Diagnostic assays based on blood sample analysis are attractive because of the simplicity of sample collection. Accurate analysis of tumor markers in blood from cancer patients could have a significant impact in facilitating the screening, diagnosis and monitoring for disease recurrence after initial therapy. The development of a simple standardized assay for cancer detection in head and neck squamous cell carcinoma (HNSCC) is appealing because of the low sensitivity of current screening procedures and the fact that HNSCC attains an annual incidence greater than 40,000 cases per year.1 More efficient and noninvasive screening methods could possibly impact the outcomes of HNSCC by improving patient compliance with screening procedures and by allowing earlier detection of the disease.
DNA is a nuclear macromolecule that can exist in an intracellular and extracellular form. In its extracellular form, DNA can appear in the blood as well as other biological fluids.2, 3, 4, 5, 6 Several studies have identified DNA alterations in circulating plasma DNA from cancer patients that match with genetic changes present in primary tumors.7, 8, 9, 10 These analyses of plasma DNA may theoretically be used for prognostic purposes or for early diagnosis and other detection strategies.11 The low sensitivities reported for detection of DNA alterations in plasma, however, make these approaches challenging for immediate clinical applications.
Jahr et al. proposed that DNA could be released from apoptotic or necrotic cells and that DNA size distribution may be used to determine the origin of DNA from either apoptotic or necrotic cells.12 Tumor necrosis is a frequent event in solid malignant neoplasms, and it generates a spectrum of DNA fragments with different strand lengths because of random and incomplete digestion of genomic DNA by a variety of deoxyribonucleases. In contrast, cell death in normal tissues occurs predominantly via apoptosis, resulting in the production of small and uniform DNA fragments. Support for this hypothesis has been found in recent studies demonstrating increased DNA length in plasma from patients with breast and gynecologic cancers.13
We attempted to use a similar approach as described by Wang et al.13 and to determine if plasma DNA integrity as measured by DNA strand length could serve as a potential marker for HNSCC and whether altered DNA strand lengths remained in the plasma compartment after surgical resection of HNSCC. We employed a real-time PCR-based assay to assess the DNA strand integrity of plasma DNA in a total of 58 HNSCC patients with paired pre- and postsurgical plasma and plasma from 47 control subjects without HNSCC.
AWD, alive with disease; DOD, dead of disease; DOU, dead of unrelated causes; HNSCC, head and neck squamous cell carcinoma; NED, no evidence of disease; QPCR, quantitative PCR; ROC, receiver operating characteristic.
Material and methods
Fifty-eight HNSCC patients with paired pre- and postoperative plasma samples were prospectively enrolled in a surveillance trial within the Department of Otolaryngology-Head and Neck Surgery, School of Medicine, The Johns Hopkins University after appropriate approval obtained from the Johns Hopkins institutional review board. Appropriate staging, outcome, exposure and clinical measures were obtained by retrospective chart review. We also obtained 47 control plasma samples from the subjects without HNSCC with a variety of ages and risk factors recruited for a head and neck screening study, which was then age matched to HNSCC patients. All control subjects were administered a confidential survey of risk factors for head and neck cancer (alcohol, tobacco use, etc.) as well as the presence of comorbid illness. Smoking and drinking were defined as at least 1 continuous year of daily smoking or drinking.
Extraction of DNA from plasma samples
Plasma samples were obtained by centrifugation of 5 ml of peripheral blood at 2,000g for 10 min. The plasma was carefully collected from the upper portion of the supernatant and placed in 1% SDS/proteinase K (0.5 mg/ml) at 48°C for 72 hr. The digested product was then subjected to phenol–chloroform extraction and ethanol precipitation.
DNA strand integrity analysis
DNA integrity index was determined as described previously.13 DNA strand integrity was measured by quantitative PCR (QPCR) using the Perkin-Elmer/ABI 7900 thermocycler to determine the integrity index, which was defined as the ratio in relative abundance of 400 versus 100 bp PCR products of a housekeeping gene, the β-actin. Primers were custom made and obtained from Invitrogen (Carlsbad, CA). Both 100- and 400-bp PCR fragments were amplified using the same forward primer: 5-GCACCACACCTTCTACAATGA-3. The nested reverse primer for the 100-bp product was 5-GTCATCTTCTCGCGGTTGGC-3 and for the 400-bp product was 5-TGTCACGCACGATTTCCC-3. PCR amplifications were carried out in a buffer containing: 16.6 mM ammonium sulfate, 67 mM Trizma, 2.5 mM MgCl2, 10 mM β-mercaptoethanol, 0.1% DMSO, 600 nM each of forward and reverse primers, 200 nM TaqMan probe, 0.6 units Platinum Taq polymerase, 2% ROX reference dye and 0.0025% SYBR Green dye. DNA isolated from white blood cell fractions, matched to each subject, was used as a reference to determine the relative DNA strand integrity in plasma DNA. The threshold (Ct) value for each reaction was calculated by the ABI 7900 software. The Ct value for the 400-bp product for a sample was subtracted from that for the reference control to obtain the normalized Ct value for the 400-bp product (ΔCt-400). Likewise, the Ct value for the 100-bp product for a sample was subtracted from that for the reference control to obtain the normalized Ct value for the 100-bp product (ΔCt-100). The normalized Ct value for the 400-bp product was then subtracted from that for the 100-bp product to obtain a final ΔΔCt value. The integrity index was calculated as exponential of (−ΔΔCt × LN 2). All samples were analyzed blindly without prior knowledge of the specimen identity.
The major statistical endpoint in this study was the comparison of the difference in plasma DNA integrity index between HNSCC patients and non-HNSCC control groups. Two sample t tests were used to compare mean DNA integrity index by sample, gender, race, smoking and alcohol groups. DNA integrity index values were transformed with the log transformation for these analyses. Means and standard deviations are reported on the natural scale. A secondary endpoint in this study was a receiver operating characteristic (ROC) analysis for DNA integrity index using SPSS software. An ROC curve was used to assess the feasibility of using plasma DNA strand integrity index as a diagnostic tool for detecting HNSC and other cancers. All statistical computations were performed using the SAS system and all p values reported are 2 sided.
QPCR was performed for 58 HNSCC patients with pre- and postoperative plasma samples and DNA integrity index was compared to 47 non-HNSCC plasma samples. Table I shows the comparison of base-line characteristics of the HNSCC and control group. The mean age (56) of the HNSCC patients was similar to that of the normal group (57). The gender distribution was significantly different in these 2 groups, with the HNSCC group being 77.6% (45/58) male when compared to the 46.8% (22/47) seen in the normal group, p = 0.002. Race and smoking proportions were similar in these groups, while drinking patterns were different. Seventy-four percent (42/57) of the HNSCC group were never drinkers, 5.3% (3/57) were former drinkers and 21.1% (12/57) were current drinkers. In the normal subjects there was a larger proportion of former drinkers, 59% (26/44), and a smaller proportion of never-drinkers, 18% (8/44), with a similar proportion of current drinkers, 23% (10/44). The majority of HNSCC patients studied were Stage III and IV (73.1%) (Table II). Of those patients, 9 received radiation therapy before tumor resection and 26 received postoperative radiation therapy. The preoperative plasma samples were collected before or on the day of tumor resection, prior to surgery. The postoperative plasma samples were collected 4.65 months (mean) after tumor resection, range from 1.75 to 13 months (median 4 months). For this cohort, the mean follow-up interval was 27.0 months and median interval was 27.9 months (range from 5.75 to 45 months).
Table I. Characteristical Comparison of HNSCC Subjects and Controls
Prior radiation, received radiation before surgery; post radiation, second plasma sampling after surgery and radiation; NED, no evidence of disease; AWD, alive with disease; DOD, dead of disease; DOU, dead of unrelated cause.
DNA integrity index as a diagnostic tool in HNSCC
Mean DNA integrity index was significantly greater in plasma from patients with HNSCC, 0.24 (95% CI: 0.11, 0.38) when compared to plasma from control subjects −2.24 (95% CI: −2.92, −1.56), p < 0.0001 (Fig. 1). Integrity index differed significantly by gender as well, −0.53 (95% CI: −0.98, −0.08) in males when compared to −1.46 (95% CI: −2.18, −0.74) in females, p = 0.02. In terms of smoking exposure, the DNA integrity index of current smokers was −0.79 (95% CI: −1.36, −0.23), similar to that of never smokers, −1.04 (95% CI: −1.62, −0.45), p = 0.52. Mean integrity index was not significantly different by race or alcohol categories (Table III). Adjusting for gender and smoking, the difference in DNA integrity index by HNSCC diagnosis remained significant.
Table III. Overall Statistical Comparison of HNSCC Subjects and Controls
Mean log (DNA integrity index)
HNSCC (n = 58)
0.24 (95% CI: 0.11, 0.38)
Control (n = 47)
−2.24 (95% CI: −2.92, −1.56)
Female (n = 38)
−1.46 (95% CI: −2.18, −0.74)
Male (n = 67)
−0.53 (95% CI: −0.98, −0.08)
White (n = 91)
−0.81 (95% CI: −1.22, −0.39)
Other (n = 14)
−1.28 (95% CI: −2.59, 0.02)
Never (n = 37)
−1.04 (95% CI: −1.62, −0.45)
Current (n = 66)
−0.79 (95% CI: −1.36, −0.23)
Never (n = 50)
−0.36 (95% CI: −0.80, −0.08)
Current (n = 22)
−0.77 (95% CI: −1.72, 0.19)
To assess the feasibility of using plasma DNA strand integrity index as a diagnostic tool for detecting HNSCC, a secondary endpoint in this study was a ROC analysis for DNA integrity index (Fig. 2). The area under the ROC curve for β-actin was 0.86 (95% CI: 0.78, 0.94). The optimal sensitivity (defined as the value for which sensitivity equals specificity) was found with DNA integrity index greater than 0.82: sensitivity, 84.5%; specificity, 83%.
Comparison of DNA integrity index in pre- and postoperative HNSCC plasma
To evaluate whether changes in plasma DNA integrity index occurred after HNSCC patients underwent resection with or without radiotherapy, we compared overall pre- and postoperative plasma of HNSCC patients. We found there was a slight decrease in the mean DNA integrity index in the postoperative HNSCC plasma, with a mean change of −0.09 (95% CI: −0.28, 0.09), which was not significant, p = 0.32. We also examined the contribution of postoperative radiation to DNA integrity index. We found that the mean DNA integrity index change between pre- and posttreatment in plasma of HNSCC patients with postoperative radiation therapy was −0.08 (95% CI: −0.32, 0.16), which was not significant when compared to the change in HNSCC patients without postoperative radiation therapy, who exhibited a mean change of −0.01 (95% CI: −0.38, 0.36), p = 0.75.
There was a small subset of 9 HNSCC patients, which received radiation therapy prior to surgery. The mean change of this group was −0.35 (95% CI: −0.74, 0.03), which was not significant when compared to the change in HNSCC patients without postoperative radiation therapy, p = 0.18.
In the present study, we compared DNA integrity index in the pre- and postoperative plasma of patients with HNSCC and control subjects using real-time PCR. Our results indicated that DNA integrity index is elevated in the plasma of patients with HNSCC, a fairly specific finding in comparison with controls. These data are consistent with a recent study of patients with gynecologic and breast malignancy that demonstrated a median DNA integrity index of 0.66 (interquartile range = 0.42–0.90) in the neoplastic group, which was significantly higher than 0.14 (interquartile range = 0.06–0.28) in the nonneoplastic group (p < 0.0001, Wilcoxon's rank-sum test).13 These findings can be explained by the fact that necrosis is a more frequent event in tumor patients other than in normal subjects. As the consequences, plasma DNA in tumor patients has a wide range of DNA fragment lengths by random and incomplete digestion of genomic DNA.14, 15 Another issue of interest is the definition of the tumor-specific contribution to of the plasma DNA in HNSCC. Mao et al. showed that 1 tumor cell harboring a microsatellite alteration(s) is detectable among 200–1,000 normal cells.16 Dilution experiments performed by Radford et al. showed that LOH is detectable provided that the amount of normal cells does not exceed 20%.17 Diehl et al. demonstrated that the median number of APC DNA fragments in such patients was 47,800 per ml of plasma, of which 8% were mutant. Mutant APC molecules were also detected in >60% of patients with early, presumably curable colorectal cancers, at levels ranging from 0.01 to 1.7% of the total APC molecules. These results support that circulating DNA is a promising marker for cancer detection.18 However, tumor cell DNA may be “enriched” in the plasma/serum of tumor patients by a mechanism unknown so far.19
It is intriguing that measures of plasma DNA integrity for HNSCC patients measured in the postoperative setting did not show a significant difference when compared to preoperative plasma in the same group of patients. One possible consideration is that there may be an alteration in systemic DNA metabolism that remains even after surgery. It is also possible that cells with altered DNA degradation mechanisms that are still present in mucosa are able to shed DNA after HNSCC surgery. Leon et al. reported that an E. coli DNase added together with a labeled DNA has almost no activity in plasma from a cancer patient, while in plasma of a healthy control the same DNase seems to work as well as in a culture medium.20 In another words, longer DNA fragments in the pre- and/or postoperative HNSCC patients may be the consequence of lower nuclease activity.
Material from dead and dying cells can also appear in the blood in other diseases, including systemic lupus erythematosus, inflammatory conditions and after trauma, thought to be a result from either an excessive burden of dead cells or impaired clearance.12, 21, 22, 23, 24 A study of 84 trauma patients showed a 60-fold median increase in plasma DNA levels in major trauma subjects within 4 hr (median 1 hr) of injury when compared with healthy, uninjured control subjects.25 In patients with less severe injuries, plasma DNA concentrations began to return to reference values within 3 hr, and the plasma DNA concentrations were higher in patients who developed multiple organ dysfunction syndrome. In patients who remained in the ICU with continuing organ dysfunction, plasma DNA remained higher than in healthy controls even at 28 days after injury. Most survivors with multiple organ dysfunction syndrome showed an initial very high peak followed by a prolonged smaller increase.26 In addition, in an animal model treated with agents including lipopolysaccharide, anti-Fas and dexamethasone, release of DNA was shown to vary depending on the death stimulus, the cell type induced to die and the activity of phagocytic cells and clearance of dead and dying cells.27, 28 In contrast to the administration of an inducer of apoptosis or necrosis, the administration of dead and dying cells provides a more defined system in which the number of dead cells is known more precisely. This approach showed that administration of apoptotic or necrotic Jurkat cells to normal mice leads to the appearance of DNA in the blood in a time- and dose-dependent manner.23
It is possible that the persistent elevation in postoperative patients with HNSCC may be related to persistent inflammation associated with postoperative healing or postoperative radiation therapy. Although the mean interval at which postoperative plasma samples were collected was 5.3 months after surgical resection, the mean interval between collection and completion of postoperative radiotherapy in radiated patients was 2 months, a time at which radiation induced mucositis may persist. Of note, we did not find an association between persistent elevation of DNA integrity index and postoperative radiotherapy, but our cohort was not large enough to test this hypothesis rigorously. In addition, separate analyses of radiated and nonradiated patients to determine whether a decrease in plasma DNA integrity index was associated with increasing time interval between surgery and sample collection were limited by the small sample size. Future studies should prospectively assess the value of plasma DNA integrity carefully at various time points after treatment with careful clinical follow-up to better understand the value of this approach in patient monitoring.
In conclusion, our results indicate that DNA integrity index in the plasma of the patients with HNSCC is significantly elevated in comparison to that in non-HNSCC control subjects. In addition, the postoperative plasma of HNSCC patients did not demonstrate a significant difference in the DNA integrity index when compared to that of the preoperative plasma.