Representative surface-enhanced laser desorption/ionization time of flight mass spectrometry (SELDI) gelview and spectra from serum samples of 5 healthy (normal) controls, 5 patients with head and neck squamous cell carcinoma (HNSCC), and 5 "healthy" smokers. The "box" identifies a peak with an average mass of 10 068 Da that corresponds to the tumor marker metallopanstimulin-1 at higher intensity in HNSCC serum compared with normal and healthy smoker sera. m/z indicates mass to charge.
Surface-enhanced laser desorption/ionization time of flight mass spectrometry (SELDI) expression level of a 10 068-Da protein in 4 pooled serum samples. The relative intensity of the SELDI peaks corresponds to the relative intensity of metallopanstimulin-1 as determined by radioimmunoassay: XHCPS, extra high concentrations of patient sera (CPS); HCPS, high; MCPS, medium; and LCPS, low. m/z indicates mass to charge.
Spectra showing the presence of a 10 068-Da peak in pooled sera containing a high amount of metallopanstimulin-1 (MPS-1) bound to an immunochip treated with rabbit polyclonal anti-MPS-1 IgG. When an unrelated rabbit IgG antibody is used, no peak at this mass is observed. The mass of this peak corresponds to the mass of the peak observed when the serum is directly bound to the IMAC3 ProteinChip pretreated with copper sulfate. m/z indicates mass to charge; XHCPS, extra high concentrations of patient sera.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Wadsworth JT, Somers KD, Stack, Jr BC, et al. Identification of Patients With Head and Neck Cancer Using Serum Protein Profiles. Arch Otolaryngol Head Neck Surg. 2004;130(1):98–104. doi:10.1001/archotol.130.1.98
New and more consistent biomarkers of head and neck squamous cell carcinoma (HNSCC) are needed to improve early detection of disease and to monitor successful patient management.
To determine if a new proteomic technology can correctly identify protein expression profiles for cancer in patient serum samples as well as detect the presence of a known tumor marker.
Direct proteomic analysis and comparison.
The surface-enhanced laser desorption/ionization time of flight mass spectrometry (SELDI-TOF) ProteinChip system was used to screen for differentially expressed proteins in serum samples from 99 patients with HNSCC, 25 "healthy" smokers, and 102 healthy (normal) controls. Protein peak clustering and classification analyses of the SELDI spectral data were performed.
Several proteins, with masses ranging from 2778 to 20 800 Da, were differentially expressed between patients with HNSCC and the normal controls. The serum protein expression profiles were used to develop a classification tree algorithm, which achieved a sensitivity of 83.3% and a specificity of 90% in discriminating HNSCC from normal and healthy smoker controls. The positive and negative predictive values were 80% and 92%, respectively. A peak with an average mass of 10 068 Da was detected in sera from HNSCC patients and identified as the known biomarker metallopanstimulin-1 (MPS-1), based on mass. Peak relative intensity of the 10 068-Da protein correlated consistently with MPS-1 levels detected by radioimmunoassay in serum samples of HNSCC patients and controls. The 10 068-Da peak was provisionally identified as MPS-1 by SELDI immunoassay.
We propose that this technique may allow for the development of a reliable screening test for the early detection and diagnosis of HNSCC, as well as the potential identification of tumor biomarkers.
HEAD AND NECK SQUAMOUS cell cancer (HNSCC) can be a devastating disease, comprising over 5% of all cancers in the United States and an even larger proportion of cancers worldwide.1 Tobacco and alcohol consumption are well-established risk factors for HNSCC. Despite many advances in the treatment of HNSCC over the past 30 years, little progress has been made in improving survival rates. These tumors are often discovered in advanced stages with treatment frequently leaving patients disfigured and with debilitating adverse effects of radiation and chemotherapy, compromised speech and swallowing, and significantly diminished quality of life.
The incidence of HNSCC in the United States has changed very little, despite increased awareness and education about the potential effects of tobacco and alcohol use. Even though prevention and early diagnosis are well established as mainstays of successful HNSCC treatment, no accepted screening test exists for this cancer. In fact, screening for HNSCC is not mentioned in the most recent screening guidelines of the American Cancer Society,2 presumably due to the lack of sufficient screening tools available to physicians. Aside from a complete head and neck physical examination and imaging studies in patients with suspicious clinical findings or symptoms, there are no accepted methods to screen for these cancers.
The ongoing search for biomarkers predictive of HNSCC has focused primarily on the detection of genetic abnormalities leading to the development of HNSCC.3,4 Despite the identification and characterization of multiple molecular aberrations in HNSCC, available technology limits their routine clinical use, and none has been determined to reliably enhance early detection of HNSCC. Recently, attention has focused on deciphering the HNSCC proteome in search of diagnostic biomarkers. The human proteome is defined as the full set of proteins encoded by the genome. Proteomics, or protein pattern analysis, is the characterization and quantitation of proteins in tissues and body fluids. Proteomic methods can be used to compare protein expression patterns in healthy patients vs those with cancer. Proteomic research has traditionally involved 2-dimensional gel electrophoresis (2D-PAGE) to detect differences in protein expression in tissue and body fluid specimens between the healthy (control) group and the disease group.5,6 In 2D-PAGE, proteins are separated on the basis of size and charge. Although 2D-PAGE has been the primary technique in conventional proteomic analyses, it has limitations in detection, particularly for proteins of low abundance and less than 10 000 Da, is labor intensive, has low throughput, and is not easily applied in the clinical setting.
Recently, there have been technological advances in proteomics such as ProteinChip surface-enhanced laser desorption/ionization time of flight mass spectrometry (SELDI-TOF-MS).7,8 In SELDI-TOF-MS analysis, patient samples, such as serum, are applied to a protein-binding chip that facilitates protein capture. The chip is irradiated with a laser, which causes the adherent proteins to "fly off" as charged ions. The ions travel through a vacuum tube and the mass-to-charge (m/z) ratios are calculated based on their "time of flight" (TOF) through the ion chamber (reviewed in Wright9 and Wulfkuhle et al10). This technology has shown great potential for the early detection of a variety of human cancers.11-16 The objective of this study was to determine if protein profiling using SELDI-TOF-MS could accurately distinguish patients with HNSCC from both healthy controls (normal controls) and "healthy" smokers (smoker controls), as well as detect a known tumor marker, metallopanstimulin-1 (MPS-1),17 potentially providing a novel approach for the detection of HNSCC.
Human serum samples from patients with head and neck cancer and normal controls were obtained through institutional review board–approved protocols at Saint Louis University School of Medicine, the Pennsylvania State University College of Medicine, and the Eastern Virginia Medical School from 1997 to the present. All patients gave informed consent under the approved protocol. All serum samples were aliquoted and frozen at −80°C until thawed specifically for SELDI analysis.
Serum samples were processed for SELDI analysis as previously described using the IMAC3 ProteinChip (Ciphergen Biosystems, Fremont, Calif) pretreated with copper sulfate.8 Briefly, 20 µL of serum is pretreated with 8M urea and 1% 3-(3-cholamidopropyl dimethylammonio)-1-propanesulfonate (CHAPS) and vortexed for 10 minutes at 4°C. A further dilution is made in 1M urea and 0.125% CHAPS, and phosphate-buffered saline (PBS). Diluted serum is then added to the ProteinChips with the aid of a bioprocessor. Each serum sample was assayed in duplicate, with duplicate samples randomly placed on different ProteinChips. ProteinChips are then incubated at room temperature followed by washes of PBS and water. Arrays were allowed to air dry and a saturated solution of sinapinic acid (Ciphergen Biosystems) in 50% (vol/vol) acetonitrile, and 0.5% (vol/vol) trifluoroacetic acid was added to each spot. The ProteinChip arrays were analyzed using the SELDI ProteinChip System (PBS-II; Ciphergen Biosystems). Spectra were collected by the accumulation of 192 shots at laser intensity 220 in a positive mode. The protein masses were calibrated externally using purified peptide standards (Ciphergen Biosystems).
Before analysis, the data were divided into 2 sets as follows: a training set that consisted of 75 samples from each group (normal control and HNSCC), and a test set of 24 HNSCC patients, 27 normal controls, and 25 smoker controls. Spectra were analyzed with the Ciphergen ProteinChip software (version 3.0) and normalized using total ion current. Peak labeling and clustering were performed using Ciphergen's Biomarker Wizard tool, exported into a spreadsheet, and the intensity values for each peak were averaged for duplicate samples. These spectral data were then analyzed by the BioMarker Patterns software (Ciphergen Biosystems) to develop a classification tree.
Details regarding classification and regression tree analysis (CART) and the bioinformatics algorithm incorporated within the BioMarker Patterns software program have been described elsewhere.18,19 Briefly, classification trees split the data into 2 nodes, using one rule at a time in the form of a question. The splitting decisions in this case were based on the normalized intensity levels of peaks from the SELDI protein expression profile. Each peak or cluster identified from the SELDI profile is therefore a variable in the classification process. The process of splitting is continued until terminal nodes are reached and further splitting has no gain in data classification. Classification trees were constructed using the training set, and following V-fold cross validation, the accuracy of each classification tree was challenged with the test set. Multiple classification trees were generated using this process, and the best performing tree was chosen for further testing.
Specificity was calculated as the ratio of the number of negative samples correctly classified to the total number of true-negative samples. Sensitivity was calculated as the ratio of the number of correctly classified diseased samples to the total number of diseased samples. Negative and positive predictive values were then calculated. Comparison of relative peak intensity levels between groups was calculated using the t test.
A total of 125 volunteers with newly diagnosed HNSCC were prospectively enrolled from a university head and neck oncology clinic. Serum MPS-1 levels in these patients were compared with 2 control groups. The first control group comprised 25 healthy, nonsmoking volunteers (normal controls). The second control group included 64 actively smoking volunteers who were screened for HNSCC and were found to be free of disease (smoker controls) during the 1999 Yul Brynner Head and Neck Screening Day in St Louis, Mo. A total of 821 serum samples (709 from cases [HNSCC], 48 from normal controls, and 64 from smoker controls) collected over a 24-month period were analyzed by the MPS-1 radioimmunoassay (RIA).
Technical details for the preparation of reagents for MPS antigen determinations, RIA procedure, and patient sample preparation are published elsewhere.20,21 Each serum sample was run in duplicate by technicians who were blinded to specimen identity. Quality control was done following the National Committee for Clinical Laboratory Standards recommendations.21 To create MPS-1 concentration standards, the serum samples were pooled based on their RIA results individually into differing concentrations of patient sera (CPS): extra high, XHCPS >100 ng/dL; high, HCPS = 50-100 ng/dL; medium, MCPS = 20-50 ng/dL; and low, LCPS <20 ng/dL.
Protein G (Sigma-Aldrich Co, St Louis, Mo) at 0.5 mg/mL in 50mM sodium bicarbonate (pH 8.0) was added to the spots of a preactivated ProteinChip array (PS1) (Ciphergen Biosystems, Inc, Palo Alto, Calif). The PS1 chip array consisted of a carbonyldiimidazole surface that docks proteins by covalently reacting with their amine groups. The chips were incubated 2 hours at room temperature with shaking. After blocking residual active sites with 20 µL of 1M ethanolamine, the chip was washed in a 15-mL conical tube twice with 1% Triton X100 in PBS and twice with PBS (pH 7.4).
Preparation of anti–MPS-1 rabbit polyclonal antibodies has been described elsewhere.21 Anti–MPS-1 IgG (1 mg/mL) or normal rabbit IgG was added to individual spots (2 µL) and incubated in a humid chamber 2 hours with shaking. Unbound IgG was removed by washing the chips with 0.1% Triton X100 in PBS 2 times followed by PBS 3 times in bulk.
For the immunoassay, a bioprocessor was used to increase the volume that could be added to each spot on the immunochip. The chip was conditioned by washing with 200 µL of 0.5% Triton X100 in PBS. The pooled MPS-1 serum samples were diluted 1:1 with 0.5% Triton X100 in PBS (200 µL) added to each chip array, and the chips incubated at 4°C overnight with mixing. The spots were washed with 0.5% Triton X100 in PBS, followed by PBS and then rinsed in high-pressure liquid chromotography water twice. Sinapinic acid was added, and mass analysis performed with a combination of 192 laser shots throughout each spot.
Peak detection using Ciphergen's ProteinChip software identified an average of 90 peaks/spectrum. Of these, 80 common peaks or clusters were generated from the training set, with masses ranging from 2000 to 21 000 Da. Because the majority of the peaks detected were in this mass range, it was considered the most useful for protein profiling. Each cluster was determined with a mass window of 0.2% and represents one protein peak. As shown in Table 1, 33 of these peaks were found to have significant differential expression levels between the HNSCC and control sera. One of these peaks, at 10 068 Da, corresponded to MPS-1 based on mass.
Classification trees were created using the training set with V-fold cross validation using all 80 peaks. This type of cross validation uses random numbers to split up the data in the training set for testing each tree. Based on the CART analysis, no single peak was identified as having the ability to separate HNSCC sera from normal and smoker control sera alone. However, the underexpression of one protein peak at 5064 Da was used in all the classification trees as the first primary splitter.22
Of the many classification trees generated, the most accurate tree, based on internal cross validation results, was used for further analysis. The most accurate tree correctly classified 90.7% of the HNSCC sera in the training set. This classification tree algorithm was then challenged with a test set blinded to the algorithm consisting of serum samples from 27 normal controls, 24 patients with HNSCC (distinct from the training set), and 25 smoker controls. All normal control (100%), 83.3% of HNSCC, and 80% of smoker control (identified as noncancer) samples were correctly identified by the algorithm (Table 2). This yielded an overall sensitivity of 83.3% and overall specificity of 90%. Negative predictive value was therefore 92% and the positive predictive value of the test was 80%.
A peak with an average mass of 10 068 Da was consistently identified in many of the serum samples. Figure 1 represents a combined gelview and spectra of 5 representative samples from each group (normal controls, HNSCC cases, and smoker controls). Common detection of the 10 068-Da peak in the HNSCC group compared with its relative absence in the normal controls and variable presence in the smoker control group is clearly shown.
Compared with the MPS-1 pooled serum samples, the 10 068-Da peak in experimental sera was, by mass, consistent with the same peak in the pooled sera. Furthermore, the MPS-1 pooled sera peak at 10 068 Da was graded in relative intensity proportional to the relative level of MPS-1 present detected by RIA (Figure 2).
Further provisional identification of this known tumor marker was performed using a SELDI immunoassay. As seen in Figure 3, the MPS-1 peak seen in sera using the IMAC3 ProteinChip was detected at the same mass using the anti–MPS-1 IgG immunochip. No 10 068-Da peak was detected when normal rabbit IgG replaced the anti–MPS-1 IgG. These data provide further indirect evidence to indicate that the peak seen at 10 068 Da is MPS-1.
Detection of head and neck cancer at early disease stages is paramount to successful clinical therapy. Yet, early-stage head and neck cancer lacks specific symptoms or biomarkers that accurately and reliably distinguish patients with HNSCC from healthy controls. Many studies have described limited success in identifying HNSCC-associated protein, DNA, and RNA biomarkers that potentially could aid the early diagnosis and prognosis of HNSCC. Reverse transcription–polymerase chain reaction was used to detect metastasis-associated cytokeratin 19 positive tumor cells in sera from a small number of patients with nasopharyngeal carcinoma.23 Analysis by enzyme-linked immunosorbent assay of serum (basic fibroblast growth factor, vascular endothelial growth factor, and matrix metalloproteinase-2) showed that only increased basic fibroblast growth factor levels correlated with earlier locoregional control in patients with HNSCC following primary chemoradiation therapy.24 Other studies evaluated several conventional serologic markers in HNSCC patients and found none to be of statistical significance.25 Antibodies to p53 tumor suppressor protein were detected in the serum samples of 25% of 271 patients with oral cancer26 and at a low percentage in saliva from patients with HNSCC.27,28 Nucleic acid–based microsatellite instability and hypermethylation of promoter regions have been used as markers to detect tumor-specific alterations in serum and saliva of patients with HNSCC.29-33 These approaches are often subjective, technically challenging, and require a panel of microsatellite markers or selected genes. Generally, nucleic acid–based methods for detection of cancer have been assessed with a limited number of samples and will require further trials to confirm these early results.
Despite the identification and extensive study of several potential tumor markers, none has been found to have clinical utility as a diagnostic marker or screening tool for HNSCC. Given the complexity of the genetic and molecular alterations that occur in HNSCC cells, the proteomic fingerprint in both tissues and body fluids may hold more vital information in screening, diagnosis, and prognosis than the individual molecular changes themselves. SELDI analysis has very high throughput (minutes or hours for large sample numbers vs days for 2D-PAGE), and specimen volumes needed are miniscule relative to traditional 2D-PAGE techniques. Protein expression profiling has been used previously to detect a protein of 8670 Da using a hydrophobic surface (H4) in tumor extracts of 5 of 6 HNSCC cases but not in matched normal tissue lysates.34 In a study of 2 matched HNSCC cell lines derived from either the primary tumor or lymph node metastasis, the SELDI ProteinChip H4 was used to identify the up-regulation of 2 membrane-associated proteins (annexin I and annexin II) and glycolytic protein enolase-α as well as the down-regulation of calumenin precursor in the metastatic cell line.35 To date, SELDI ProteinChip technology has not been reported as a tool to interrogate serum from HNSCC patients compared with normal controls for protein fingerprints of HNSCC.
No standardized screening tool is available for patients with HNSCC. Patients are most often diagnosed in the late stages of disease because of the location of the tumors and because early symptoms often mimic and are treated as benign processes, such as viral upper respiratory tract infections. Continued efforts to identify protein profiles or patterns that differentiate cancer from noncancer could lead to earlier detection and development of diagnostic tests for HNSCC. Using SELDI-TOF-MS techniques, we achieved 90% specificity and 83.3% sensitivity for detection of HNSCC, as well as correctly categorized 80% of patients deemed to be at high risk, rapidly and reproducibly. Once validated with more samples, in a planned multi-institutional investigation, this approach may provide an innovative test of significant benefit for clinicians treating HNSCC.
We also note that this technology allows demonstration of the presence of known tumor markers. In this case, we looked specifically for MPS-1, given that this marker had been fully characterized in serum of the study groups. We were able to detect a peak of 10 068 Da consistent with the mass size of MPS-1. In an effort to identify the 10 068-Da peak, we used a SELDI immunoassay with rabbit polyclonal antibodies to MPS-1 to demonstrate a peak at 10 068 Da bound by the anti–MPS-1 antibody but not detected with normal rabbit IgG. To our knowledge, this is the first report of identification of a known tumor marker in HNSCC using SELDI technology. It is particularly interesting that SELDI allows for identification of a tumor marker that is in relatively low concentrations in serum. As the number of known tumor markers grows, use of this technology may allow for analyzing patient sera reproducibly and with high throughput for many biologic markers simultaneously. SELDI may therefore facilitate identification of certain patterns of biologic markers that could have relevance to the diagnosis, clinical behavior, and treatment of HNSCC.
Interestingly, the serum profiles of the smoker control group appeared to have many protein peaks in common with HNSCC patients. Indeed, the classification tree correctly identified nearly the same percentage of smoker controls as HNSCC patients. The differences between healthy smokers and HNSCC patients were expected to be less than those between normal controls and HNSCC patients, since progression from normal to cancer is multifocal and heterogeneous. Some "healthy" smokers may well be on the way to developing HNSCC without overt signs. The addition of a smoker control group in the analysis did have a minor effect on the overall sensitivity and specificity of the test compared with previous data.22 Further study and larger sample sizes are needed to further evaluate these differences.
Further studies are also under way to determine if differing protein patterns can be identified in individuals with different TNM stages and subsites of HNSCC. In addition, other studies in progress include (1) the evaluation of other chip surfaces to improve testing accuracy and to increase the chance of discovery of potential biomarkers, and (2) the analysis of serum SELDI profiles before, during, and after definitive treatment of HNSCC to determine whether this technique can be equally useful to monitor patients for persistent or recurrent disease. In the future, it may be possible to predict which treatment or combination of treatments may best serve patients with certain protein fingerprints.
In summary, SELDI protein fingerprinting represents a paradigm shift from traditional cancer diagnostic approaches. The identification of protein biomarkers may be facilitated by SELDI-TOF-MS as exemplified by the preliminary identification of a protein species at 10 068 Da as MPS-1. It appears possible to identify known tumor markers in a semiquantitative fashion. Furthermore, it is the protein pattern rather than the individual protein alteration that becomes the key for differentiating healthy individuals from those with HNSCC. The high sensitivity and specificity achieved in this study using SELDI-TOF-MS techniques coupled with a classification algorithm identified protein patterns in serum that distinguished normal controls from HNSCC patients and healthy smokers. This technique provides data that are easy to accumulate and that should be readily adaptable to clinical use. Further investigation is warranted to evaluate SELDI protein profiling as an assay for early detection, diagnosis, and prognosis, potentially leading to more accurate screening, diagnosis, and treatment of patients with HNSCC.
Corresponding author and reprints: J. Trad Wadsworth, MD, Department of Otolaryngology–Head & Neck Surgery, Eastern Virginia Medical School, 825 Fairfax Ave, Suite 510, Norfolk, VA 23507 (e-mail: email@example.com).
Submitted for publication April 24, 2003; final revision received August 1, 2003; accepted August 7, 2003.
This work was supported in part by grant CA85067 from the National Cancer Institutes Early Detection Research Network (Washington, DC), grant DAMD17-02-1-0054 from the Department of Defense Prostate Cancer Research Program (Washington, DC), and a grant from the Virginia Prostate Center (Norfolk).
This study was presented at the annual meeting of the American Head and Neck Society; May 4, 2003; Nashville, Tenn.
Create a personal account or sign in to: