Exploring biomarkers in head and neck cancer


  • Corey J. Langer MD

    Corresponding author
    1. Abramson Cancer Center, University of Pennsylvania, Philadelphia, Pennsylvania
    • Division of Hematology and Oncology, University of Pennsylvania, 3400 Civic Center Blvd, 2PCAM, Philadelphia, PA 19104
    Search for more papers by this author
    • Fax: (215) 615-5122


This article is corrected by:

  1. Errata: Erratum: Exploring biomarkers in head and neck cancer Volume 118, Issue 23, 6014, Article first published online: 8 May 2012


Personalized medicine based on predictive markers linked to drug response, it is hoped, will lead to improvements in outcomes and avoidance of unnecessary treatment in squamous cell carcinoma of the head and neck (SCCHN). Recent research has shown that expression of ERCC1 may predict resistance to treatment with platinum agents. Future testing for this marker may help select the optimal type of chemotherapy. Infection with human papillomavirus (HPV) is associated with less aggressive disease and better prognosis in locally advanced SCCHN treated with chemoradiation or radiation alone; HPV-positive patients may ultimately benefit from less intensive, less toxic therapy. K-RAS mutations, occurring in about 40% of colorectal cancers and associated with lack of benefit from epidermal growth factor receptor (EGFR) antibodies in this disease, are found in <5% of SCCHN patients, making routine testing for K-RAS mutations unwarranted at this time. Virtually all head and neck tumors overexpress EGFR, which limits the usefulness of EGFR expression as a marker for treatment selection. Although the incidence of EGFR tyrosine kinase domain mutations is very rare, a better understanding of the role of EGFR mutations, expression, amplification, and downstream effects in SCCHN may help define the role of EGFR in this setting. These observations caution against extrapolating results obtained with biomarkers in other types of cancer to SCCHN. Validation of each biomarker in the context of SCCHN clinical trials will be required before a specific marker can be incorporated into daily practice. Cancer 2012;. © 2012 American Cancer Society.

An estimated 49,260 new cases of squamous cell carcinoma of the head and neck (SCCHN) were diagnosed in the United States in 2010.1 Most patients diagnosed with SCCHN present with locally advanced disease, for which a combined modality treatment approach with curative intent is usually prescribed. Cure rates with current combined-modality strategies are favorable, but therapies are indiscriminately aggressive, often resulting in severe short-term toxicity and long-term functional impairment. Thus, a clear role for better patient selection to avoid unnecessary treatment and minimize long-term toxicity is warranted.

By tailoring treatment to the specific genetic or molecular profile of a tumor in an individual patient, response and survival outcomes could be improved through refined treatment decisions, treatment-related costs and complications reduced, and unnecessary interventions avoided. Such refinements may also facilitate the development of newer, more effective therapies. The understanding of biological markers in SCCHN is not as well defined as in other tumor types (Table 1),2 such as breast or lung cancer, but recent advances have brought us closer to providing personalized medicine for these patients. The following review provides an overview of some of the most relevant biomarkers related to SCCHN.

Table 1. Most Relevant Biological Alterations in Head and Neck Cancer by Frequency2
  1. Abbreviation: HPV, human papillomavirus.

Other markers
 ERCC1 high levels71%-73%


The ERCC1 (excision repair cross-complementation group 1) protein plays an important role in repairing DNA damage caused by platinum agents, and may therefore be useful in predicting which patients will benefit from platinum-based therapy. Platinum agents work by creating DNA adducts that inhibit DNA replication and transcription in cancer cells.3 Cells that repair the damage efficiently are more likely to be resistant to platinum therapy.4, 5 Removal of cisplatin-DNA adducts is mediated by the nucleotide excision repair (NER) complex, a multistep process involving >30 distinct proteins that recognize DNA damage, incise the lesion, resynthesize DNA, and ligate the repair patch; 1 of the key enzymes in this pathway is ERCC1.6, 7 Several groups have linked low ERCC1 expression to platinum sensitivity in vitro,8, 9 and some have shown that blocking the expression or activity of ERCC1 increases sensitivity to platinum agents.10-12 For example, Selvakumaran et al11 demonstrated that by modifying platinum-resistant ovarian cancer cells to constitutively express antisense ERCC1 RNA, the cells were made to inhibit the capacity to repair DNA damage and consequently reduce IC50 values for platinum agents. The study also showed that mice transplanted with antisense cancer cells lived longer after cisplatin treatment than those bearing control cells.

In primary tumor samples, pretreatment levels of ERCC1 have been shown to predict outcomes after platinum therapy in patients with ovarian cancer,13 gastric cancer,14 colorectal cancer,15 esophageal cancer,16 and nonsmall cell lung cancer (NSCLC).17-19

In lung cancer, Olaussen et al19 assessed whether ERCC1 status predicted outcome after platinum-based adjuvant chemotherapy in patients with completely resected NSCLC. ERCC1 status was determined in 761 samples (of the 1867 patients originally enrolled in the International Adjuvant Lung Cancer Trial) using a semiquantitative H score based on the diffuseness and intensity of ERCC1 staining by immunohistochemistry (IHC); the median H score was used as the cutoff between ERCC1-positive (44%) and ERCC1 negative (56%) disease. Platinum-based adjuvant chemotherapy was found to be beneficial in patients with ERCC1-negative tumors, but not in those with ERCC1-positive tumors. In the control group, which did not receive adjuvant chemotherapy, 5-year survival was significantly greater in patients with ERCC1-positive disease than in those with ERCC1-negative disease (46% vs 39%). Hence, ERCC1 was both prognostic and predictive. Subsequent to this study, the reliability and specificity of the key reagent used to measure ERCC1 expression, the antibody 8.F1, has been called into question, which would cast doubts over the robustness of these results.20

Cobo et al21 conducted 1 of the first prospective randomized trials to select treatment based on ERCC1. In this study, patients with stage IV NSCLC were randomized to a control group, which received standard therapy with cisplatin/docetaxel, or an experimental genotypic group, in which patients with low ERCC1 expression, this time measured by mRNA expression, were treated with cisplatin/docetaxel and those with high ERCC1 expression received a nonplatinum-containing regimen (docetaxel/gemcitabine). Among the 346 evaluable patients, the response rate was markedly increased in the genotypic group (50.7 vs 39.3% in the control group; P = .02); within the genotypic group, response rate was 53.2% for patients with low ERCC1 and 47.2% for the high ERCC1 subgroup. However, no difference in survival was observed between the 2 treatment groups, although the median follow-up time was short (9 months). Although these results are encouraging and suggest the potential to base treatment decisions on ERCC1 expression levels in the future, the high dropout rate (17.6%), due primarily to insufficient tissue available for ERCC1 mRNA analysis, suggests that better protocols for ERCC1 testing are needed.

ERCC1 in head and neck cancer

Preliminary studies suggest ERCC1 has a role in predicting patient response to cisplatin-based therapy in SCCHN.22, 23 Handra-Luca et al22 assessed the impact of ERCC1 expression level (measured by IHC with the aforementioned 8.F1 reagent) on outcomes after cisplatin-based chemotherapy in patients with locally advanced disease. Of the 96 pretreatment tissue samples analyzed, 68 (71%) had high expression of ERCC1. Patients with low levels of expression had a 4-fold greater chance of achieving an objective response to cisplatin-based chemotherapy compared with those with high levels of expression. After adjusting for age, disease stage, tumor differentiation, and location, low ERCC1 expression was a significant predictor of prolonged cancer-specific survival. These findings suggest that pretreatment ERCC1 expression levels are inversely correlated with response and survival after cisplatin-based therapy.

Jun et al23 evaluated whether pretreatment levels of ERCC1 expression (also by 8.F1-based immunostaining) predicted survival in patents with locally advanced SCCHN undergoing cisplatin-based concurrent chemoradiotherapy. ERCC1 expression was measured using IHC in formalin-fixed, paraffin-embedded tissue blocks from 45 patients. Most tumors (73%) had high ERCC1 expression. With a median follow-up of 53.6 months, the estimated 3-year progression-free survival (PFS) was significantly longer in those with low ERCC1 expression compared with those with high expression (83.3% vs 49.4%). The 3-year overall survival (OS) was also significantly greater in those with low ERCC1 expression (91.7% vs 45.5%). In a multivariate analysis, low ERCC1 expression was a significant and independent predictor of prolonged survival.

Notably, the prevalence of ERCC1 overexpression in both SCCHN studies was similar (71%-73%) and considerably higher than that reported in NSCLC (approximately 40%)19 using a similar scoring technique. This suggests that the proportion and pattern of ERCC1 expression may vary by tumor type. Moreover, the response rates observed in these studies suggest that a proportion of patients with high ERCC1 expression may still respond to platinum therapy. Thus, the sensitivity of ERCC1 as a predictor of platinum response in SCCHN appears to be relatively low, and other factors—within the NER pathway or beyond—may also influence tumor response. Although the argument could be made that in these studies cisplatin was combined with 5-fluorouracil or taxanes, and therefore the magnitude of the ERCC1 effect could be at least partially confounded by the use of multiple chemotherapeutic agents, the differences in outcomes between the ERCC1 groups in the Jun study were consistent regardless of the chemotherapeutic agent partnered with cisplatin.22

From a practical perspective, it appears that high ERCC1 expression in SCCHN tumors may suggest a low probability of benefiting from platinum therapy. Patients with this marker may be candidates for alternative approaches, including nonplatinum therapy. However, further validation is needed, including prospective trials that evaluate strategies for assigning treatment in SCCHN based on ERCC1 expression status. The most common method for evaluating ERCC1 mRNA expression is quantitative real-time reverse transcription polymerase chain reaction (PCR), which requires the collection of fresh formalin-fixed, paraffin-embedded tumor samples.13-15, 17 Commercial kits are available for testing ERCC1 expression (via IHC or PCR-based techniques) in other tumor types, such as lung and colon cancer, but not in SCCHN. Samples must be taken from tumors that have not been irradiated (because fibrosis or radiation-induced necrosis may interfere with interpretation), a requirement that may hinder the potential application of ERCC1 as a biomarker in SCCHN.

In addition to the effect of ERCC1 levels on response to platinum therapy, certain polymorphisms on the ERCC1 gene may influence response of patients to radiotherapy. In a study of 108 patients with stage II SCCHN treated with radiotherapy only, Carles et al24 identified the homozygous genotype Thr259Thr as significantly associated with worse outcomes, for both time to progression (median, 11.6 months vs >85 months for genotypes Lys259Thr or Lys259Lys; P = .00005) and OS (median, 27.9 months vs >88 months; P = .0089). This same study evaluated other genes encoding DNA repair-related proteins, such as ERCC5 or XPA. Apart from ERCC1, only 1 of the ERCC5 polymorphisms showed a significant correlation with outcome; patients with homozygous T/T nucleotide at codon position 46 had shorter time to progression (median, 56.7 months vs >81 months for those with C/T or C/C; P = .049) and worse survival (median, 61.0 months vs >87 months; P = .0066). If validated, these findings may prove particularly relevant to clinical decision making in a setting where there may be a choice between radiation and surgical management.


The RRM1 (ribonucleotide reductase M1) gene encodes a regulatory subunit of the enzyme ribonucleotide reductase that is the key molecular target of gemcitabine.18, 25, 26 Preclinical studies have shown that high levels or specific mutations on RRM1 are associated with gemcitabine resistance.3, 27-30

Clinically, low RRM1 expression or single nucleotide polymorphisms have been found to correlate with better outcomes in patients treated with gemcitabine-based chemotherapy in several solid tumors.18, 29-31

The role of RRM1 in determining chemosensitivity in SCCHN is unclear. However, its correlation with ERCC1 expression and platinum sensitivity (although weaker than its association with gemcitabine sensitivity) suggest that the role of this biomarker in SCCHN should be explored further. Studies evaluating not only RRM1 but multiple related genetic markers, such as RRM2 and BRCA1, may also provide a more comprehensive picture of the role of RRM1 in predicting treatment outcomes.32

β-Tubulin Isotypes

Taxanes exert their anticancer activity by binding to β-tubulin polymers and inhibiting microtubule depolymerization and mitotic progression. In preclinical models, aberrant expression levels of the isotype III of β-tubulin have been associated with resistance to paclitaxel in several cell types.33, 34

In the clinic, this relationship has been explored mostly in NSCLC. Monzo et al35 showed the association between tubulin mutations and lack of response and inferior outcomes in 49 patients with advanced NSCLC treated with paclitaxel. More recently, Seve et al36 and Dumontet et al37 have provided consistent results in 2 groups of patients (91 patients and 19 patients, respectively) treated with taxane-based therapy (paclitaxel or docetaxel in combination with platinum, or docetaxel as a single agent). Patients were classified as low or high β-tubulin-III expressing, according to whether their levels of expression (by IHC) were below or above the median value in the entire population37 or whether the number of stained cells in the sample was < or >50%.36 Regardless of the nuances of the scoring system, the low-expressing patients exhibited higher response rates and longer PFS and OS.

In SCCHN, docetaxel has become part of the standard induction regimen in combination with cisplatin and 5-fluorouracil. A correlative study to Taxotere® (TAX) 324,38 the phase 3 trial comparing induction with 5-fluorouracil to induction with cisplatin and 5-fluorouracil alone indicated that expression of β-tubulin II measured by IHC (samples were scored by percentage of stained cells and intensity of staining; positive results were those above the median score and negative those below) may be associated with outcomes. In the subset of evaluable patients (265 of 501 patients enrolled), patients with low expression had better PFS and OS, regardless of treatment with or without docetaxel (OS hazard ratio [HR], 2.39; P = .0001), suggesting a potential prognostic effect. Moreover, for this same subgroup with low expression, the benefit from the specific addition of docetaxel to the induction therapy regimen seemed to be greater, whereas in the high-expression group, there were modest or no differences in outcomes between the 2 induction regimens.38

As the use of taxanes in SCCHN grows, this marker may assume a greater role in therapeutic decision making. To date, there are no commercial assay kits to determine expression of β-tubulin isotypes, and as described above, there is some variation in the scoring systems reported in the literature, indicating that a potential practical application may have to first overcome some standardization issues.

Human Papillomavirus

Virtually all cases of cervical cancer, and a subset of SCCHN tumors, are linked to human papillomavirus (HPV) infection.39, 40 HPV-positive tumors in SCCHN are almost always oropharyngeal, and the reported HPV infection prevalence varies widely, depending on the method of detection used. About 35% of overall head and neck tumors have been reported to be HPV positive, as detected by PCR.40 Specifically, HPV has been detected most often in cancers of the tonsils (43.6%) and the base of the tongue (38.4%).41 One of the largest and most recent analyses of patients with oropharynx cancer from RTOG (Radiation Therapy Oncology Group) 0129 demonstrated a 64% incidence of HPV-positive tumors using in situ hybridization techniques, suggesting that the correlation of this infection and oropharynx cancer may be greater than previously appreciated.42

The prevalence of HPV-related SCCHN may also vary by geographic region. A global survey of oropharynx cancer found that the prevalence of HPV infection was 18%43; another report indicated the prevalence was 38% and notably higher in North America (47%) than in Europe (28%).43 Furthermore, the prevalence of types of SCCHN potentially related to HPV infection appears to be increasing in the United States.40 The increase in HPV-related tumors (eg, base of tongue, tonsil, oropharynx) has been noted in recent decades, particularly among younger-age cohorts, whereas the incidence of cancers typically unrelated to HPV (eg, tongue, gum, lip, other oral cancers) has decreased. These trends may be explained in part by changes in both sexual behavior and smoking.

Of those head and neck tumors that contain HPV DNA, up to 90% present with the oncogenic variant HPV-16.44-46 Compared with tumors of the oropharynx and oral cavity, tonsillar squamous cell carcinoma has been most strongly and consistently associated with HPV-16 infection; in these tumors, molecular analyses have revealed variable HPV copy numbers, compared with low copy numbers in other sites.46 Hence, a causative relationship between HPV and SCCHN is plausible, if not likely. HPV-16 has been shown to immortalize epithelial cells of both cervical and oral origin in vitro.47, 48 This effect may be mediated by inactivation of the tumor suppressor proteins p53 and pRb.49

Numerous studies have now shown that HPV infection is an independent risk factor for SCCHN. In an epidemiological study of nearly 900,000 serum samples from healthy individuals in Norway, Sweden, and Finland, 292 cases of SCCHN were documented, occurring after a median of 9.4 years.50 The prevalence of seropositivity of the oncogenic HPV-16 was nearly twice as high in SCCHN patients than in controls (12% vs 7%). In a nested case-control study involving 130 consecutive patients with oropharynx cancer, the risk of oropharynx cancer was linked to sexual behavior, lifetime exposure to HPV-16 (determined by seropositivity for HPV-16 LI capsid protein), and the presence of HPV infection (any HPV or HPV-16).44 Notably, the relationship between HPV-16 and cancer risk was independent of tobacco and alcohol use. After adjusting for age, sex, and alcohol use, HPV-16 seropositivity was found to be an independent predictor of oropharynx cancer and accounted for 55% of all cases.

Beyond its role as a risk factor for developing SCCHN, HPV status appears to define a distinct subtype of SCCHN tumors, and ultimately may aid in shaping treatment choices. Retrospective studies have consistently shown that HPV-positive tumors have better response and survival after treatment, possibly because of the propensity of these tumors to maintain an apoptotic response to radiation and chemotherapy.51 Alternatively, these patients tend to be younger and healthier compared with other head and neck cancer patients, with fewer smoking-related comorbidities; however, it is important to note that in multivariate analyses of clinical trials, the prognostic effect of HPV is not driven by demographics alone.

The impact of HPV status on response and survival was evaluated by the Eastern Cooperative Oncology Group (ECOG) in a prospective multicenter phase 2 study of chemoradiation as organ-preserving therapy in 96 patients with resectable stage III/IV laryngeal or oropharynx cancer. By using in situ hybridization techniques, HPV DNA was detected in 40% of tumor samples; 63% of oropharyngeal tumors were HPV positive, whereas none of the laryngeal tumors evidenced HPV DNA. Most positive cases involved HPV-16 (95%). Patients with HPV-positive tumors had a higher response rate after induction chemotherapy (82% vs 55%; P = .01) and after chemoradiation (84% vs 57%; P = .007). OS was significantly longer in HPV-positive patients than in HPV-negative patients (alive at 2 years: 95% vs 62%). The lower death and progression risk for HPV-positive patients was still present after adjusting for age, disease stage, and performance status.45

Recent retrospective studies of large phase 3 trials have solidified the clinical importance of HPV as a prognostic marker for SCCHN (Table 2). The phase 3 TROG (Trans Tasman Oncology Group) 02.02 (HeadSTART) trial randomized patients with stage III/IV SCCHN to receive cisplatin/radiotherapy ± tirapazamine; 465 patients had oropharynx cancer. Of the evaluable samples, 54 of 195 (28%) were HPV positive, and 107 of 186 (58%) were p16 positive. At the 2-year mark, HPV-positive tumors were associated with better OS versus HPV-negative tumors (94% vs 77%; HR, 0.27; P = .007) as well as improved failure-free survival (FFS; 85% vs 75%; HR, 0.43; P = .035). In addition, patients with p16-positive tumors had improved 2-year OS (92% vs 74%; HR, 0.35; P = .004) and improved 2-year FFS (87% vs 72%; HR, 0.38; P = .003). Patients with both HPV and p16-positive tumors had increased OS and FFS rates compared with double-negative patients (OS: 95% vs 71%; P = .003; FFS: 89% vs 69%; P = .002).52 Similarly, analysis from the RTOG 0129 study, which randomized patients to receive standard fractionation 70 grays (Gy) + cisplatin or accelerated fractionation with concomitant boost 72 Gy + cisplatin, correlated HPV status with patient prognosis. Of the evaluable samples, 206 of 323 (64%) were HPV positive, of which 96% were HPV-16 positive. OS, PFS, and locoregional failure were all significantly superior among HPV-positive patients at 2 years (OS: 87.9% vs 65.8%; P < .001; PFS: 71.8% vs 50.4%; P < .001; locoregional failure: 13.6% vs 24.8%; P = .004). Interestingly, heavy smoking status (≥20 pack-years) seemed to negate partially the benefit conferred by an HPV-positive tumor status; compared with HPV-positive/<20 pack-year patients, HPV-positive/≥20 pack-year patients had an OS HR of 1.91 (95% confidence interval [CI], 1.20-3.05); for HPV-negative/<20 pack-year subjects OS HR was 2.25 (95% CI, 1.44-3.50), and for HPV-negative/≥20 pack-year subjects OS HR was 4.30 (95% CI, 2.40-7.71).41

Table 2. Summary of Data on the Prognostic Effect of HPV on Different Phase 3 Trials in Locally Advanced Squamous Cell Carcinoma of the Head and Neck42, 52-54
StudyTreatment SettingHPV or p16 Positive, %Outcome Effect, HPV Positive vs HPV Negative
  1. Abbreviations: HPV, human papillomavirus; HR, hazard ratio; OS, overall survival; RT, radiotherapy; RTOG, Radiation Therapy Oncology Group; TROG = Trans Tasman Oncology Group.

RTOG 0129Cisplatin plus standard fractionation RT vs accelerated fractionation RT64%2-year OS <87.9% vs 65.8%; tobacco use interfered with the prognostic effect of HPV
TROG 02.02Cisplatin/RT ± tirapazaminep16 positive, 58%; HPV positive, 28%OS HR = 0.27; P = .007
RTOG 9003Standard fractionation RT vs hyperfractionation vs accelerated fractionation with split vs accelerated fractionation with concomitant boost RT39.5%5-year OS, 49% vs 19.6%
TAX, Taxotere 324Induction therapy cisplatin/5-fluorouracil ± docetaxel50%∼82-month OS, 79% vs 31%

More recently, Gillison and colleagues53 have performed an identical analysis of an older study, RTOG 9003, a randomized phase 3 trial comparing various radiation delivery schedules in locally advanced SCCHN. A total of 1068 patients were randomized to standard fractionation, hyperfractionation, accelerated fractionation with split, or accelerated fractionation with concomitant-boost radiotherapy, all without chemotherapy. The final analysis demonstrated no differences in PFS, locoregional control, and OS among the trial arms. Consistent with the observations from chemoradiotherapy trials, the retrospective analysis of 646 evaluable samples showed that HPV-positive patients (identified by the surrogate marker p16) had a 5-year survival rate of 49% compared with 19.6% for those who were HPV negative (P < .0001).

Finally, the observations from the subset of evaluable patients (111 of the 264 oropharynx cancer patients enrolled) from the phase 3 trial TAX 324 that compared 2 induction chemotherapy regimens (cisplatin and 5-fluorouracil with or without docetaxel total N = 501) before definitive chemoradiotherapy also were consistent, showing a strong prognostic correlation with HPV status. The rate of HPV-positive tumors was 50%. At a median follow-up of 83 months, 79% of HPV-positive patients were alive, versus 31% of those with HPV-negative tumors (P < .0001).54

Together, these studies have definitively established HPV as a meaningful prognostic marker for oropharyngeal SCCHN, and are sufficient to warrant accounting for HPV status in the design and analysis of future clinical trials using combined or sequential chemotherapy and radiation. We have no such prognostic data in the setting of surgery only.

Implications in SCCHN

The prevailing consensus at this time is that patients with HPV-related disease should probably be considered a distinct entity compared with patients with HPV-unrelated disease. Whether to stratify future trials on the basis of HPV status or actually institute separate trials for HPV-positive and HPV-negative patients remains unclear, although consensus is clearly moving toward separate trials, particularly because the clinical goals of therapy are different. In patients who are HPV positive, with less aggressive cancer and reasonably good survival, the goal is to deintensify therapy and thereby reduce toxicity. In HPV-negative patients, whose prognosis is decidedly worse, alternative approaches focusing on new targeted approaches and tolerable therapeutic intensification are desired.

Possible strategies for deintensification in HPV-positive disease include the administration of radiation or platinum therapy at lower doses with or without taxanes on a weekly basis, or the replacement of platinum therapy with cetuximab. For instance, ECOG is completing accrual to a phase 2 study of induction therapy followed by concurrent cetuximab and either low-dose radiation or standard radiation (intensity-modulated radiotherapy [IMRT]) in HPV-positive, resectable oropharynx cancer (see www.clinicaltrials.gov NCT01084083). As of September 2011, 79 of 86 patients targeted for accrual have been enrolled. The phase III intergroup trial RTOG 1016 (see www.clinicaltrials.gov NCT01302834), dedicated to patients with HPV-positive oropharyngeal tumors, will compare combination radiation and cetuximab to combination radiation and cisplatin. Patients are eligible if their tumors are HPV positive as determined by p16 status, and the radiation modality used is IMRT. The trial includes an extensive panel of quality of life-related measures as secondary endpoints to assess any potential difference or advantage in functional outcomes with cetuximab versus cisplatin, because the goal of this study is to pursue better functional outcomes without jeopardizing efficacy.

As the prevalence of HPV infection and HPV-related SCCHN increases,40, 44, 55 questions regarding screening and prevention are raised. Screening serum samples for HPV-16 is reliable and requires only basic laboratory techniques.56 But the value of screening high-risk patients for HPV infection has not been demonstrated, and the best approach to treating HPV-related premalignant lesions detected by screening is unknown. Moreover, whether identifying patients at high risk of contracting HPV may improve early detection of SCCHN by raising patient and physician awareness and increasing monitoring is unclear; the rising incidence of HPV infection raises other issues, including the potential blanket institution of vaccination strategies. To address these issues, a new multipronged approach to research is likely needed, involving continued retrospective analyses and epidemiological studies, prospective data collection and validation, and the initiation of the first trials to stratify SCCHN patients according to HPV status, as described above. Cooperative trials that are collecting samples for HPV assessment, such as the RTOG 0522 trial that compared chemoradiation with or without cetuximab, showing no differences overall with the addition of cetuximab,57 will help to determine the impact of HPV status on outcomes following current standards of care (see www.clinicaltrials.gov NCT00265941).


The ras family of genes, including H-RAS, K-RAS, and N-RAS, encode a protein that resides on the cytoplasmic side of the plasma membrane. The protein transmits mitogenic signals in response to a variety of physiological stimuli.58 Currently, there is considerable interest in the role of activated mutant forms of K-RAS in carcinogenesis. Most of the attention has focused on colorectal cancer, where the prevalence of K-RAS mutations is high (about 40%).59-61 In colorectal cancer, certain K-RAS mutations are predictive of lack of benefit from epidermal growth factor receptor (EGFR)-targeting agents, and testing for K-RAS mutation status is now recommended before treatment with EGFR therapy in this setting.62, 63 Assays for detecting K-RAS mutations include real-time PCR and direct sequencing analysis62; commercial kits are available for assaying K-RAS mutation status in colorectal cancer.

Whether K-RAS mutation status predicts response to EGFR-targeted therapy in other types of cancer is unknown. In SCCHN, the prevalence of K-RAS mutations is low. One study reported a prevalence of 8%,64 but most reports are generally <5%, or as low as 2%.65, 66 Although some evidence suggests that K-RAS mutations may indicate enhanced proliferation and an aggressive disease course in SCCHN,65, 66 their extremely low prevalence will likely preclude a major role in helping to define treatment. At this time, there is no clinical benefit to testing for K-RAS mutation status before administering EGFR-targeted therapy to patients with SCCHN. Further evaluation of K-RAS, including the implications of overexpression of wild-type K-RAS,67 is needed to better define the potential prognostic and predictive role, if any, of this factor in SCCHN.


The EGFR is a transmembrane protein that generates an intracellular signaling cascade that results in cell proliferation in response to extracellular binding with its natural ligands, such as epidermal growth factor and transforming growth factor (TGF)-α. Cells that acquire the ability to overproduce these ligands or increase the number of EGFRs on their surface can create an autocrine growth pathway, resulting in uncontrolled growth.68, 69

Virtually all SCCHN cases express EGFR according to currently accepted assays,69, 70 usually at high levels (2+, 3+). Indeed, mRNA expression of EGFR and its ligand TGF-α was increased in 24 tumor samples compared with normal mucosa samples taken from nonaffected individuals.69 There was no increase in EGFR gene copy number. Notably, levels of EGFR expression were also upregulated in normal mucosa from SCCHN patients, suggesting that increased EGFR expression is an early event in the development of SCCHN, and may help to explain the high incidence of synchronous and metachronous disease observed in this type of cancer.

Overexpression of EGFR is a negative prognostic factor associated with poor local control and survival in patients with SCCHN.68, 70, 71 In a randomized trial evaluating cisplatin with or without cetuximab in patients with advanced SCCHN, the highest levels of EGFR expression by IHC were associated with diminished responsiveness to cetuximab.72 It was also noted that cetuximab-related skin rash tended to be less frequent in those with the highest levels of EGFR expression. These findings suggest that, in patients with very high levels of EGFR, standard doses of cetuximab may be insufficient to saturate the target receptor, resulting in less efficacy and less toxicity. This may set the stage for dose-escalation studies of cetuximab using either direct measurement of EGFR expression or the clinical surrogate marker of skin rash to gauge biological target effects.72

The very high frequency of EGFR expression in SCCHN greatly limits the usefulness of classifying tumors as simply positive or negative as a marker for selecting treatment. The ability to quantify the EGFR expression levels more precisely, however, may enhance the usefulness of this biomarker. The most commonly used method for assessing EGFR expression levels is IHC of paraffin-embedded tumor samples, although results can vary depending on differences in technique, the type of antibody or fixative used, and the storage time of the sample.73 Fluorescent in situ hybridization can be used to determine the EGFR gene copy number, but increased EGFR gene copy number appears to be neither the main cause of increased EGFR expression nor predictive of outcomes in patients with SCCHN.69, 74 A new technique for assessing protein expression known as the automated quantitative analysis of protein expression may provide a more accurate measurement of total and phosphorylated EGFR expression in tumor samples.75 Use of this assay, however, is in its fledgling stages and has yet to be thoroughly evaluated for its predictive ability.

EGFR variants

A mutant form of EGFR known as EGFRvIII has been detected in up to 40% of SCCHN cases.76 This truncated, constitutively active receptor is ligand independent and does not bind with antibodies that target the extracellular domain of wild-type EGFR. In vitro, cells that express EGFRvIII have been shown to be less sensitive to the growth-inhibiting effects of cetuximab. Notably, EGFRvIII is typically only seen in cells that overexpress wild-type EGFR, which has led to the hypothesis that EGFRvIII mutations are a late stage event caused by the rapid proliferation induced by wild-type EGFR overexpression.76

Other alterations in EGFR include activating mutations, which have been detected in 10% to 30% of patients with NSCLC and predict increased sensitivity to small molecules that target the intracellular portion of EGFR.77, 78 However, these mutations are present in just 7.3% of Asian patients with SCCHN who have been evaluated for the mutation and in only 1% of evaluated Caucasian patients with SCCHN, indicating that it is a relatively rare mutation for this tumor type.77, 79 Although the practical applications of EGFR mutation status are currently limited, further assessment of EGFRvIII and other mutations in SCCHN is warranted.

Serum Profiling

Serum-based diagnostics offer a different type of approach, less invasive and not dependent on the collection of tumor specimens. Markers in serum provide information on host factors that may determine response to therapy.

Serum samples can be analyzed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. This analysis generates a proteomic peak profile, because most commonly the species detected in the sample are proteins or protein fragments. In NSCLC, a profile of this kind has been characterized that may be prognostic and/or predictive of benefit with EGFR tyrosine kinase inhibitor (TKI) therapy.80 A retrospective study in 108 patients with recurrent/metastatic SCCHN treated with cetuximab or EGFR TKIs revealed that patients with the same proteomic profile (66% of the sample) derived greater benefit from anti-EGFR therapy than those negative for the profile (HR for OS, 0.20 to 0.41, depending on the agent). There was also a small numeric difference between positive and negative patients treated with docetaxel chemotherapy (n = 34), which complicates the interpretation of these data, suggesting that this marker possibly could have a prognostic component, in addition to a predictive effect.81

A different strategy is the analysis of a defined panel of biomarkers in serum, rather than a proteomic approach. Byers et al82 and Ferris et al83 have analyzed defined panels of cytokines and angiogenic factors by enzyme-linked immunosorbent assay on serum from patients with locally advanced SCCHN treated with induction 5-fluorouracil or docetaxel/cisplatin plus cetuximab. In both studies, marker signatures including vascular endothelial growth factor and interleukin 6 were identified that appeared to correlate with progressive disease or recurrence. Follow-up studies could confirm if this effect allows for prognostic stratification, or is predictive of specific treatment benefits.


There are a growing number of molecular markers that may potentially be used as either prognostic or predictive tools in the treatment of SCCHN (Table 3). In particular, 2 factors—ERCC1 and HPV—have been studied extensively and have produced consistent results in both SCCHN and other types of cancer. Pending further clinical study, these factors are likely to soon alter how we select treatment for patients with SCCHN.

Table 3. Summary of Key Implications About Candidate Biomarkers in SCCHN
  1. Abbreviations: FISH, fluorescent in situ hybridization; HPV, human papillomavirus; RT, radiotherapy; SCCHN, squamous cell carcinoma of the head and neck.

ERCC1 expressionExpression may be relevant for response to platinum therapy, needs further validation.
ERCC1 polymorphismsMay be relevant to outcomes after radiotherapy treatment, needs further validation.
RRM1Expression may be relevant to response to gemcitabine. It may also correlate with ERCC1 expression, and other DNA repair-related markers, unclear relevance.
β-TubulinExpression of certain isotypes may influence response to taxanes, needs further validation.
HPVStrong prognostic factor, warrants dedicated trial designs. No specific treatments for the HPV-positive population yet.
K-RAS mutationsLow prevalence, no predictive value documented.
EGFRExpression is universal in SCCHN; overexpression is a negative prognostic factor after RT. Gene copy amplification (by FISH) is not predictive of outcomes after cetuximab treatment.
EGFRvIIIMay affect sensitivity to cetuximab, not yet validated in the clinic.
EGFR kinase domain mutationsLow prevalence, unclear relevance.

With regard to other potential markers, however, it is becoming increasingly apparent that what is relevant in 1 type of cancer may not necessarily apply to others. Despite their importance in colorectal cancer, K-RAS mutations, for example, do not appear to play any role in SCCHN, and routine testing for K-RAS mutations in SCCHN patients is not warranted at this time. EGFR unquestionably plays an important role in the development of SCCHN, but its usefulness as a biomarker in this setting may depend on further refinement of testing methods and better definition of candidate abnormalities. The cooperative groups will continue to study these and other markers to better select the most appropriate treatment for patients with SCCHN.


C.J.L. is fully responsible for the content of this article, but wishes to acknowledge the assistance of the Clinical Insights, Inc. editorial team, with funding from Bristol-Myers Squibb, in researching references, preparing tables, editing the draft, and formatting it for submission.


Speaker's bureaus: BMS, ImClone, Lilly, Genentech/OSI (all curtailed as of December 2010); advisory boards: BMS, ImClone, Lilly, Genentech/OSI, Sanofi-Aventis, Novartis, Biodesix, Clarient, Astra-Zeneca, Celgene, Abraxis, Allos, Pfizer, Boehringer-Ingelheim, Bayer-Onyx; research funding: BMS, ImClone, Lilly, Genentech/OSI, Celgene, Abraxis, Pfizer, Boehringer-Ingelheim, GSK.