPH and healthy individuals. GSTP1 hypermethylation was only detectable in a small fraction of cell-free DNA proving a mostly noncancerous origin. Furthermore, we demonstrate that short PTGS2 DNA fragments and the apoptosis index are valuable diagnostic and prognostic indicators in PCA.


AI, apoptosis index; AUC, area under curve; bp, base pair; BPH, benign prostatic hyperplasia; GSTP1, glutathione-S-transferase pi 1; PCA prostate cancer; PSA, prostate specific antigen; PTGS2, prostaglandin-endoperoxide synthase 2; ROC, receiver operating characteristic; TIG1, tazarotene induced gene 1; TURP, transurethral resection of the prostate.

Material and methods

Patients, sample collection and DNA isolation

A total of 168 consecutive patients with PCA undergoing radical prostatectomy, 5 patients with incidental PCA diagnosed after transurethral resection of the prostate (TURP), 42 patients with benign prostatic hyperplasia (BPH) undergoing TURP with histologically verified nonmalignant prostate tissue and 11 healthy individuals were recruited into this study. Prostatectomy specimens were reviewed by 2 independent pathologists with uro-pathological experience. In cases of disagreement the cases were reviewed by a panel including a reference pathologist for the German division of the International Association of Pathologist for prostate pathology. The clinicopathological parameters of all individuals are shown in Table I. Blood samples from patients were collected before anaesthesia at the Department of Urology, University Hospital Bonn, Germany. Sixteen samples from BPH patients were obtained from the Department of Urology, St. Josef-Hospital, Troisdorf, Germany. The period between prostate biopsy and radical prostatectomy ranged from 3 to 6 weeks. An increase of cell-free DNA due to prostate biopsy at the time of radical prostatectomy is unlikely: cell-free DNA levels are increased in trauma patients, but DNA levels decrease towards normal range within 3hr in less severe injured patients.11 All blood samples were collected in serum S-Monovette Gel tubes (Sarstedt, Nürnbrecht, Germany) containing a clot activation additive and barrier gel. Clotting of serum samples was allowed for at least 60 min before centrifugation (1,800g, 10 min) and supernatants were stored at −80°C. All patients had given written informed consent according to the institutional guidelines before inclusion into the study. The QIAamp Ultrasens Virus Kit (Qiagen, Hilden, Germany) was used to isolate cell-free DNA from 2 ml serum samples. DNA isolation was performed according to the manufacturer's protocol (elution volume: 60 μl AVE-buffer).

Table I. Clinicopathological Characteristics of Patients with Prostate Cancer, Benign Prostate Hyperplasia and Healthy Individuals
 PCA n = 168 (%)PCA subset with available follow-up n = 124 (%)Incidental PCA after TURP n = 5 (%)BPH n = 42 (%)Healthy individuals n = 11 (%)
  1. n.a. not applicable; TURP, transurethral resection of the prostate.

Age (years)
Prostate weight (gram)
Preoperative PSA
 <4 ng/ml13 (7.7)9 (7.3)2 (40.0)15 (35.7)0 (0.0)
 4–10 ng/ml93 (55.4)72 (58.1)1 (20.0)15 (35.7)0 (0.0)
 >10 ng/ml59 (35.1)40 (32.3)2 (40.0)9 (21.4)0 (0.0)
 Unknown4 (2.4)3 (2.4)0 (0.0)3 (7.1)11 (100)
Preoperative tumor stage
 cT00 (0.0)0 (0.0)5 (100)5 (100)n.a.
 cT1104 (61.5)77 (62.1)0 (0.0)0 (0.0)n.a.
 cT255 (32.5)42 (33.9)0 (0.0)0 (0.0)n.a.
 cT31 (0.6)1 (0.8)0 (0.0)0 (0.0)n.a.
 Unknown9 (5.3)4 (3.2)0 (0.0)0 (0.0)n.a.
Pathological tumor stage
 pT10 (0.0)0 (0.0)5 (100)n.a.n.a.
 pT2113 (66.9)82 (66.1)0 (0.0)n.a.n.a.
 pT353 (31.4)39 (31.5)0 (0.0)n.a.n.a.
 pT43 (1.8)3 (2.4)0 (0.0)n.a.n.a.
 Lymph node invasion8 (4.7)6 (4.8)n.a.n.a.n.a.
 Capsular penetration55 (32.5)42 (33.9)n.a.n.a.n.a.
 Seminal vesicle infiltration24 (14.2)19 (15.3)n.a.n.a.n.a.
 Extraprostatic extension55 (32.5)41 (33.1)n.a.n.a.n.a.
 Positive surgical margins63 (37.3)47 (37.9)n.a.n.a.n.a.
 G116 (9.5)12 (9.7)4 (80.0)n.a.n.a.
 G2115 (68.0)85 (68.5)1 (20.0)n.a.n.a.
 G338 (22.5)27 (21.8)0 (0.0)n.a.n.a.
Gleason score
 2–418 (10.7)13 (10.5)4 (80.0)n.a.n.a.
 5–691 (53.8)66 (53.2)1 (20.0)n.a.n.a.
 735 (20.7)25 (20.2)0 (0.0)n.a.n.a.
 8–1025 (14.8)20 (16.1)0 (0.0)n.a.n.a.

Measurement of DNA fragment levels and the apoptosis index by quantitative real-time PCR (QPCR)

Initial experiments conducted with primer sets amplifying similar sized regions at GSTP1, TIG1, PTGS2 and Reprimo revealed that all probes gave similar linear relationships of DNA levels across the samples used and that these levels were highly correlated (r > 0.84; p < 0.001; data for GSTP1 and PTGS2 shown in Fig. 1). Amplification of the 124 bp PTGS2 and 271 bp Reprimo fragment, however, demonstrated excellent PCR efficiency and were used in this study (see supplementary information: Fig. S1, Fig. S2 and Table S1). Neither gene is located on a chromosome, which has been identified as developing genetic mutations during the development of PCA. Therefore, the observed difference in PTGS2 and/or Reprimo DNA fragment levels can be considered the result of different concentrations of small and/or large fragments in cell-free serum DNA. It was previously reported that electrophoretic patterns of cell-free DNA in serum/plasma biochemically characterize DNA derived from apoptotic cells.7, 10 The apoptosis index (AI) was calculated as the ratio of PTGS2 to Reprimo DNA fragments.

Figure 1.

Correlation plot revealing the degree of correlation in DNA levels in serum measured with the short GSTP1 and PTGS2 probes (n = 50; r = 0.91, p < 0.001).

Quantitative real-time PCR was carried out in triplicate on an ABIPrism 7900HT (Applied Biosystems, Foster City, CA). Each 10 μl reaction consisted of 1× SYBRGreenER Mix (Invitrogen, Paisley, Scotland), 200 μM sense and antisense primer and 1 μl of DNA sample. PCR was done by the following procedure: 90°C for 10 min, followed by 40 cycles at 95°C for 15 sec and 60°C for 60 sec. Melting curve analysis was performed to confirm specificity of the PCR products. Each run included serial dilutions of an external standard and water blanks. The external standard was prepared according to the protocol “Creating Standard Curves with Genomic DNA or Plasmid DNA Templates for Use in Quantitative PCR” (Applied Biosystems). Samples were analyzed without prior knowledge of the specimen identity.

Methylation-sensitive real-time PCR

CpG island hypermethylation of the promoter region at GSTP1 is present in ∼ 90% of PCA.12 To provide insight about the origin of cell-free serum DNA, we performed methylation-sensitive PCR. Cell-free DNA of 30 PCA patients (pT2: n = 20; pT3: n = 10) was digested with methylation-sensitive restriction enzymes which cut the unmethylated sequence whereas cleavage is completely blocked by methylation. We incubated 10 μl purified serum DNA with each 50 U HpaII and HinP1I (New England Biolabs, Frankfurt, Germany) in a total volume of 25 μl for 4 hr at 37°C. To ensure complete digestion, additional 10 U HpaII/HinP1I were added. Following additional 14-hr incubation, the restriction enzymes were inactivated at 65°C for 20 min. Identically, an input control was set-up without restriction enzymes. The PCR product (size: 131 bp) covered one restriction site of each HpaII and HinP1I, and thus only hypermethylated cell-free DNA is amplified. Quantitative PCR was performed essentially as described above. Levels of methylated DNA were normalized to the input DNA.

Statistical analysis

Differences of serum DNA levels/AI between healthy individuals, BPH and PCA patients were analyzed using the Mann–Whitney test. Prostate weight was correlated with cell-free DNA levels/AI using the Spearman test. Receiver-operating characteristic (ROC) analysis was used to determine the area under curve (AUC), sensitivity and specificity of circulating DNA fragments and the AI. Correlations between clinicopathological parameters and serum DNA levels or AI were performed using Mann–Whitney and Kruskal–Wallis-tests, as appropriate. Kaplan–Meier analysis was used to correlate the period of PSA-free survival and total DNA levels/AI. Statistical tests were performed using SPSS 11 (SPSS, Chicago, IL).


Quantification of cell-free serum DNA fragments (Table II, Fig. 2)

We analyzed the concentration of 2 different sized DNA fragments in serum of patients with PCA, BPH and healthy individuals. Mean PTGS2 DNA fragment (124 bp) levels were significantly increased in PCA patients in comparison to BPH patients and healthy individuals (both p < 0.0001). Noteworthy, all patients with incidental PCA had short PTGS2 DNA fragment above the threshold distinguishing PCA from BPH patients most accurately. There were no significant differences (p = 0.151) between BPH and healthy individuals. Reprimo DNA fragment (271 bp) levels were slightly increased in PCA patients, but differences were not significant (vs. BPH: p = 0.601; vs. healthy individuals: p = 0.759). Reprimo DNA levels were similar in BPH and healthy individuals (p = 0.930). These results suggest that increases in serum DNA in PCA is predominantly derived from apoptotic cells. We consequently defined the AI as the ratio of 124 bp PTGS2 to 271 bp Reprimo DNA fragments. The AI was significantly increased in PCA vs. BPH patients and healthy individuals (p < 0.0001; Table II). The AI was not significantly different between BPH patients and healthy individuals (p = 0.056). See Table II and Figure 2 for details.

Figure 2.

Circulating PTGS2 and Reprimo DNA fragment levels were determined as described in Methods by quantitative real-time PCR using primer sets for a (a) 124-bp gene fragment at the PTGS2 promoter and (b) a 271-bp gene fragment at the Reprimo promoter. The apoptosis index (c) was defined as the ratio of PTGS2 to Reprimo fragment levels. Receiver operator characteristic analysis was performed to evaluate the diagnostic information of PTGS2 DNA fragments (d), Reprimo DNA fragments (e), apoptosis index (f) and serum PSA (g).

Table II. Levels of Serum PTGS2 and Reprimo DNA Fragments and the Apoptosis Index in Patients with Prostate Cancer (PCA), Benign Prostatic Hyperplasia (BPH) and Healthy Individuals
 PCA (n = 168)Incidental PCA(n = 5)BPH (n = 42)Healthy individuals (n = 11)
  • 1

    p-value in comparison to prostate cancer patients (Mann–Whitney-test).

  • 2

    Reprimo DNA was detected only in one patient with incidental PCA; n.d. not done.

PTGS2 (124bp fragment) 
 p-value2 n.d.<0.001<0.001
 Median (ng/ml)70.2388.2110.497.08
 Mean (ng/ml)85.9584.7925.2610.88
 Minimum (ng/ml)2.7826.110.000.00
 Maximum (ng/ml)668.52144.4893.1540.86
 10th percentile (ng/ml)14.1526.111.220.18
 90th percentile (ng/ml)181.46144.4774.9238.74
Reprimo (271bp fragment) 
 p-value1 n.d.0.6010.759
 Median (ng/ml)9.1326.639.44
 Mean (ng/ml)20.052.9616.6910.81
 Minimum (ng/ml)0.0020.000.51
 Maximum (ng/ml)472.79297.5025.14
 10th percentile (ng/ml)1.3521.130.73
 90th percentile (ng/ml)43.35241.1624.46
Apoptosis index    
 p-value1 n.d.<0.001<0.001
 10th percentile1.2920.580.31
 90th percentile29.0429.851.58

GSTP1 CpG islands hypermethylation in cell-free serum DNA (Fig. 3)

To determine whether cell-free DNA was derived from cancer cells, we treated 30 PCA samples with methylation-sensitive restriction endonucleases, and investigated CpG island hypermethylation at the GSTP1 promoter using QPCR (see also supplementary material: Fig. S3). Hypermethylated DNA was detected in 21 (70%) serum samples. However, hypermethylated DNA accounted for only a minor part of the serum DNA (mean 1.92%) and was not associated with pathological tumor stage (pT2 vs. pT3: p = 0.812; Fig. 3). Cell-free DNA in PCA patients is therefore mainly derived from normal cells.

Figure 3.

The non-cancerous origin of cell-free DNA in serum of patients with prostate cancer: GSTP1 hypermethylation was detected in 70% of serum samples of PCA patients, but less than 10% of total-cell free DNA harboured hypermethylation.

Diagnostic capacity of serum DNA fragments (Table III, Fig. 2)

To assess the clinical usefulness of serum DNA fragments we calculated their sensitivity and specificity using ROC analysis. Data from patients with PCA and BPH were analyzed; healthy individuals (n = 11) and patients with incidental PCA (n = 5) were excluded from this analysis because of their differences in age and the low number of patients, respectively. A calculated cut-off level of 19.7 ng/ml for the short PTGS2 DNA fragment distinguished between BPH and PCA patients with good sensitivity (87.5%) and moderate specificity (64.0%). On the other hand, the AI discriminated PCA and BPH patients more specifically (81.3%), but less sensitively (67.5%). In comparison, the diagnostic accuracy of serum PSA levels was distinctly low (Table III and Fig. 2).

Table III. Diagnostic Information on Serum PSA Levels and Cell-Free Circulating DNA Characteristics
 Area under curve (95% CI)Sensitivity (%)Specificity (%)Negative predictive value (%)Positive predictive value (%)
  • 1

    As calculated using receiver operating characteristic analysis.

PTGS2 fragments (cut-off: 19.7 ng/ml)10.824 (0.754–0.893)87.564.056.390.7
Apoptosis index (cut-off: 2.28)10.786 (0.703–0.868)67.581.346.691.2
PSA (cut-off: 4ng/ml)0.615 (0.505–0.724)92.138.553.686.6

The diagnostic accuracy increased if PSA and PTGS2 DNA fragment measurements were combined. In this case, patients with PCA and elevated serum PSA levels (>4 ng/ml) were discriminated from BPH patients with very good sensitivity (80%) and specificity (83%). The combination of PSA and AI showed similar results (Table III).

Prognostic relevance of serum DNA (Fig. 4)

We analyzed whether the 124 bp PTGS2 DNA fragment levels or the AI were correlated with clinicopathological parameters (i.e. age, preoperative PSA, prostate weight, pT-stage, capsular penetration, seminal vesicle infiltration, extraprostatic extension, positive surgical margins, Gleason score, grading). PTGS2 DNA fragment levels were not correlated with any parameter. Grading was correlated with the AI (p = 0.044), i.e. in dedifferentiating PCA, the AI was increased (mean ± standard error: 5.42 ± 1.46 (G1), 11.74 ± 1.94 (G2) and 21.66 ± 6.09 (G3), respectively). None of the other parameters were correlated with the AI.

Figure 4.

Kaplan–Meier analysis for 124 bp circulating PTGS2 DNA fragments and apoptosis index in prostate cancer (PCA) patients as specified in Methods. (a) Comparison of circulating DNA fragments above and below 19.7 ng/ml threshold (p = 0.0395). (b) Apoptosis index above and below 2.28 threshold (p = 0.0482).

Follow-up information was available for 124 patients (Table I). The maximum period of follow-up period was 72 months. Biochemical recurrence following radical prostatectomy was defined as serum PSA >0.2 ng/ml. Of the 124 cases with follow-up, 24 patients suffered from biochemical recurrence. Kaplan–Meier analysis revealed that increased free circulating PTGS2 DNA fragments as well as an increased AI were correlated with biochemical recurrence (log rank p = 0.0395 and p = 0.0482; see Fig. 4). Noteworthy is that none of the patients with DNA fragment levels below 19.7 ng/ml suffered from biochemical recurrence.


Cell-free serum DNA has been shown to be increased in various malignancies.7 Importantly, patients with nonmalignant disease (e.g. infectious, cardiovascular, autoimmune and musculoskeletal disease) that can increase serum DNA levels have substantially lower concentrations of DNA in plasma than patients with cancer.5 Therefore, serum DNA seems to be a potential universal tumor marker. This approach is of special interest in cancer types lacking peripheral blood tumor markers that can be detected with satisfying accuracy or/and reliability.13

The origin of cell-free DNA remains enigmatic. Circulating tumor DNA seems to be present in serum of most patients with PCA: in a subset of 30 patients with clinically localized PCA we detected hypermethylation at the GSTP1 promoter in 70% of cases revealing its tumor origin. However, tumor DNA accounted for only a very small fraction (mean: 1.9%) of total serum DNA. The predominant source of serum DNA was noncancerous cells. We therefore hypothesize that peripheral cells increasingly undergo apoptosis because of the release of proapoptotic cytokines into the circulation by PCA cells. Increased levels of noncancerous DNA would therefore indirectly indicate the presence of PCA. This idea is supported by previous studies: (i) electrophoretic patterns of cell-free DNA show biochemical characteristics of apoptosis (e.g. presence of mono-, di- and tri-nucleosomal DNA fragments).7, 10 (ii) Most PCA cells express and secrete Fas-ligand,14 thus PCA cells are presumably the origin and cause of increased serum Fas-ligand levels and apoptosis in PCA patients.15 Indeed the AI is ∼4-fold increased in PCA patients. Notably, the 271 bp Reprimo DNA fragment concentrations (characterising DNA of nonapoptotic cell-death) were nearly similar in all subgroups. Seemingly controversial to our results it was reported that long DNA fragments (>300 bp) are increased in sera of patients with breast,8, 9 colon,16 ovarian8, 16 and head and neck cancer.17 These authors suggest that circulating DNA mainly originates from necrotic tumor and bystander cells and used primer sets amplifying short and long fragments of the repetitive Alu-element.9, 16 In this context, it was reported that the proportion of Alu fragments compared to b-globin fragments was significantly greater in serum DNA than in lymphocyte DNA, and thus a preferential release of long repetitive Alu sequences is due to an active DNA release.18 Furthermore, different cancer cells may induce the release of either necrotic or apoptotic DNA fragments into circulation depending upon specific cancer entity.

Nevertheless, our data shows that circulating short PTGS2 DNA fragments and the AI are considerately increased in PCA patients and that this fact enables to distinguish between PCA and BPH patients (including healthy controls). Using ROC analysis, we showed that by measuring PTGS2 DNA fragments and the AI a test accuracy can be achieved (sensitivity of 88% and a specificity of 81%), which was well above the diagnostic information of PSA testing in our patient cohort. A recent report suggests that the combination of serum DNA and PSA measurement is useful to increase the predictive accuracy.19 Somewhat controversial, Boddy et al. observed increased plasma DNA levels also in patients with benign prostatic disease. But to our opinion, PCA was not definitively excluded at the time of blood withdrawal by histological examination and increased DNA levels may be explained by undetected PCA. This assumption is supported by our finding that even patients with incidental PCA had increased serum DNA fragment levels.

Undoubtedly, PCA-specific mortality would be the most reasonable factor to assess the prognostic efficacy of serum DNA fragments. We chose PSA-free survival as a reasonable compromise because sufficient follow-up time was not yet reached. In a subset of 124 patients with a follow-up period up to 72 months, increased short PTGS2 DNA fragment levels and the AI were, to a significant extent, associated with a shorter period of PSA-free survival. Although the average period of follow-up is relatively short in our study, we believe that this finding has important prognostic implications. Patients with early PSA recurrence within the first 3 years following radical prostatectomy have an increased risk of cancer-specific mortality.20 Therefore, the analysis of serum PTGS2 DNA fragments may help to identify patients who may benefit from early multimodal adjuvant therapy. It was also shown by Jung et al. that increased plasma DNA levels of patients with metastatic PCA were associated with disease specific survival.6 Hence, serum PTGS2 DNA fragments could serve as a postoperative surveillance marker, especially for patients with PSA-negative primary PCA.

Initial experiments showed that DNA levels were similar regardless of the gene site targeted by the primer set and implicating that the release and degradation of cell-free DNA occurs randomly (see Fig. 1). We analyzed short and long DNA fragment levels using primer sets amplifying different gene sites (e.g. Reprimo and PTGS) because the PCR efficiency was optimal in these cases. Theoretically, DNA fragment levels could be affected due to deletion or amplification of the target sites in PCA. However, neither deletion nor amplification of the PTGS2 (chromosome 1q31.1) or Reprimo (chromosome 2q23.3) gene site has been reported in PCA so far. Even if such deletions or amplifications occur, this is of limited relevance for the quantification of cell-free DNA: tumor-specific DNA accounts only for a small percentage of total cell-free DNA (mean: 1.9%). In summary, we assume that the increase of short PTGS2 DNA fragments is not caused by amplifications of the PTGS2 gene in PCA tissue.

Although cell-free DNA was detected in peripheral blood 30 years ago,4 procedures concerning collecting and processing of blood samples are still controversial. DNA levels are lower in plasma than in serum and it was assumed that DNA release during clotting of fragile cells is the reason for this difference. A recent report clearly shows that DNA concentrations in serum compared to plasma are ∼6-fold higher, but the contribution of extraneous DNA from ruptured cells is only minor.21 The authors advocate that serum is better than plasma as source for circulating DNA in cancer.21 The comparison of different studies is also complicated by the fact that the amount of serum/plasma used for extraction varies markedly (600 μl–20 ml)6, 7, 19, 22 and commercial DNA isolation kits have a different DNA extraction efficiency.23 Therefore, standardized protocols are mandatory to evaluate the significance of different DNA fragments in cancer, i.e. PCA.

In conclusion, circulating PTGS2 DNA fragment levels and the AI are promising noninvasive biomarkers in the diagnosis and prognosis of PCA. However, DNA quantification needs meticulous standardization, and prospective, large-scale studies are essential to confirm the clinical value of circulating DNA fragments in PCA.


We thank Ms. Doris Schmidt for excellent technical assistance, and Prof. Hans-Peter Bastian and Mr. Rudolf Moritz for the collection of serum samples at the St.-Josef Hospital, Troisdorf, Germany. Thestudy was supported by a BONFOR grant to Jörg Ellinger (grant number O-116.0015) and Patrick J. Bastian (grant number O-116.0019).