Cyfra 21-1, a new marker of lung cancer

Nucl.

Pergamon

Med.

Biol.

Elsevier

Vol. 21, No. 3, pp. 471-476,

Science

lYY4

Ltd. Printed in Great Britain 0969-8051194 $6.00 + 0.00

0969-8051(94)E0025-E

Cyfra 21-1, a new marker of lung cancer J. GRENIER*,

* * *

F. GUILLEUX*, F.B. MICHEL**

J.L. PUJOL**, H. PUJOL* and

J.P.

DAURES***,

Centre R&gional de Lutte contre le Cancer, Laboratoire de Radioanalyse, Montpelkr, France Cliniques des Maladies Respiratoires, H&pita1 Arnaud de Villeneuve, Montpellier, France Dbpartement de I’lnformation Medicale, HGpital Lapeyronie, Montpellier,

France

ABSTRACT

The are few outstanding sensitivity

and specificity

serous markers in the treatment

of bronchial

cancer,

ACE lacks

and cannot be used as a diagnositic marker. It has been described as

a marker of tumoral mass, although but only in a univariate

analysis.

not

in our study. Prognostic significance

Sensitivity

of SCC TA4 varied

population, the ROC curve for SCC TA4 showed poor discrimination

was observed,

between studies.

In our

potential. NSE was shown

to be a useful marker in the treatment of patients with small cell cancers.Cytokeratins expressed

by all bronchial

cancers. Cytokeratin

and their neoplastic counterparts.

19 is a sub-unit

detected in simple epithelia

During tumoral cell lysis, certain fragments of this cytokeratin

may be liberated. The immunoradiometric

assay described here is able to detect fragments of

cytokeratin

19 (called Cyfra 21-1)

Cyfra 21-1

levels and the cancer stage for NCPC, but not for CPC. In addition,

concentration

are

in serum. Our study established

a correlation

between

Cyfra 21-1

may be used as a tumoral mass marker. Patients with high Cyfra 21 levels must

undergo special treatment to find remote tumors.

All correspondence should be addressed to: J. Grenier, Centre Regional de Lutte contre le Cancer, Laboratoire de Radioanalyse, 34094 Montpellier Cedex 5, FRANCE. 471

472

.J. GRENIERet al.

INTRODUCTION

The increasing rate and seriousness of bronchial cancer make it the leading cause of death by cancer in the male population (1). Among the different histologies, epidermoid cancer appears most frequently; surgical exeresis, although restricted to patients with localized forms of the cancer, is the therapeutic method that offers the highest possibility of long-term survival (2). For other patients, therapeutic techniques such as the combination of radiotherapy and chemotherapy for the locally advanced stages are still under evaluation. At present, diagnostic and therapeutic decisions are based on the carcinological stage and on the evaluation of a general state of being using the performance index. Diverse serous markers have been proposed to contribute to the treatment of patients suffering from lung cancer. The carcinoembryonic antigene is a marker available for organ specificity, its use in bronchial cancer limited due to its lack of sensitivity. The “Squamous cell carcinoma antigen” (SCC TA4) has been evaluated in epidermoid bronchial cancers. Its sensitivity varies from 33 to 61 % according to different studies (3). The “Neuron Specific Enolase” (NSE) is a useful marker for small cell bronchial cancer (4).

Cytokeratins (5) The cytokeratins (a cytoskeletal protein), are part of one of the five classes of the intermediate filament families. These fibres often have one of their extremities in the nuclear membrane, either in the desmosomes or in the plamatic membrane. All intermediate filaments have in common various structural characteristics. One may classify the cytokeratins according to a low molecular weight, found in the epithelium (cytokeratins 7, 8, 18, 19) and a high molecular weight, found in the epidermis. Their filaments are made up of heteropolymers, whose sequence is a tetramer composed of two double helixes containing two polypeptides of type I (acid) and type II (basic); the central alpha helix, formed of 310 amino acids, is relatively constant while the variable regions are situated in the N and C terminal regions. The cytokeratins form a family of distinct yet interrelated genetic products who are synthesized in different epithelial cells. In normal tissue, the cytokeratin filaments are recognized by a large spectrum of antibodies in those cells considered as epithelial. However, even if we can detect these filaments in other tissues, the cytokeratins have to be considered as a specific marker of epithelial differentiation. Each cell expresses different cytokeratins during its evolution; their synthesis depends on the initial amount and the progression of cellular division. In tissue of malignant epithelial tumors and their metastasis, preservation of the expression of cytokeratin filaments has been shown. MATERIALS

AND METHODS

Patients A hundred and sixty five (165) patients suffering from lung cancer were recruted in various Montpellier hospitals from February 1990 to February 1992 as part of the study. Among them were 37 CPC (SCLC), 85 epidermoids (SQC), 28 adenocarcinomas and 15 large

Euro-Immunoanalyse

‘93

473

cell carcinomas. The performance index was estimated according to the “Eastern Cooperative Oncology Group” criteria, and the weight loss percentage was recorded for a 4 month period. The stage of development of the illness was established according to the most recent international TNM classification (6) and according to the cartography of mediastinal adenopathies proposed by the American Thoracic Society (7). The regrouping of the different stages was carried out according to the Montain method, which takes into account the anatomical stages T and N (8). The following exams were performed on all of the patients: a complete clinical exam, a standard radiography, a computed tomographic scan of chest thorax and upper part of the abdomen, a bronchial fibroscopy, and a bone scanner. Controls Cyfra 2 1 was measured for 104 patients suffering from a benign pulmonary pathology.

Cyfra 2 1 assay Each patient’s serum was aliquoted and conserved in liquid nitrogen. Cyfra 21 (Centocor Diagnostic; Cis Bio International) is a radioimmunometric assay. The cytokeratin 19 is recognized by two monoclonal antibodies (KS 19-1 and BM 19-21). The assay is based on the “sandwich” technique on a solid phase. The first antibody is absorbed on the solid phase and the second antibody, radiolabelled with iodine 125 is used as a tracer. The Cyfra 21 molecules present in the standards, samples or controls to be tested are “sandwiched” between the the two antibodies. Following the formation of the coated antibodylantigenfiodinated antibody sandwich, the unbound tracer is easily removed by a washing step. ihe bound radioactivity, measured with a gamma counter, is proportional to the concentration of Cyfi-a 21 present in the sample, standard or control. The standard has the following values: 0 @ml; 3 ng/ml; 8 @ml; 25 @ml; 50 ng/ml. Quality control of reagents The dilution test (linearity), inter/intra trial reproductibility and a sensitivity evaluation were performed. Reading the standard curve in the low value zone presents no difficulty due to the fact that the difference between the cpm at 0 ng and 3 ng is large enough to prevent overlapping. Statistics Use of ROC curves (9) (1,2). The ROC curves allow the organization of a decision-making strategy which is to determine a threshold according to sensitivity and specificity. This strategy is, in other words, to scan the spectrum of possible thresholds and to obtain for each threshold a corresponding sensitivity and specificity. Sensitivity is defined in the sick population as the % of true positives; specificity is defined in the healthy population as the % of true negatives. The curve represents the relationship between the % of true positives (sensitivity) and the % of false positives (specificity), analyzed for different thresholds of the marker. The markers were compared according to the area under the curve. The marker assays were expressed in median and interquarterly (spread of 25th - 75th percentiles) due to the fact that their distribution does not obey normal law. The comparison of the value of each marker between the different patient subgroupes, defined according to the stage of sickness or ganglionic status, was

474

J. GRENIER et al.

performed using non parametric tests (Mann and Witney test and Wilcoxon test) (Kruskal Wallis test). RESULTS

Assay sensitivity and dependability The assay’s analytic sensitivity was calculated according to the MDC principle, its value established at 0.6 ng/ml. The linearity test shows the linear relationship between concentrations of Cyf?a 21 and successive dilutions performed in the case of elevated serum levels. Analysis of the low value zone of the standard curve shows that there is no overlapping between 0 and 3 ng/ml. Value distribution The mean and variability expression (expressed interquarterly between the 25th and 75th percentile) are significantly higher in patients suffering from cancer than in the control population. The use of ROC curves has permitted the determination of the threshold at 3.6 ng/ml. Cyl?a 21 and histology The mean of Cyfra 21 is: 3.4 @ml for the CPC (SCLC), 5.6 rig/ml for epidermoids (SQC), 2.9 rig/ml for adenocarcinomas and 3.8 ng/ml for large cell carcinomas. The level of Cyfra 21 varies according to the histology (p< 0.05). The level of Cyfra 21 is much higher in the epidermoids than the other histologies. Cyfra 21 and clinical evaluation In the NCPC (MSCLC) the level of Cyfia 21 is much higher for patients in metatstatic evolution (7.4 ng/ml) than in the case of a contained cancer (3.8 ng/ml). When the marker was studied according to the stage of the illness, one observes a significant augmentation between stages I and IV (p < 0.005); there is no significant difference between stages IIIa and IlIb. In addition, patients with mediastinal ganglions (N2 or N3) have a higher Cyfra 21 level (5 @ml) than those without ganglionic invasion (NO or Nl : 3.2 ng/ml). The distribution of Cyfra 21 for the NCPC varies significantly according to the performance index. Cyfra 21 in the CPC (SCLC) does not allow a differentiation between a localized cancer (3 ng/ml) and a spread cancer (4.5 ng/ml) (p = 0.07). Study of Survival In a univaried analysis, it is shown that patients with a Cyfia 21 level that is higher than 3.6 rig/ml have a similar survival rate than those whose Cyf?a 21 is under this threshold. This negative prognostic Cyfra 21 factor is highly significant in the epidermoid group (SQC) with p < 0.001. In a study following the Cox model, Cyf?a 21 places itself after the performance index and the stage of illness, and before the other biological markers used currently. DISCUSSION

This study allowed the comparison of traditionally used markers such as ACE, SCC, TA4, NSE with Cyfra 21. Cyfi-a 21 was the most discriminating marker. A threshold of 3.6

Euro-Immunoanalyse

‘93

47s

ng/rnl was chosen after studying the curve. The level of Cyfra 21 is significantly in correlation with the tumoral mass. Few serous markers are in fact performant in the treatment of bronchial cancer. ACE lacks sensitivity and specificity and can not be used as a diagnostic marker. ACE has been described as a marker of tumoral mass although not so in our study. A prognostic significance was observed, but only in a univaried analysis. The SCC TA4 seems to have a sensititvity that varies from one study to the next. In our population; the ROC curve of SCC TA4 shows a weak ability of discrimination. NSE has been shown to be a marker in the treatment of patients with small cell cancer. The cytokeratins are expressed by all bronchial cancers. Cytokeratin 19 is a sub-unit detected in simple epitheliums and their neoplasic counterparts. During a lysis of tumoral cells, certain fragments of this cytokeratin may be liberated. It has also been shown that the CPC (SCLC) are able to simultaneously express cytokeratins and neurofilaments;.this idea has led to the hypothesis that there may be a known cellular origin for all histological types of bronchial cancer. The immunoradiometric assay described here is able to detect a fragment of cytokeratin 19 (called Cyfra 21-1) in serum. The discrimination threshold of 3.6 ng/ml was determined using the ROC curve, which allows us to find the best compromise between sensitivity and specificity. The positionning of the ROC curve in the upper left octant shows the clinical potential of Cyfra 2 1- 1. Our study has established a correlation between the level of Cyf?a 21-1 and the stage of the cancer for the NCPC but not for the CPC; in addition, level of Cyfra 21-1 increases significantly in the NCPCs with ganglionic invasion. This results suggest that Cyfra 21-1 may be considered as a tumoral mass marker. Patients with a high level of Cyfra 21 must follow a particular treatment in order to find tumors localized at a distance. Despite these important results, it is impossible to predict the operability of NCPC using Cyf?a 21 values due to the overlapping of results between stages IIIa (normally RESECABLE) and IIIb. The absence of this discriminating ability may appear surpising at first because Cyfra 21 seems to correlate well with tumoral mass. In fact, the Montain classification, often used for therapeutic indications in bronchial cancers, takes into account anatomical criteria (of utmost importance in terms of operability) while tumoral mass intervenes very little. Thus, a small tumor that invades CARENE is classed T4 and therefore IIIb while a mass of 8 cm. circumscribed by pulmonary tissue is classed T2 and belongs to a localized stade if there are no mediastinal adenopathies. The slight elevation of Cyfra 21 in stages I and II does not allow its use in early diagnosis. Due to the fact that Cyfra 21 correlates with the extent of the sickness and the performance index, it is not surprising that univaried analysis shows that patients with a high level of Cyfi-a 21 have a smaller survival rate. This is especially true in epidermoid cancers (SQC). Finally, the Cox model shows that the prognostic value of Cyfra 21 is independant and adds a supplementary information to the hierarchy of prognostic factors.

476

J. GRENIERet al.

REFERENCES

1. Stjemsward, J., and Stanley, K. Etiology, epidemiology and prevention of lung cancer. Lung Cancer, 4 : 1l-24, 1988. 2. Mulshine, J.L., Glastein, E., and Ruckdeschel, J.C. Treatment of non-small cell lung Cancer. J. Clin. Oncol., 4 : 1704-1715, 1986. 3. Iannuzzi, M.C., and Scoggin, C.H. State of the art : small cell lung cancer. Am. Rev. Respir. Dis., 134 : 593-608, 1986. 4. O’Connel,J.P. Kris, M.G., Gralla, R.J., Groshen, S., Trust, A., Fiore, J.J., Kelsen, D.P., Heelan, R.T., and Golbey, R.B. Frequency and prognostic importance of pretreatment clinical characteristics in patients with advanced non-small cell lung cancer treated with combination chemotherapy. J. Clin. Oncol., 4 : 1604-1614, 1986. 5. Moll, R., Franke, W.W., Schiller, D.L., Geiber, B., and Krepler, R. The catalog of human cytokeratins : patterns of expression in normal epithelia, tumors and cultured cells. Cell, 3 1 : 1l-24, 1982. 6. Sobin, L.H., Hermanek, P., and Hutter, R.V.P. TNM classification of malignant tumours. 4th edition, UICC Geneva 1987. 7. Tisi, G.M., Friedman, P.J., Peters, R.M., Pearson, G., Carr, D., Lee, R.E., and Selawry, 0. American Thoracic Society : clinical staging of primary lung cancer. Am. Rev. Respir. Dis. 125 : 659-64, 1982. 8. Moutain, CF. A new international staging system for lung cancer. Chest, ‘(s) ; 225s-233s 1986. 9. Beck, J.R., and Schultz, E.K. The use of relative operating characteristics (ROC) curves in test performance evaluation. Arch. Pathol. Lab. Med., 110 : 13-20, 1986.