Each year nearly 30,000 Americans are diagnosed with oral cancer. 90% of these lesions are oral squamous cell carcinomas 1. Despite advances in surgery, radiation and chemotherapy, the five-year survival rate is 54%, one of the lowest of the major cancer sites, and this rate has not improved significantly in recent decades 234. Worldwide, the problem is much greater, with over 350,000 to 400,000 new cases being found each year 5. The disease kills one person every hour – more people than cancers of the cervix, brain, ovary, testes, liver, kidney, malignant melanoma or Hodgkin's lymphoma 56. In the United States, African American males suffer the highest incidence and lowest survival rates of any group. From 1985 to 1996, the five-year survival rate for tongue carcinoma in African-American men was 27%, compared with a 47% five-year survival rate among white men 7. In 2001, similar five-year survival rates were found in a study of oral and pharyngeal cancer among African-American and White men 8. Notably, incidence in young adults (<40 years) is increasing in the U.S. 910 and worldwide 1112.

Early detection followed by appropriate treatment, can increase cure rates to 80 or 90%, and greatly improve the quality of life by minimizing extensive, debilitating treatments 513. Despite the accessibility of the oral cavity to direct examination, these malignancies are often not detected until a late stage 51415. Oral cancer is unusual in that it carries a high risk of second primary tumors. Patients who survive a first cancer of the oral cavity have up to a 20-fold increased risk of developing a second primary oral cancer and that risk lasts 5–10 years and sometimes longer 16.

Major risk factors for oral cancers in the United States are use of tobacco and alcohol, which account for 75 to 80% of all oral cancers 517. Although tobacco is a well-recognized risk factor for OSCC, the public is generally unaware that alcohol synergizes with tobacco. Those who both smoke and drink have 15 times the risk of developing oral cancer 5. Notably, some oral cancer patients have no known risk factors, and the disease in this population may pursue a particularly aggressive course 18.

The American Cancer Society recommends that doctors and dentists examine the mouth and throat during routine examinations 2 as early cancer lesions are often asymptomatic and may mimic benign lesions 1920. General population screening, however, has not been shown to reduce the incidence of and mortality from oral cancer. The reasons include the low prevalence and incidence of OSCC, the potential for false-positive diagnoses and poor compliance with screening and referral 621. Thus the National Institute of Dental and Craniofacial Research and The Oral Cancer Foundation have recommended that research efforts focus on developing novel detection techniques 516.

Studies have reported that certain common oral bacteria are elevated on or in oral and esophageal cancer lesions and their associated lymph nodes 22232425262728. Although increased colonization of facultative oral streptococci have been reported most often 24252627, anaerobic Prevotella, Veillonella, Porphyromonas and Capnocytophaga species were also elevated 252628. Currently, studies are examining whether bacteria may be incidentally or causally associated with oral cancer. Additional research is determining whether various salivary markers may be used as early diagnostic indicators for oral cancer.

The reason for these shifts in bacterial colonization of cancer lesions is unclear. Mechanistic studies of bacterial attachment provide some insights, however. Research has repeatedly shown that oral bacteria demonstrate specific tropisms toward different biological surfaces in the oral cavity such as the teeth, mucosa, and other bacteria 29303132333435. The non-shedding surfaces of the teeth offer a far different habitat than the continually shedding surfaces of the oral mucosa. Due to the repeated shedding of epithelial cells, there is less time for a complex biofilm to develop on soft tissue surfaces; thus, a premium is placed on potent mechanisms of adhesion. The differences in bacterial tropisms for specific oral sites suggest that different intra-oral surfaces and bacterial species have different receptors and adhesion molecules that dictate the colonization of different oral surfaces.

It is now recognized that bacteria bind to and colonize mucosal surfaces in a highly selective manner via a "lock- and key" mechanism. Adhesins on bacteria bind specifically to complementary receptors on the mucosal surfaces of the host. These adhesins differ from species to species leading to specificity in attachment to different surfaces. Studies have shown that even within genera, colonization patterns of individual species may differ markedly 29303132. Streptococcus salivarius, for example, preferentially colonized the oral soft tissues and saliva compared to the teeth, while the reverse was true of Streptococcus sanguis.

Cancer has been referred to as a molecular disease of cell membrane glycoconjugates, 363738. Certain glycoconjugates serve as receptors for specific bacteria and recent reports support the notion that shifts in the colonization of different cancer cells are associated with observed changes in cell surface receptors 364041. An in vitro study of S. sanguis, a common oral inhabitant, demonstrated that its binding capacity to normal exfoliated human buccal epithelial cells (HBEC) depended upon the availability of surface sialic acid residues 36. Desialylation of HBEC invariably abolished adhesion of S. sanguis to these epithelial cells. In similar experiments carried out with a buccal carcinoma cell line, S. sanguis did not reliably attach. It was determined that the tumor cells did not express the sialylated membrane glycoprotein of normal cells suggesting that changes in the surface receptors had occurred in the buccal carcinoma cell line.

In a previous study of 225 OSCC-free subjects we found a high degree of specificity in the "preferred" intra-oral localization of species, even within a single genus such as Streptococcus 42. This specificity in localization of individual species agreed with that described in previous studies. Our investigation extended earlier findings by describing the distribution of multiple species within the same genus on a wider range of intra-oral surfaces. For example, S. oralis, S. constellatus, S. mitis, S. intermedius and S. anginosus colonized the soft tissues in higher proportions than the teeth; however, their "preferred" soft tissue habitats differed. S. sanguis colonized different soft tissue locations in similar proportions, but was found in higher mean proportions on the teeth, particularly in the supragingival plaque.

The availability of a large amount of data from the OSCC-free subjects permitted this group to be subset according to periodontal and smoking status and the colonization patterns on the soft tissues compared among groups 43. The clinical parameters among the populations were in accord with those found in previous studies and results were similar to previous investigations 44454647. Few differences were found in the salivary or soft tissue microbiota among the subset populations. It was concluded that the presence or absence of periodontal infections or a smoking habit had minimal effects on salivary and soft tissue colonization. These findings were in accord with studies by Danser et al. 1996 and Lie et al. 1998 4849 but contrasted with earlier reports by Colman et al. 1976 and van Winkelhoff et al. 1986 5051. Importantly, we found that when the microbiota of teeth, soft tissues and saliva were compared, the microbial profile of saliva was similar to that of the soft tissues, but saliva and soft tissue colonization differed markedly from that of dental plaque. These findings were similar to those of other investigations 465152.

As previously mentioned, studies have reported that the microbiota of OSCC lesions differs from that found on the soft tissues of OSCC-free individuals. Little was known, however, about the salivary microbiota of oral cancer subjects. Thus, the purpose of the present investigation was to determine whether the salivary microbiota in subjects with an oral squamous cell carcinoma (OSCC) lesion would differ from that found in OSCC-free controls.

Results from this investigation demonstrated that oral cancer subjects had elevated counts (p < 0.001) of C. gingivalis, P. melaninogenica and S. mitis in saliva compared to OSCC-free subjects. These results are borderline in significance after adjusting for multiple comparisons. However, when each species was ≥0.4 × 10⁵ they indicated the presence of an OSCC lesion with 80% diagnostic sensitivity and ≥82% specificity in both matched and unmatched populations.

The reason for this finding is unclear. One explanation may relate to the altered cell surface receptors observed in cancer cells 363941. It seems reasonable that alterations in tumor cell receptors could change the adhesion of certain species of bacteria. This was shown, as previously discussed, in an in vitro study of HBEC and buccal cell carcinoma cell lines using the common oral bacterium S. sanguis by Neeser 36. One might expect that as Neeser found decreased colonization of S. sanguis, a similar study using S. mitis would result in a reduced colonization of oral cancer cells. Interestingly, our previous investigation of 225 OSCC-free subjects provided evidence to the contrary. Colonization of different oral sites differed among the 40 test species, even among those of the same genera, such as streptococci. For example, S. sanguis and S. mitis both colonized the oral soft and hard tissues; however, marked differences in their proportions at these sites were noted. Highly species-specific oral colonization by streptococci has been reported by other investigators 293031325758.

Saliva was found to be similar in microbial profile to the soft tissues. This was a significant finding from the study of the OSCC-free population. In contrast, the microbiota of the teeth and saliva differed markedly. These results agreed with previous studies 305758. Thus, if alterations in bacterial adhesion to OSCC cells observed in vitro exist in vivo, colonization of OSCC lesions would be affected. Shifts in the soft tissue microbiota of the oral cavity appear likely to affect salivary levels as well.

A screening test for oral cancer based on salivary counts of bacterial species is appealing. Saliva is now meeting the demand for inexpensive, noninvasive, and easy-to use diagnostic aids for oral and systemic diseases, and for assessing risk behaviors such as tobacco and alcohol use. Detection of HIV by the presence of virus-specific antibodies in saliva, for example, has led to the development of commercially available test kits 16. If increased numbers of certain salivary species are shown to be a signature of oral cancer, an early diagnostic test for OSCC may be developed, reducing the morbidity and mortality of this devastating cancer.

Studies to examine the validity of these findings are planned. If the results of this study are validated it will be important to address whether oral bacteria can be used as indicators of oral cancer and whether certain oral species contribute to carcinogenesis.

Results of the present study suggest that high salivary counts of C. gingivalis, P. melaninogenica and S. mitis may be diagnostic indicators of OSCC. These findings taken with those of an earlier study indicate that the presence of an OSCC has a more powerful effect on the salivary microbiota than either smoking or periodontal infections.

DLM conceived of this investigation and the preliminary studies, constructed the study design, developed clinical sampling techniques, wrote hospital protocols and K-23 grant application that funded the project, collected and processed the majority of OSCC subject samples, modified laboratory protocols and drafted the manuscript. ADH made substantial contributions to the conception and study design. She coordinated the study of OSCC-free subjects, conducted the statistical analysis of the OSCC-free data and made substantial contributions to the interpretation of the data for the OSCC-free and OSCC populations. PMD, CMN and MRP made substantial contributions to the conception, design and coordination of the study during the preliminary studies and initiation of this investigation and were instrumental in coordinating clinical evaluations with the recruitment and sampling of OSCC subjects. JMG was instrumental in the conception and design of the study and performed the majority of statistical analyses and interpretations of data for the OSCC subjects. JMG made critical revisions for important intellectual content of the manuscript. All authors have given final approval of the manuscript.