The Lack of Utility of a Tumor Marker Panel in Head and Neck Carcinoma Squamous Cell Carcinoma Antigen, Carcinoembryonic Antigen, Lipid-Associated Sialic Acid, and CA-125 Matthew B. Straka; Robin L. Wagner; Jonas T. Johnson, MD; \s=b\

An ideal tumor marker should be sensitive in tumor\x=req-\

bearing patients while having adequate specificity so that controls do not demonstrate the marker. To date, a single circulating marker has not been identified for squamous cell carcinoma of the head and neck. This study evaluates a panel including squamous cell carcinoma radioimmunoassay, lipid-associated sialic acid, carcinoembryonic antigen, and CA-125. In this population of patients with cancer, serum samples from 101 patients and 88 controls were evaluated. The squamous cell carcinoma radioimmunoassay

the most sensitive marker identified (47.5%), while carcinoembryonic antigen level was elevated in 40.6%, lipid-associated sialic acid level in only 16.8%, and CA-125 level in 7.9%. False-positive results were found with all markers, including squamous cell carcinoma radioimmunoassay (18.2%), carcinoembryonic antigen (18.2%), lipid\x=req-\ associated sialic acid (10.2%), and CA-125 (15.9%). Various combinations of markers did not significantly improve either specificity or sensitivity. Available tumor markers are inadequate for diagnostic purposes in patients with squamous cell carcinoma of the head and neck. (Arch Otolaryngol Head Neck Surg. 1992;118:802-805)

was

studies have been conducted attempts Numerous marker for the presence of identify cell carcinoma (SCC) the head and neck in

a

to

serum

in re¬ squamous gion. An ideal marker would be both a specific and sensi¬ tive indicator of the presence and extent of a cancer. A marker showing high specificity would be present in ele¬ vated levels in a low percentage of patients without disease, and a marker that is sensitive would be elevated in a high percentage of patients with disease. An ideal marker would correlate with stage of disease and be sen¬ sitive enough to detect the earliest stages of cancer. Such a marker could be an advantage in the diagnostic evaluations of malignant neoplasia as well as for use in the follow-up of patients after therapy to identify early recur¬ rence. If able to be obtained by relatively noninvasive

Accepted for publication November 5, 1991. From the Department of Otolaryngology, University of Pittsburgh (Pa) School of Medicine, The Eye and Ear Institute of Pittsburgh. Reprint requests to Department of Otolaryngology, University of Pittsburgh School of Medicine, The Eye and Ear Institute of Pittsburgh, 203 Lothrop St, Suite 500, Pittsburgh, PA 15213 (Dr Johnson).

Kimberly Keane Kachman; David E. Eibling,

MD

means, such

as from blood or urine, and employed for low cost, this ideal marker could be used as a screening proce¬

dure for large populations, leading to earlier diagnosis and better cure rates. While various substances have been evaluated as possi¬ ble tumor markers for SCC of the head and neck, none have been demonstrated to meet the requirements of an ideal marker. In an attempt to improve specificity and sensitiv¬ ity by combining multiple markers, a panel of four mark¬ ers was studied. It was postulated that when examined and correlated, the combined data might yield a significant advantage in specificity and sensitivity over the use of in¬ dividual markers. Previous investigators have demon¬ strated an additive effect in utilizing multiple markers. Goodwin and coworkers1 demonstrated improved speci¬ ficity and sensitivity in a study using two markers, SCC radioimmunoassay (RIA) and lipid-associated sialic acid

(LASA).

This study employed a panel of four known markers: SCC antigen, carcinoembryonic antigen (CEA), LASA, and CA-125. Serum levels were examined simultaneously in 101 patients with cancer and 88 healthy controls. Each of these four markers has demonstrated independent clinical applicability in a variety of histologie types (Table 1). MATERIALS AND METHODS Carcinoma RIA Johnson et al2 and Eibling et al3 studied the usefulness of SCC antigen in head and neck cancers using a commercially available RIA.2 The SCC antigen levels were elevated (£2 ng/mL) in only 45% of patients with cancer, but 67% of the patients who remained disease free after surgery showed corresponding decreases in levels of the marker. In patients experiencing recurrence of dis¬ ease, 92% had corresponding elevations of SCC antigen level in

Squamous Cell

samples. Similar results were noted in other studies of SCC antigen.4 serum

Carcinoembryonic Antigen Silverman et aF studied CEA levels in the serum of 439 patients with SCC of the head and neck and a control group consisting of 154 healthy smokers and 122 nonsmokers. Carcinoembryonic an¬ tigen levels were found to exceed 5 ng/mL in only 5% of nonsmokers, but in 36% of tumor-bearing patients. Carcinoem¬ bryonic antigen levels exceeded 7 ng/mL in 5% of healthy smok¬ ers and 17% of tumor-bearing patients. Carcinoembryonic anti¬ gen levels did appear to correlate with stage, but this was only

Downloaded From: http://archotol.jamanetwork.com/ by a Georgetown University Medical Center User on 05/27/2015

clearly apparent

when patients with stage IV disease were included.6 The CEA levels appeared to be unable to predict local recurrence, metastatic spread, or appearance of new primary cancer.6

Table 1.—Marker Manufac-

Use

turer

Studied

Marker*

CA-125 Antigen Serum levels of CA-125 antigen have been demonstrated to be elevated in greater than 80% of nonmucinous ovarian cancers.7 Levels have also been shown to correlate with the course of dis¬ ease in about 80% of cases, in that an increase of serum CA-125 level precedes relapse by a few months. In another study by Kawahara et al,8 CA-125 antibody reacted with 43% of the 82 se¬ rum samples for ovarian cancer, 18% for breast cancer, 7% for colorectal cancers, and 19% for miscellaneous cancers, which in¬

SCC

antigen

Lipid-asociated sialic acid was tested as a serum marker by Voigtmann et al.' The total sialic acid content was determined for a group of healthy donors to have a mean value of 59.95 mg/dL,

while that of a group with cancer was determined to be 95.91 mg/dL. Since sialic acid is a constituent of glycoproteins and glycolipids, which alter their composition on malignant cell sur¬ faces, it is believed that sialic acid could enter the blood in signif¬ icant amounts in patients with cancer. The study showed that glycoproteins containing sialic acid coextracted with glycolipids constituted the LASA level.

Methods In

1981, a serum bank was initiated in the Department of Oto¬ at the Eye and Ear Institute of Pittsburgh (Pa).

laryngology

Patients with biopsy-proved SCC of the head and neck were ac¬ crued in accordance with the guidelines of the Institutional Review Board. The serum sample was drawn before treatment, centrifuged, pipetted into aliquots, immediately frozen, and stored at 70°C. Data on each patient, including site, stage, treat¬ ment, and outcome, were obtained and maintained on a comput¬ erized database. Serum samples from 101 patients with cancer and 88 controls were the topic of this study. Squamous Cell Carcinoma RIA.—Samples were assayed with the research SCC-RIA kit distributed by Abbott Laboratories, North Chicago, 111 (manufactured for Abbott Laboratories by Dainabot Co Ltd, Tokyo, Japan) to detect the presence of SCC an¬ tigen. The purified antigen used in the RIA kit is a subfraction of TA-4, the antigen employed by Kato and Torigoe.10 Carcinoembryonic Antigen.—Cancer group and control se¬ rum samples were thawed and assayed with the CEA-RIA kit distributed by Abbott Laboratories (Diagnostics Division) to de¬ tect the presence of CEA. The Abbott CEA-RIA is a solid-phase RIA in which beads coated with anti-CEA (mouse, monoclonal) are incubated with specimen dilution buffer and specimen sam¬ ple, standard, or controls. Lipid-Associated Sialic Acid.—The total sialic acid content of the serum samples was determined by a photometric test kit pro¬ duced by Boehringer Mannheim Biochemica (Mannheim, Ger¬ many). The glucosidically bound sialic acid was hydrolyzed by neuraminidase to release free N-acetylneuraminic acid. CA-125 Antigen.—The CA-125 RIA is a simultaneous sand¬ wich, solid-phase RIA manufactured by Centocor Ine (Malvern, Pa).7-8 Polystyrene beads coated with mouse monoclonal antibody to CA-125 are incubated with the specimen, standards, or control, and tracer.

Range,

Range,

%

%

Lung

40-55

90-95

Abbott

Colon,

60-83

85-89

60-72

90-92

ovaries, etc

Boehringer

LASA

Head and

neck,

lung,

cluded head and neck carcinomas.

Lipid-Associated Sialic Acid

Sensitivity Specificity

Abbott

radioimmu¬ noassay CEA

Sensitivity and Specificity

etc

Centocor

CA-125

34-58 91-99 Ovaries *SCC indicates squamous cell carcinoma; CEA, carcinoembryonic antigen; and LASA, lipid-associated sialic acid.

Table 2.—Marker Level Elevations for

Study Population*

Elevations SCC values >2.00 ng/mL

Cancer (n=101)

Control (n=88)

48 (47.5)

19 (22)

CEA values >4.00 ng/mL

41 (40.6) 16 (18) LASA values 17 (16.8) >125.00 mg/dL 9(10) CA-125 values 8 (7.9) 14 (16) £25.00 U/mL *Values are number (percent). SCC indicates squamous cell carci¬ noma (antigen detected by radioimmunoassay); CEA, carcinoembryonic antigen; and LASA, lipid-associated sialic acid.

—

RESULTS Serum SCC antigen levels greater than 2.00 ng/mL were considered to be elevated based on the previous SCC study by Johnson et al.2 The SCC antigen level was elevated in 48 (47.5%) of the 101 patients with cancer, which was similar to previous studies.1 Circulating SCC antigen was elevated in 19 (22%) of the 88 controls. The mean SCC level in the

Table 3.—Mean Marker Values Marker*

SCC, ng/mL CEA, ng/mL LASA, mg/dL CA-125, U/mL

Cancer

Control

3.025

1.675

4.066

2.924

100.500

100.000

13.660 *SCC indicates squamous cell carcinoma

15.790

(antigen detected by radio¬ immunoassay); CEA, carcinoembryonic antigen; and LASA, lipid-

associated sialic acid.

group was 3.02 ng/mL, while that of the control group was 1.77 ng/mL. Serum CEA levels greater than 4.00 ng/mL were ob¬ served in 41 (40.6%) of the patients with cancer and 16 (18%) of the controls. These findings were similar to those of Silverman et al.5 The mean CEA level in patients with cancer was 4.066 ng/mL, while that for the control group cancer

was

2.92

ng/mL.

The mean level of LASA was 100.00±25.79 mg/dL in the 88 control patients. Therefore, serum sialic acid levels greater than 125.00 mg/dL were established as elevated. This control level was not in agreement with previous studies, which showed a mean of 59.95 mg/dL for healthy controls and 95.91 mg/dL for patients with cancer. Sialic acid level was elevated in 17 (16.8%) of the 101 patients with cancer and in nine (10%) of the 88 controls. The mean for patients with biopsy-diagnosed cancer was 100.50

mg/dL.

Downloaded From: http://archotol.jamanetwork.com/ by a Georgetown University Medical Center User on 05/27/2015

Table 5.—Correlation of SCC and CEA*

Table 4.—Distribution by Number of Elevated Marker Levels* Total No. of Elevated Marker Levels

(n=101 )

Control (n=88)

At least 1

74 (73.3)

46 (52)

3

6 (5.9)

2

28(27.7) 40(39.6)

(2.3) 8(9.1) 36(41)

Only 'Values

are

1

Cancer

Elevated Marker Levels SCC

CEA

or

CEA

Cancer 69 (68.3)

Control 33

(38)

(40.6) 48(47.5) 20(19.8)

16(18) SCC 19(22) SCC and CEA 2(2.3) *Values are number (percent). SCC indicates squamous cell carci¬ noma (antigen detected by radioimmunoassay); CEA, carcinoembryonic

2

number (percent).

41

antigen. Table 6.—SCC and CEA Controls (n=88)

Stage

33 19 16 2

6(75) 6(75) 1 (12)t 1 (12)

by Stage* Stage

I

(n=8)

II

(n=19)

Stage

III

(n=32)

Stage

IV

(n=33)

Elevated marker level SCC or CEA SCC CEA SCC and CEA Mean level, SCC CEA

ng/mL

(38) (22)

(18) (2.3)

1.675 2.924

12 (63)

7(37) 7(37) 2 (10)

4.695 3.159

2.28 3.99

21 (66) 13 (41) 12 (38)

26 (79) 18 (54)

4(12)

9(27)

2.09 3.98

3.84 4.08

17(52)

*Except for SCC and CEA mean levels, values are number (percent). SCC indicates squamous cell carcinoma (antigen detected by radioimmunoas¬ say); CEA, carcinoembryonic antigen. tOne

patient with stage I disease who

was a

nonsmoker had

a

level of 3.71

A serum level of CA-125 above 25.00 U/mL was estab¬ lished as elevated based on the control mean (±SD) of 15.76±11.22 U/mL. The mean level of CA-125 was 13.66 U/mL in the cancer group and 15.70 in controls (P>.5). The CA-125 level was elevated in eight (7.9%) of the patients with cancer and 14 (16%) of the control group. Table 2 demonstrates the number of patients with elevations for each of the four markers. The mean values for each of the markers is demonstrated in Table 3. All four determinations per patient were combined to determine the sensitivity of all markers examined together. The total number of patients with levels one or more of the substances elevated was 74 (73.3%) for those with cancer and 46 (52%) of the controls (P=.15). Only six (5.9%) of the 101 patients with cancer had lev¬ els of three markers elevated. Only 28 (27.7%) of the can¬ cer group and eight (9.1%) of the control subjects had lev¬ els of two markers elevated. The total number of patients with only one level elevation was 40 (39.6%) of the patients with cancer and 36 (41%) of the control group (P=.91) (Ta¬ ble 4). CA-125 and LASA CA-125 of and LASA

serum levels was utility to markedly compromised owing multiple false-positive results (poor specificity). Hence, the similarity between levels in the cancer and control populations led to nonsig¬

The

nificant results.

SCC and CEA The SCC and CEA levels were studied in combination. The level of at least one of the two substances was elevated in 69 (68.3%) of the patients with cancer and 33 (38%) of the controls (P=.02). Specificity was improved by identifying those patients with levels of both markers elevated. Twenty (19.8%) of 101 patients with cancer, and only two (2.3%>) of 88 controls, had levels of both markers elevated

(P=.0008) (Table 5).

ng/mL.

Patients with either SCC or CEA level elevations were then stratified by site of disease, dividing the patients with cancer into five categories. The categories consisted of lar¬ ynx, oropharynx, hypopharynx, oral cavity, and miscella¬ neous (unknown primary, maxillary sinus, ear, skin). In 34 patients with laryngeal sites, SCC level was elevated in 19 (56%), while CEA level was elevated in only nine (26%). Of the 14 patients with oropharyngeal sites, SCC level was el¬ evated in only four (29%), but CEA level was elevated in nine (64%). This was similar to the group of 13 patients with hypopharyngeal sites, which had SCC level elevated in just four (31 %) but CEA level elevated in eight (62%). Of 34 patients with oral cancer, SCC level was elevated in 16 (47%), while CEA level was elevated in 12 (35%). Of the six patients with miscellaneous sites, SCC and CEA levels were both elevated in four (67%). The SCC and CEA levels were then stratified by the stage of disease (Table 6). The SCC level seemed to correlate well with stage, with the exception of patients with stage I dis¬ ease. The SCC levels were elevated in six (75%) of eight patients with stage I disease, and three patients had the highest mean level of SCC of any stage. Carcinoembryonic antigen level correlated well with stage of disease both in number of elevated levels and in mean levels of CEA. At least one of the two markers was elevated in 33 (38%) of the controls. Similar elevations were noted in 26 (79%) of 33 with stage IV disease, 21 (66%) of 32 with stage III disease, 12 (63%) of 19 with stage II disease, and six (75%) of eight with stage I disease. Both CEA and SCC levels were elevated in only two (2.3%o) of the controls, nine (28%) of 33 with stage IV disease, four (12%) of 32 with stage III disease, two (10%0 of 19 with stage II disease, and one (12%) of eight with stage I disease.

COMMENT The search for a reliable circulating serum marker for SCC of the head and neck remains elusive. Although nu¬ merous markers have been identified that are found in el-

Downloaded From: http://archotol.jamanetwork.com/ by a Georgetown University Medical Center User on 05/27/2015

evated levels in a percentage of patients with SCC of the head and neck, none of these markers has either the spec¬ ificity or sensitivity to serve as useful clinical indicators of the presence of disease. It was postulated that by combining different markers, sensitivity and specificity could be im¬ proved to the degree that clinical usefulness could be ad¬ justed. We therefore selected four markers, each having been demonstrated to be effective in epithelial malignant neoplasms and each apparently measuring a different cir¬ culating substance. Unfortunately, our results failed to pro¬ duce the results that we had anticipated. Using four makers only resulted in a modest improvement in sensitivity when compared with single markers; moreover, over 50% of the control patients had elevated levels of at least one marker. Hence, the specificity of the battery of the four markers was no better than a coin toss. Attempts to utilize two markers added slightly to the sensitivity, but again at great compro¬ mise to the specificity of the test battery. Two of the four markers, LASA and CA-125, demon¬ strated essentially no utility in this patient population, with elevations of one or the other found in more than one third of the control population, compared with approxi¬ mately two thirds of the cancer-bearing population. Al¬ though the results were somewhat better utilizing SCC and CEA, nevertheless, the addition of CEA appeared to add little to the SCC level determination, which was elevated in approximately 55%. Inadequate specificity was noted with all markers. Nineteen (22%) of 88 controls had elevated SCC levels, and 16 (18%) had elevated CEA levels. Although the numbers of controls with elevated levels were lower with LASA and CA-125, the numbers of elevated levels were also lower in tumor-bearing patients. In fact, elevated levels were found more commonly in control patients than in tumor-bearing patients when CA-125 was studied. The search for a marker with sufficient specificity and sensitivity for patients with carcinoma of the head and neck continues, since the battery of four markers studied herein did not provide data that would help in patient treatment.

CONCLUSION The identification of a circulating tumor marker has great potential for clinical application. Unfortunately, efforts to date to identify a single tumor marker or a com-

bination of markers for SCC of the head and neck have been frustrated by the relatively poor sensitivity and lack of specificity demonstrated by existing markers. In this study, a panel of markers failed to improve diag¬ nostic sensitivity or specificity. Accordingly, we conclude that application of SCC-RIA, CA-125, LASA, and CEA ei¬ ther singly or in combination for screening purposes is not indicated. The potential usefulness of these markers to fol¬ low up the clinical course of marker-positive patients has been previously reported and deserves some consider¬ ation. However, the community of head and neck oncol¬ ogists is still in need of a highly specific marker with req¬ uisite sensitivity to enjoy clinical application. This investigation was supported in part by the Mary Hillman Jen¬ nings Foundation, which is part of The Eye & Ear Institute of Pitts¬ burgh (Pa). References 1. Goodwin WJ, Kirchner JC, Sasaki CT. Plasma tumor markers associated with squamous carcinoma of the upper aerodigestive tract. Head Neck Oncol Res. 1988:67-73. 2. Johnson JT, Wagner RL, Eibling DE, Su S. Radioimmunoassay for SCC antigen in the diagnosis of squamous cell carcinoma of the head and neck: a preliminary report. In: Kato H, deBruijn HWA, Ebert W, Herberman RB, Johnson JT, eds. 5CC Antigen in the Management of Squamous Cell Carcinoma.

Princeton, NJ: Excerpta Medica; 1987:112-123.

3. Eibling DE, Johnson JT, Wagner RL, Su S. SCC-RIA in the diagnosis of squamous cell carcinoma of the head and neck. Laryngoscope. 1989;99: 117-124. 4. Fischbach W, Meyer T, Barthel K. Early results concerning the clinical usefulness of SCC antigen in oral and facial squamous cell carcinoma. In: Kato H, deBruijn HWA, Ebert W, Herberman RB, Johnson JT, eds. SCC Antigen in the Management of Squamous Cell Carcinoma. Princeton, NJ: Excerpta Medica; 1987:124-129. 5. Silverman NA, Alexander JC, Chretien PB. CEA levels in head and neck cancer. Cancer. 1976;37:2204-2211. 6. Schneider M, Demard F, Chauvel, et al. Carcinoembryonic antigen determinations in head and neck cancer. In: Krebs BP, Lalanne CM, Schneider

M, eds. Clinical Applications of Carcinoembryonic Antigen Assay: Proceedings of a Symposium, Nice, October 7-9, 1977. Princeton, NJ: Excerpta

Medica; 1978:384-387.

7. Szymendera JJ. Review: clinical usefulness of three monoclonal antibody-defined tumor markers: CA19-9, CA 50, and CA125. Tumour Biol.

1986;7:333-342. 8. Kawahara M, Terasaki PO, Chia D, Johnson C, Hermes M, Tokita K. Use of four monoclonal antibodies to detect tumor markers. Cancer. 1986;58: 2008-2012. 9. Voigtmann R,

the

cer.

Pokorny J, Meinshausen A. Evaluation and limitations of lipid-associated sialic acid test for the detection of human cancer. Can1989;64:2279-2283.

10. Kato H, Torigoe T. Radioimmunoassay for tumor antigen of human cervical squamous cell carcinoma. Cancer. 1977;40:1621-1628.

Downloaded From: http://archotol.jamanetwork.com/ by a Georgetown University Medical Center User on 05/27/2015