Early Detection and Diagnosis
Feasibility of quantitative PCR-based saliva rinse screening of HPV for head and neck cancer
Article first published online: 31 MAY 2005
Copyright © 2005 Wiley-Liss, Inc.
International Journal of Cancer
Volume 117, Issue 4, pages 605–610, 20 November 2005
How to Cite
Zhao, M., Rosenbaum, E., Carvalho, A. L., Koch, W., Jiang, W., Sidransky, D. and Califano, J. (2005), Feasibility of quantitative PCR-based saliva rinse screening of HPV for head and neck cancer. Int. J. Cancer, 117: 605–610. doi: 10.1002/ijc.21216
- Issue published online: 14 SEP 2005
- Article first published online: 31 MAY 2005
- Manuscript Accepted: 14 MAR 2005
- Manuscript Received: 8 DEC 2004
- Damon Runyon Cancer Research Foundation. Grant Number: CI-#9
- National Institute of Dental and Craniofacial Research. Grant Number: 1R01DE015939-01
- head and neck squamous cell carcinoma;
- quantitative PCR;
Human papilloma virus (HPV) 16 is likely to be an etiologic factor in a subset of head and neck squamous cell carcinomas (HNSC). We investigated the ability of a quantitative PCR-based assay to detect HPV 16 in salivary rinses as a screening method for HNSC. Real time quantitative PCR was applied to detect HPV16 E6 and E7 DNA level in 92 primary tumors and salivary rinses from patients with HNSC and 604 control subjects without HNSC. A total of 45.6% (42/92) of primary HNSC and 32.6% (30/92) of saliva rinse samples from HNSC patients had any detectable HPV 16 DNA, 30.4% (28/92) of tumors had HPV 16 levels >0.1 copies/cell. HPV16 was detected in 50% (confidence interval [CI] = 34.2–65.8%) (21/42) of the saliva rinse samples from HNSC patients with detectable HPV16 level in their tissue, 18% (CI = 8.4–30.9%) (9/50) of saliva rinse samples from patients with HPV 16 negative primary HNSC, in 2.8% (CI = 1.7–4.5%) (17/604) of the normal controls (p < 0.001). Using a cutoff of HPV 16 >0.001 copies/cell in saliva rinse yielded a sensitivity of 30.4% and a specificity of 98.3%. Nonsmokers had significant higher HPV 16 level than smokers in the cohort of cancer patients. HPV 16 DNA in saliva rinses can reflect HPV 16 status of primary HNSC. Quantitative analysis of HPV 16 DNA in salivary rinses allows for detection of HPV-related HNSC, however, specific limitations exist that prevent the application of this as a screening technique for a broad population. © 2005 Wiley-Liss, Inc.
Head and neck squamous cell carcinomas (HNSC) account for 6.5% of annual cancer cases worldwide. In United States, HNSC comprises about 4% of all malignancies.1, 2, 3 The etiology of squamous cell carcinoma of the head and neck includes well-known risk factors such as tobacco smoking and alcohol drinking,4, 5 as well as human papillomavirus (HPV) high-risk subtypes, predominantly HPV16.6, 7, 8, 9, 10 Several investigators have attempted to determine the prevalence of high-risk HPV subtype infection in head and neck cancer patients, with estimates ranging from 2–90%, depending on the patient population as well as the sampling and assay methods used, most recent estimates demonstrate high-risk HPV in approximately 25% of all HNSC. HPV is present in a higher rate in tumors of the oropharynx, with rates of 50%.6, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 HPV-related tumors are diagnosed at later stages, due to their predominant occurrence in oropharyngeal sites, where early cancers are difficult to distinguish from lymphoid tonsillar and base of tongue tissue.22 Common clinical presentations include signs symptoms attributable to later stage tumors, including dysphagia, pain, and neck masses, an the incidence of oropharyngeal tumors is increasing in relation to other head and neck cancers.22 Development of a saliva-based screening test for oropharyngeal cancer, therefore, may be able to detect early stage oropharynx cancer before development of clinical symptoms. Integration of high-risk HPV DNA into the human cellular genome is an important step that leads to malignant transformation. Molecular studies have shown that high-risk HPV integration results in production of the viral oncoproteins E6 and E7, which promote tumor progression by inactivating the p53 and Rb tumor suppressor gene products.23, 24, 25, 26
Application of a real-time PCR technique in determination of HPV DNA level in head and neck tumor tissue and premalignant tissue has been evaluated, demonstrating more sensitivity than other conventional methods, including in situ hybridization, Southern and Northern blotting, PCR, reverse transcription-PCR and DNA sequence analysis.6, 25, 27, 28 Recent studies have also shown that viral DNA sequences may be detectable in serum.27 Recently, investigators have also used regular PCR-based techniques to detect HPV DNA in saliva rinses from patients with HNSC. Herrero et al.21 found HPV DNA in exfoliated cells was not associated HPV DNA detection in tumor biopsy specimens. On the contrary, Smith et al.29 found HPV high-risk types detected in oral exfoliated cells were predictive of HPV-high-risk types in tumor tissue, and have shown that HPV infection is a risk factor for head and neck cancer, independent of alcohol and tobacco use, and acts synergistically with alcohol consumption.30 Variability in prior reports may be based on a variety of factors, including lack of quantitative analysis of HPV. To clarify issues related to HPV testing of saliva rinses for HNSC, we hypothesized quantitative PCR-based techniques may improve sensitivity and specificity of HPV DNA detection in saliva rinses from patients with HNSC.
We selected HPV-16 E6 and E7, the most frequently identified types of HPV in HNSC as target genes. The aim of our study is to explore a quantitative, sensitive and specific method to determine the prevalence of high-risk HPV infection in HNSC and normal population, to determine the feasibility of using this technique to as a screening tool for patients with head and neck cancer in a normal population.
Material and methods
Sequential snap frozen tumor tissue samples from 92 patients were collected from the Johns Hopkins Hospital Department of Otolaryngology & Head and Neck Surgery between 1984 and 2002, using appropriate informed consent obtained after institutional review board approval. Study patients included those patients with primary squamous cell carcinoma of the upper aerodigestive tract, including oral cavity, oropharynx, hypopharynx, larynx or unknown primary squamous cell carcinoma metastatic to the neck. No patients were excluded. The average age of the patients was 60 (range = 32–82), including 65 men and 27 women. Corresponding pretreatment saliva was collected by 20 cc 0.9% NaCl rinses that included brushing bilateral buccal, lateral tongue, floor of mouth, soft palate/tonsillar pillar, followed by a rinse, gargle and expectoration. Samples were centrifuged at 2,000g for 10 min, supernatant discarded and DNA was extracted from the cell pellet as described below. Clinical and exposure data were extracted from retrospective clinical chart review.
Control saliva rinses were collected and DNA was extracted in an identical fashion from a control group of subjects during a community screening study for head and neck cancer, according to study protocols approved by institutional review boards. Patients were English speaking individuals with >18 years of age. The average age of this population was 61 (range = 18–92), with 380 men and 222 women. A single case was excluded after initial enrollment due to voluntary subject withdrawal. Control subjects with prior head and neck cancer were excluded from our study, other comorbidities were noted but did not result in study exclusion. Two control subjects did not respond to questions defining gender. All samples had adequate DNA available for testing. A smoker was defined as someone that has smoked daily for >1 year, and a former smoker is a smoker who has quit smoking completely for >2 weeks.
Sections (40–50) of 10-μm thickness were cut from frozen tissue blocks and microdissected using a microsurgical forceps and blade. Samples or cell pellets were placed in 3 ml 1% SDS/proteinase K overnight at 45°C. The DNA was extracted with phenol chloroform, ethanol precipitated and stored at −20°C. The samples were diluted to 10 ng/4 μl and ready for quantitative PCR.
Specific primers and probes have been designed to amplify the E6, E7 regions of HPV 16. HPV 16 E6 forward primer, 5′-TCAGGACCCACAGGAGCG-3′,. HPV 16 E6 reverse primer, 5′-CCTCACGTCGCAGTAACTGTTG-3′, HPV 16 E6 TaqMan probe, 5′-(FAM)- CCCAGAAAGTTACCACAGTTATGCACAGAGCT-(TAMRA)-3′, HPV 16 E7 forward primer, 5′-CCGGACAGAGCCCATTACAA-3′, HPV 16 E7 reverse primer, 5′-CGAATGTCTACGTGTGTGCTTTG-3′, HPV 16 E7 TaqMan probe, 5′-(FAM)- CGCACAACCGAAGCGTAGAGTCACACT-(TAMRA)-3′, β-actin forward primer, 5′-TCACCCACACTGTGCCCATCTACGA-3′, β-actin reverse primer, 5′-CAGCGGAACCGCTCATTGCCAATGG-3′, β-actin TaqMan probe, 5′-(FAM)-ATGCCCTCCCCCATGCCATCCTGCGT-(TAMRA)-3′. All the samples were run in duplicate. Primers and probes to a housekeeping gene (β-actin) were run in duplicate and parallel to normalize input DNA. Samples in which two results were not concordant were repeated twice in duplicate and were usually due to failed PCR in one of the initial reactions. Each reaction was run 50 cycles. By using serial dilutions, standard curves were developed for the HPV 16 viral copy number using CaSki (American Type Culture Collection, Manassas, VA) cell line genomic DNA, known to have 600 copies/genome (6.6 pg of DNA/genome). Standard curves were developed for HPV 16 E6 and E7, using serial dilutions of DNA extracted from CaSki cells with 50,000 pg, 5,000 pg, 500 pg, 50 pg and 5 pg of DNA. Standard curves were developed as well for the β-actin housekeeping gene (2 copies/genome), using the same serial dilutions of the CaSki genomic DNA. This additional step allowed for relative quantification of the input DNA level and final quantity as the number of viral copies/genome/cell. HPV copy number ≥0.1 copy/cell for tumor samples was regarded as positive.28 For all samples, triplicate control β-actin amplifications of 10 ng of total DNA were positive.
The major statistical endpoint in our study was the use of quantitative PCR-based saliva rinse screening of HPV for the detection of head and neck cancer. Sensitivity, specificity, likelihood ratios and positive predictive values are presented for 3 cutoffs for saliva rinse HPV. The effect of prevalence on positive predictive value is shown.
Factors associated with head and neck cancer and positive HPV tumor tissue were selected based on cross tabulations and logistic regression models, initial factors considered included age, gender, ethnic origin, tobacco exposure, ethanol intake, comorbidity, site, stage, therapy and clinical outcome.31 HPV16 E6 and E7 values in tumor patients were correlated using the non-parametric correlation, Spearman's ρ. Wilcoxon's signed ranks test was used to compare paired tissue and saliva rinse quantities. The Wilcoxon Mann-Whitney test was used to test differences in HPV 16 E6, E7 quantities between HNSC patients and the controls. Event time distributions for disease-free survival were estimated with the method of Kaplan and Meier32 and compared using the log-rank statistic33 or the proportional hazards regression model.34 The simultaneous effect of 2 or more factors was studied using the multivariate proportional hazards model. All p-values are 2-sided and all confidence intervals (CI) are 95%. Computations were carried out using the Statistical Analysis System35 or EGRET.36 Alcohol consumption was not well reported within the HNSC case cohort, but available data were included.
We examined 92 HNSC tumor and paired saliva rinse samples and saliva rinse samples from 604 control subjects from a community screening study for the presence of HPV 16 E6 and E7 DNA using quantitative PCR. Both groups of people were similar with respect to age and gender. The cancer patients included 65 (70.7%) men and 27 (29.3%) women and they ranged in age from 32–83 (mean = 60, median = 61) years. There were 380 (63.1%) males and 222 (36.9%) females in the controls with an age range of 18–92 (mean = 61, median = 62) years.
We found 42 (45.6%) of the 92 head and neck cancer patients with a detectable level of HPV16 (either E6 or E7 ≥0 copies/cell) in their primary tumor tissue. Twenty-eight (30.4%) of the tumor tissues were considered positive (either E6 or E7 ≥0.1 copies/cell). E6 and E7 were usually detected simultaneously in HPV 16 positive tumors, E6 and E7 correlation within tissue samples (tumor patients only, n = 92) was significant, Spearman's ρ = 0.71, p < 0.0001 as well as the E6 and E7 correlation within saliva rinse samples (tumor and controls, n = 696), Spearman's ρ = 0.50, p < 0.0001.
Tables I–III summarizes the head and neck cancer clinical data by tissue HPV16 status. Analysis of the distribution of HPV-positive tumors demonstrated findings consistent with prior studies, HPV-positive tissue was found more frequently in oropharynx tumors 58% (95% CI = 37–77%) when compared to all other sites combined, 19% (95% CI = 11–31%), p = 0.003. There were no significant differences in terms of HPV 16 status for stage or clinical outcome. We did, however, find a higher HPV 16 positive rate for nonsmokers in comparison to smokers. Using the cutoff of HPV16 copy numbers ≥0.1 copies/cell to define tumors as positive, HPV16 was noted in 57% (95% CI = 34–78%) (12/21) of never smoker HNSC patients, 23% (95% CI = 11.8–38.6%) (10/43) of former smoker, and 21% (95% CI = 8.3–41%) (6/28) of the current smoker HNSC patients, p = 0.01. This is consistent with prior observations that HPV is an etiologic agent for HNSC, particularly in a nonsmoking population.
|HPV16 positive||HPV16 negative||p-value|
|>60||13 (37.7)||34 (72.3)||0.55|
|≤60||15 (33.3)||30 (66.7)|
|Male||22 (33.8)||43 (66.2)||0.23|
|Female||6 (22.2)||21 (77.8)|
|I–II||7 (25)||21 (75)||0.37|
|III–IV||21 (32.8)||43 (67.2)|
|Oropharynx||15 (57.7)||11 (42.3)||0.005|
|Oral cavity||6 (15.8)||32 (84.2)|
|Larynx||3 (18.8)||13 (81.3)|
|Hypopharynx||1 (16.7)||5 (83.3)|
|Unknown primary||3 (50.0)||3 (50.0)|
|N+||17 (30.4)||39 (69.6)||0.895|
|N−||11 (30.6)||25 (69.4)|
|Never||12 (57.1)||9 (42.9)||0.01|
|Former||10 (23.3)||33 (76.7)|
|Current||6 (21.4)||22 (78.6)|
|No information||29 (4.8)|
|No information||60 (9.9)|
|Not sure||37 (6.1)|
Quantitative analysis of HPV 16 DNA was also carried out on all paired 92 salivary rinses from patients with HNSC. Thirty (32.6%) saliva rinse samples were HPV 16 positive(>0 copies/cell). There were 16 of 28, 57% (95% CI = 37–76%) HPV16 positive saliva rinse samples(>0 copies/cell) from HPV16 positive HNSC patients compared to 14 of 64, 21.9% (95% CI = 12–34%) HPV16 positive saliva rinse samples from HPV16 negative HNSC patients, p = 0.001. The quantity of HPV16 in primary HNSC, as determined by E6 and E7 levels, was consistently higher than in matched saliva rinse samples from the same patients (Wilcoxon signed rank tests, p = 0.002 and p = 0.03 respectively). For the 28 HPV16 positive tumors, 16/28 (57.1%) had a positive HPV16 in their saliva rinses (>0 copy/cell of E6 and/or E7) (Table IV).
|Saliva rinse HPV 16 positive||Saliva rinse DNA HPV 16 negative|
|Tumor HPV 16 positive||16 (57.1%)||12 (42.9%)|
|Tumor HPV 16 negative||14 (21.9%)||50 (78.1%)|
We noted 17 (2.8%) of the 604 controls had positive HPV16 E6 or E7 DNA in their saliva rinse samples. Two of these controls (0.3%) demonstrated levels >0.1copies/cell. Table II shows the demographic characteristics and risk factors of controls.
We analyzed HPV 16 E6 and E7 copy numbers of the 3 groups of samples: saliva rinse from HPV 16 positive HNSC patients, saliva rinse from HPV-negative HNSC patients, and saliva rinse from normal controls. This showed significant differences between saliva rinse from control subjects and saliva rinse from HPV 16 positive patients (p < 0.001, Mann-Whitney test). These results indicated that an elevated HPV 16 DNA level in saliva rinse is found in saliva rinse from tumor patients when compared to normal controls (Fig. 1). We also found there was no significant difference in HPV status between saliva rinse from patients with HNSC and HPV status of tumor tissue (p = 0.97). The HPV-positive rate for primary HNSC, saliva rinse from HNSC patients and normal control saliva rinse samples were 30.4%, 32.6% and 2.8% respectively (p < 0.0001) (Table V).
|Cutoff copies/cell||HPV16 positive case saliva||HPV16 positive control saliva||Sensitivity 95% CI||Specificity 95% CI||Likelihood ratio positive 95% CI||Positive predictive value 95% CI|
|>0||32.6% (30/92)||2.8% (17/604)||32.6 (22.8, 41.8)||97.2 (95.9, 98.5)||11.5 (6.6, 19.9)||63.8 (50.1, 77.6)|
|≥0.001||30.4% (28/92)||1.3% (8/604)||30.4 (20.8, 39.4)||98.7 (97.8, 99.6)||22.7 (10.7, 48.4)||77.8 (64.2, 91.4)|
|≥0.1||14% (13/92)||0.33% (2/604)||14.0 (6.9, 21.0)||99.7 (99.2, 100)||42.2 (9.7, 184.1)||86.7 (69.5, 103.9)|
Saliva rinse HPV16 sensitivity, specificity, positive likelihood ratio and positive predictive value for head and neck cancer using control subjects and based on 3 cutoff values for HPV copy/genome is given in Table IV. Positive predictive value, the probability that the disease is present given a positive test, is a property of the test as well as the population it is tested on and changes based on the prevalence of the disease in a screened population. Table V shows the effect of prevalence on the positive predictive value of saliva rinse HPV16 given the sensitivity and specificity observed in our study. The number of true- and false–positives is based on a screened population of 10,000. Only when specificity is very high (>98%), and the prevalence is ≥5%, are the number of true positives greater than the number of false positives.
Factors significantly increasing the probability of head and neck cancer included saliva rinse HPV >0 (odds ratio [OR] = 16.44; 95% CI = 8.6 31.5; p < 0.0001), saliva rinse HPV ≥0.1 copies/cell (OR = 48.9; 95% CI = 10.8, 220.5; p < 0.001), and current smoking (OR = 5.6; 95% CI = 3.5, 9.0; p < 0.0001). Males were slightly more likely to have head and neck cancer (OR = 1.4; 95% CI = 0.9, 2.2; p = 0.19). A multivariate regression model using smoking status and HPV presence in saliva rinse as variables demonstrated that status as a current smoker (OR = 5.0; 95% CI = 2.9–8.3; p < 0.0001) and presence of any detectable level of HPV 16 in saliva rinse (OR = 15.8, 95% CI = 7.8–32.2, p < 0.0001) were associated independently with HNSC.
Within the group of head and neck cancer patients, factors found to have an association with an HPV-positive tumor included HPV-positive saliva rinse HPV, and a site of origin of HNSC in the oropharynx site when compared to the normalized oral cavity site. Cancer patients with any detectable HPV in saliva rinse samples were 4.9 (95% CI = 1.89, 12.6) times as likely to have HPV-positive tumors, p = 0.001. Saliva rinse HPV ≥0.1 copies/cell increased the probability of HPV-positive tumor tissue as well (OR = 7.2; 95% CI = 2.0, 26.1; p = 0.003). Compared to the oral cavity group, patients with the oropharynx site were 7.5 (95% CI = 2.3, 24.1) times as likely to have HPV-positive tumors, p = 0.001. Former (OR = 0.21; 95% CI = 0.06, 0.7; p = 0.01) and current (OR = 0.23; 95% CI = 0.07, 0.7; p = 0.01) smokers were about 20% as likely as never smokers to have HPV-positive tumors. Stage of disease, gender and age were not associated with HPV-positive tumors.
A consensus has developed recently that high-risk HPV infection is a potential risk factor for head and neck cancer.7, 8, 13, 15, 37 The aim of our study is to determine the association between high-risk HPV 16 subtype presence in saliva rinse from patients with HNSC and normal control subjects, using a quantitative PCR-based method to investigate the feasibility of screening using this technique.
In our study, the HPV 16-positive rate for HNSC is 30.4% (28/92) and is similar to other reports.12, 38, 39 We found a similar site specific preponderance of HPV 16 presence in oropharynx tumors as reported by other investigators.20, 21, 40, 41, 42, 43 In addition, we found that the rate of HPV presence differed between smoker and nonsmokers with HNSC, HPV 16 infection rate in nonsmokers is 57.1%, higher than smokers (22.5%; p = 0.003). This confirms other reports.
Other investigators have been able to detect HPV 16 DNA in saliva rinses from patients with HPV-related HNSC.27, 29 Smith et al.30 have found an independent association of high-risk HPV presence in oral exfoliated cells with HNSC, using nonquantitative PCR techniques, as well as a synergistic relationship between alcohol consumption and HPV. Perhaps due to incomplete information regarding alcohol consumption in HNSC patients in our study, we did not demonstrate this synergistic relationship. Similar to Smith et al.,29 we found the HPV DNA detection in tumor tissue and saliva rinse from patients with tumors demonstrated a tight correlation (p < 0.001, Table IV). We noted that some tumors that were HPV-positive did not yield HPV-positive saliva rinse, and that the saliva rinse level of HPV 16 was invariably lower than that of the tumor, presumably due to the diluting effect of normal exfoliated upper aerodigestive tract cells unrelated to the primary tumor. Statistical analysis showed HPV infection was not only related to the presence of head and neck cancer but that higher HPV 16 DNA levels in saliva rinse were associated with the presence of cancer. In addition we noted that we were able to detect more HPV 16 E7 in the saliva rinse of HPV-positive HNSC patients, HPV-negative HNSC patients, and controls in comparison to the same groups when using HPV 16 E6 to detect HPV. This may be due to improved sensitivity of the E7 probe and primer set, in that very low level HPV may have been detected due to infection of other cell populations other that tumor. Other more rare, high-risk, HPV types were not assessed by our analysis.
Note that our control population had a statistically significant difference in smoking status when compared to HNSC patients. The risk associated with HPV-positive salivary rinse was found to be independent of smoking status on multivariate analysis.
In contrast to our data and data reported by Smith et al.29 recent investigation from Herrero et al.21 have examined the presence of HPV 16 DNA in oral cavity brushings using conventional PCR and did not find a correlation between HPV presence in oral brushings and HPV status of head and neck tumors. Herrero et al.21 collected brushings only from the oral cavity, did not sample the soft palate or any part of the oropharynx, and did not have subjects gargle or spit a rinsing solution. Because the majority of HPV-positive tumors are located in the oropharynx, it is likely that the lack of correlation between HPV status in oral brushings and tumor noted in Herrero et al.21 is a reflection of their sampling method, and that oropharynx and other upper aerodigestive exfoliated cells were not well sampled.
We chose to use a quantitative method of HPV 16 detection to examine the possibility that application of threshold criteria for detection of HPV in saliva rinse could improve the specificity of the assay. Application of a threshold of 0.001 copies/cell in saliva rinse yielded minimal decreases in sensitivity, from 32.6–30.4%, with an improvement in specificity, from 97.2–98.7 % (Table V). Screening strategies are useful for diseases like HNSC with high morbidity or mortality for which effective therapies for early stage disease exist, and ideally with a high prevalence in the screening population. When screening a population in which the prevalence of the disease is very low, specificity is particularly important to avoid false–positive results. The incidence of HNSC in the general population is low, with approximately 50,000 new cases diagnosed in the United States this year, of which approximately 25% are HPV-associated. The incidence of HNSC is higher in subpopulation with exposure to causative factors for HNSC. Table V tabulates false-positives for four different prevalence levels of HNSC, holding sensitivity and specificity constant.
From the results of our study, no more than 45.6% of head and neck cancer patients have any detectable level of HPV16 in their tumor tissue, although some of these patients had a very low level of HPV 16 copy number/genome. It is possible that some of these tumors with minimal amounts of HPV 16 genome are not caused by HPV 16 genomic integration in a dominant clonal population, and tumors with low HPV copy numbers may not be caused by HPV 16. Based on the sample in our study, even if HPV16 were present in 100% of the saliva rinse samples taken from these tumor patients with any detectable HPV16 in their tumor tissue, the overall sensitivity of this test for head and neck cancer could not be greater than about 45.6%. As one would expect, HPV16 levels were consistently lower in the saliva rinse than in tumor tissues from the same patients. Using any detectable saliva rinse HPV16 as cutoff, only 57% of the positive tumor patients had detectable saliva rinse HPV16, which resulted in a sensitivity of 32.6% for the saliva rinse test for head and neck cancer.
The sensitivity of the saliva rinse test for head and neck cancer was moderately reduced, from 32.6% to 30.4% as the cutoff for determining positivity was increased from 0 to 0.001 copies/cell. Specificity, however, was only increased slightly, from 97.2% to 98.7% (Table V). Requiring ≥0.1 copies/cell in the saliva rinse for a positive test further decreased the sensitivity to 14% and increased the specificity to 99.7%. Even with this specificity, a prevalence of 5% or greater would be needed in the screened group to have a positive predictive value close to 70% (Table VI). The number of true positives to false–positives in this case would be slightly greater than 2:1.
|Cutoff copies/cell||Sensitivity %||Specificity %||Prevalence %||PV+ %||True positives||False positives|
It is notable that cancer cases were not drawn from the community screening that provided control patients. Therefore, controls may not purely represent the study population from which the cases arose, potentially introducing some selection bias, distorting the distribution of measured factors and the effect estimates.
We also noted no association with stage in terms of ability to detect HPV in saliva rinse from patients with HPV-positive tumors, or with the quantity of HPV genome in salivary rinses. This indicates that early and late stage tumors are detected with equal facility using this assay, indicating that early stage detection remains possible.
For saliva rinse screening of HPV for head and neck cancer to be feasible, identification of a group of higher risk patients with prevalence close to 5%, and a more sensitive test that also remains very specific, ≥ 99%, and more sensitive, would be optimal. Additional mechanisms to achieve this goal could include limited screening of a high-risk group based on factors associated with HPV infection, including high-risk sexual activity. This screening could be used concomitantly with other screening tests based on detection of disease in populations with other exposures associated with HNSC, including tobacco use. These tests could include other molecular markers as well as markers of exposure. Another consideration includes developing plans for further testing after an initial positive result. For example, retesting patients after an appropriate interval may be able to eliminate those patients who demonstrate a false–positive due to an HPV 16 infection without presence of an epithelial neoplasm. This test could be combined with other detection methods directed at detecting molecular alterations in saliva rinse that are specific to HPV-negative HNSC, particularly methods used to detect p53 gene alteration or mutation.
Application of a high throughput test for detection of high-risk HPV in salivary rinses has potential for development of molecular screening for HPV-related HNSC. In its current form, however, even 99% specificity would result in significant expense and a high rate of false–positive results that make this test difficult to apply to a broad population. Additional methods of defining a high-risk population that is smaller in number, or increasing the specificity of this test would be helpful for broad application.
Dr. Califano is a Damon Runyon-Lilly Clinical Investigator supported by the Damon Runyon Cancer Research Foundation (CI-#9), a Clinical Innovator Award from the Flight Attendant Medical Research Institute, and the National Institute of Dental and Craniofacial Research (1R01DE015939-01).
- 31Regression models and life-tables (with discussion). J R Stat Soc B 1972; 34: 187–220..
- 33The analysis of binary data. London: Methuen, 1970..
- 35SAS Inc. SAS user's guide: Statistics version 5,. Cary, NC: SAS Institute Inc., 1985.
- 36SaER Corporation. EGRET users' manual. Seattle, Washington: 1988.