Preoperative carcinoembryonic antigen level as a prognostic indicator in resected primary lung cancer

Preoperative Carcinoembryonic Antigen Level as a Prognostic Indicator in Resected Primary Lung Cancer Philippe Icard, MO, [ean-Francois Regnard, MO, Arthur Essomba, MO, Vincenzo Panebianco, MO, Pierre Magdeleinat, MO, and Philippe Levasseur, MO Centre Chirurgical Marie Lannelongue, Le Plessis Robinson, France

The aim of this study was to evaluate the prognostic significance of elevated preoperative carcinoembryonic antigen (CEA) levels in cases of resected primary lung cancer. Between 1985 and 1989, 152 patients with tumors and CEA levels above 10 nglmL underwent operation. One hundred twenty-five of them underwent resection of their tumors and the other 27 underwent exploratory thoracotomy only. Fifty-two percent of cancers were adenocarcinomas and 33% were epidermoid. Forty-two resected tumors were classified as stage I, 29 as stage II, 45 as stage lIla, 7 as stage IIIb, and 2 as stage IV. The 3-year actuarial survival rate was 54% for patients with stage I tumors, 28% for those with stage II, 18% for those with stage IlIa, 44% for those with stage IIIb, and 0% for those with stage IV tumors. The 5 year actuarial survival was 40% for those with stage I tumors, 28% for those with stage II, 7% for those with stage IlIa, and 0% for those with stage IIIb tumors. Preoperative CEA levels increased from stage I to stage lIla (p < 0.05). However,

based on preoperative CEA levels we were not able to predict resectability, because levels were not significantly different between stage lIla and exploratory thoracotomy-only groups. Adenocarcinoma was not significantly associated with higher CEA levels than was epidermoid, except in stage lIla disease (p < 0.05). We found a critical unfavorable level of prognostic significance at 30 ng/mL. Within patients who underwent resection of stage I or II tumors, those with preoperative CEA levels under 30 nglmL demonstrated significantly prolonged survival over those with CEA above 30 ng/mL (p < 0.05). Virtually all patients with marked elevations of CEA levels (>50 ng/mL) died within 2 years. Therefore, these patients must be highly suspected of having metastases even if operative staging may appear limited. Determining preoperative CEA levels provides prognosis information which may supplement that available by staging. (Ann Thorae Surg 1994;58:811-4)

C

Material and Methods

arcinoem bryonic antigen (CEA) is an oncofetal antigen that normally is present during fetal life, occurs at low concentrations in adults, and circulates in high concentrations in patients with certain malignancies, particularly epithelial tumors [l]. Because CEA did not appear useful in predicting the operability of patients with primary lung cancer [2-6], it appeared to be of limited interest. Furthermore, its prognostic value has been studied rarely and the survival of resected primary lung cancers with elevated preoperative CEA levels (> 6 ng/mL) was reported to be very poor, not exceeding 3 years [2, 3] with a median survival of 3 months when the CEA level was higher than 15 ng/mL [4] or 20 ng/mL [5]. The purposes of this study were to evaluate both the survival of patients with resected primary lung cancers and elevated preoperative CEA levels, and the prognosis interest of this tumoral marker.

Accepted for publication Jan 24, 1994. Address reprint requests to Dr Regnard, Centre Chirurgical Marie Lannelongue, 133 Avenue de la Resistance, 92350 Le Plessis Robinson, France.

© 1994 by The Society of Thoracic Surgeons

Our study is based on patients with elevated preoperative CEA levels who were undergoing operation of bronchogenic carcinoma at Marie Lannelongue Hospital between 1985 and 1989. There were 152 patients (127 men and 25 women) with an average age of 60 years (range, 44 to 79 years). These patients represented 8.3% of the 1830 patients with primary lung cancers who underwent operation in our institution during the same period. CEA levels were measured by the modified Z-gel technique [7]. All patients had preoperative CEA levels greater than 10 nanograms per milliliter (ng/mL); the normal range according to other studies [1,8,9] is less than 5 ng/mL in nonsmoker patients and less than 10 ng/mL in smoker patients. Resectability was studied by physical examination and morphologic studies. Morphologic examinations systematically included chest roentgenogram, computed tomography of chest and upper abdomen (including liver and adrenal glands), and ultrasonography of the liver. Routine brain scanography was not performed in the absence of neurologic symptoms suggestive of brain involvement. 0003-4975/94/$7.00

812

ICARD ET AL CARCINOEMBRYONIC ANTIGEN IN RESECTED LUNG CANCER

Table 1. Preoperative Carcinoembryonic Antigen Levels and Extent of Cancer

Stage I

Mean CEA (::': SEM) (ng/mL)

42 29 45 7 2

II

IlIa I1Ib

IV a

No. of Patients

26::': 61 ::': 163::': 41 ::': 69::':

3 21 56 5 3.5

Log CEA 2.96 3.34 3.75" 2.87 4.09 a

=

carcinoembryonic antigen;

differences between several classes, and linear regression analysis was performed for correlation studies. The distribution of CEA levels between groups was compared using a >? test. Survival rates were calculated by the KaplanMeier method and compared by the log rank test [11]. Results were considered to be significant if the p value was less than 0.05.

Results Preoperative CEA, Staging, and Resectability

p < 0.05, compared with stage I.

CEA

Ann Thorac Surg 1994;58:811-4

SEM

~

standard error of the mean.

Surgical Procedure and Extent of Cancer All patients underwent operation. One hundred and twenty-five underwent resection of their cancer and 27 patients underwent exploratory thoracotomy only. Ninety-three out of the 125 resections (74%) were considered "complete and curative," whereas 32 of the 125 resections (26%) were considered incomplete and palliative because of macroscopic or microscopic remnants of cancer. The resections consisted of 72 lobectomies (58%), 45 pneumonectomies (36%), 4 bilobectomies (3%) and 4 wedge resections (3%). Eighteen of the 45 pneumonectomies were extended pneumonectomies due to invasive cancers. These extended resections included the vena cava (1), the chest wall (2), the atrium (3), the carina (3), the pericardum (3), the mediastinal pleura (5) and the proximal pulmonary artery (1). According to the TNM classification [10], 42 (34%) of the resected tumors were described as stage I (14 T1 NO, 28 T2 NO), 29 (23%) as stage II (3 T1 N1, 26 T2 Nl), 45 (36%) as stage IlIa (28 T1 T2 N2, 10 T3 NO, 7 T3 N2), 7 (6%) as stage I1Ib (7 T4 NO), and 2 (1%) as stage IV (2 M1; brain metastases previously operated). Histologic examination of resected cancers showed an adenocarcinoma in 66 patients (53%), an epidermoid carcinoma in 42 (35%), a mixed epidermoid and adenocarcinoma in 15 (12%), and a small cell lung cancer in 2 patients (1.5%). Two of the 66 adenocarcinomas were of bronchioloalveolar type. Histologic examination of tumor tissues of the 27 patients who underwent exploratory thoracotomy only revealed adenocarcinoma in 13 patients (48%), epidermoid cancer in 8 patients (30%), a mixed form in 1 patient (4%), small cell lung cancer in 3 patients (11%), and undifferentiated cancer with large cells in 2 patients (7%).

Follow-up and Statistical Analysis Follow-up was obtained for all patients until either death or completion of the study (May 1992). All deaths were attributed to cancer disease. Mean CEA values were calculated for each individual stage, or histologic pattern. When the distribution of CEA measurements was found to be asymmetric and skewed to the right due to the influence of markedly elevated CEA levels, the asymmetry was normalized by the logarithm of the mean CEA. The statistical significance of differences in the logarithm of the mean CEA for each class was calculated using the two-tailed Student's t test. Variance analysis was used to calculate

The mean preoperative CEA value was 64 ± 17 ng/mL (mean ± SEM) (10 to 2,000 ng/mL). The relationship between preoperative CEA levels and the extent of tumor spread is shown in Tables 1 and 2. There was a significant difference (p < 0.05) between preoperative CEA levels in patients with stage I and those with stage IlIa tumors. Linear regression analysis disclosed a positive correlation (r = 0.272) between stage (I through IlIa) and log CEA (p = 0.01). Furthermore, there were more patients with marked elevated CEA values (>30 ng/mL) in stages IlIa and II than in stage I; levels higher than 30 ng/mL were found in 17 of 45 (38%) stage IlIa patients and 11 of 29 (38%) stage II patients, significantly (p < 0.05) more than were found in stage I (7 of 42, 17%) patients (Table 2). CEA levels were lower in stage I1Ib than in stages II or Illa, but all 7 stage IIIb patients had NO disease. Mean preoperative CEA level was 107 + 14.1 ng/mL in the exploratory thoracotomy-only group. Preoperative CEA levels were not significantly different between patients who underwent resection of a stage IlIa cancer and those who underwent no resection (exploratory thoracotomy only). Fourteen of the 27 patients (52%) who underwent exploratory thoracotomy only had preoperative CEA levels less than 20 mg/mL.

Preoperative CEA and Histology The relationship between preoperative CEA and histology (adenocarcinoma, epidermoid cancer, or mixed forms) is shown in Table 3. Preoperative CEA levels were significantly higher for adenocarcinoma than for epidermoid tumors (p < 0.05). However, within each stage, this significant difference held only for stage IlIa. Moreover, as shown in Table 4, levels higher than 30 ng/mL were found in 24 of 64 (37.5%) adenocarcinomas and in 8 of 42 (19%) epidermoid tumors (p < 0.1).

Table 2. Distribution of Preoperative Carcinoembruonic Antigen Levels by Stage

Preoperative CEA Levels (ng/mL) Stage I II

IlIa I1Ib

IV CEA

=

10-20

21-30

31-50

>50

28 (66%) 13 (45%) 23 (51%) 6(86%) 0

7 (l7%) 4 (l4%) 5 (11%) 1 (l4%) 0

3(7%) 5 (l7%) 5 (l1 %) 0 0

4 (l0%) 7(24%) 12 (27%) 0 2 (l00%)

carcinoembryonic antigen.

ICARD ET AL CARCINOEMBRYONIC ANTIGEN IN RESECTED LUNG CANCER

Ann Thorac Surg 1994;58:811-4

Table 5. Survival by Stage and Preoperative Carcinoembryonic Antigen Levels

Table 3. Preoperative Carcinoembryonic Antigen and Histology by Stage Stage

N

Total Adenocarcinoma Epidermoid Mixed Stage I Adenocarcinoma Epidermoid Stage II Adenocarcinoma Epidermoid Stage IlIa Adenocarcinoma Epidermoid

Mean CEA (± SEM) (ng/mL)

66 42 15

107.5 ± 32 39.3 ± 12 57 ± 9

23 17

24.6 ± 3.5 22.1 ± 5

17 11

63 ± 25 63 ± 42

26 14

222 ± 83 47.6 ± 21

CEA = carcinoembryonic antigen; standard error of the mean.

NS

=

Log CEA <0.05 3.53 3.053 3.21 NS 3.01 2.88 NS 3.44 3.17 <0.05 4.05 3.16 SEM

not significant;

~

Conversely, mildly elevated CEA levels (under 20 mg/mL) were observed significantly more frequently in epidermoid cancers (30 of 42, 71%) than in adenocarcinomas (32 of 66, 48%; P < 0.02).

Preoperative eEA and Survival Six patients died within 30 days after operation (operative mortality, 3.9%). Survival rates were significantly better (p < 0.05) for patients with preoperative CEA levels under 30 ng/mL (versus levels above 30 mg/mL). We have to point out that 23 of 25 (92%) patients with preoperative CEA levels exceeding 50 ng/mL died within 2 years after surgical resection, and only 1 patient from this group was alive 5 years after operation. The relationships between staging, preoperative CEA levels, and survival rates are shown in Table 5. The 3 year survival rate was respectively 54% for patients with stage I tumors, 28% for those with stage II, 18% for those with stage Illa, 44% for those with stage IIIb, and 0% for those with stage IV tumors. The 5 year actuarial rate was respectively 40% for those with stage I tumors, 28% for those with stage II, 7% for those with stage lIla tumors. For stage I and stage II tumors, survival of patients with preoperative CEA levels greater than 30 ng/ mL was significantly lower (p < 0.02 for stage I, p < 0.05 for stage II) than survival of patients with CEA levels less than 30 ng/mL.

Table 4. Distribution of Carcinoembryonic Antigens in Adenocarcinoma and Epidermoid Tumors Preoperative CEA Levels (ng/mL) Type of Tumor Adenocarcinoma Epidermoid CEA

=

10-20

21-30

(30-50)

>50

32 (48%) 30 (70%)

10 (15%) 4 (9.5%)

24 (36%) 8 (19%)

14 (21%) 6 (14%)

carcinoembryonic antigen.

813

Survival (%)

Total No. of Patients

2 years

3 years

10-30 >30

42 35 7

63 (18)" 69 (1) 40 (2)

54(9) 58 (8) 40 (1)

10-30 >30

29 17 12

33 (8) 44 (6) 27(2)

28 (4) 44 (3) 14 (1)

IlIa 10-30 >30 IlIb 10-30 >30 IV

45 28 17 7 7 0 2

25 (8) 23 (5) 26 (3) 44 (1) 44 (1)

18 (6) 19 (4) 18 (2) 44 (1) 44 (1)

Stage/CEA Level (ng/mL)

II

a

0

5 years <0.02a 40 (2) 49 (2) 0 <0.05 28 (2) 44 (2) 0 NS 7 (1) 0 9(1)

0

Numbers in parentheses are number of patients eligible.

Comment This study confirms previous reports [2-6] indicating that preoperative CEA levels cannot predict the resectability of lung cancer, although we observed a trend in preoperative CEA levels reflecting the extent of cancer. Adenocarcinoma was the main histologic type in our series (53%). CEA levels were lower for patients with stage IIlb tumors than for those with stage lIla or stage II tumors, and this may reflect the NO status of our stage IIlb patients. There were no significant differences in preoperative CEA levels between adenocarcinomas and epidermoid cancers, except for those with stage IlIa tumors. Adenocarcinomas outnumbered epidermoid tumors and were associated with higher preoperative CEA levels than epidermoid tumors in patients with stage lIla tumors. Survival after lung resection in patients with elevated preoperative CEA levels (>10 ng/mL) was markedly better than rates previously reported [3-5]. In two studies [4, 5] survival was very poor, not exceeding a few months, when preoperative CEA levels were greater than 15 to 20 ng/mL. Concannon and coworkers [3] indicated that a1147 of their patients with CEA levels above 6 ng/mL in their study died within 3 years after operation. In our series, we report a 5 year survival rate of 40% for those with stage I tumors, 28% for those with stage II and 7% for those with stage IlIa tumors. However, these survival rates, particularly for stage I and II tumors, appear less favorable than those reported recently for lung cancers resected at equal stages [12-18]. Merlier and coworkers [15, 16] reported a series of 1285 patients operated from 1960 to 1973. The 5 year survival rate was 53% for those with stage I tumors (66% in TINO), 42% for those with stage II tumors, and 12% for those with N2 disease. In a recent review of 1737 staged patients, Naruke and coworkers [18] have reported 5 year survival rates as

814

Thorac Surg 1994;58:811-4

ICARD ET AL CARCINOEMBRYONIC ANTIGEN IN RESECTED LUNG CANCER

follows: patients with stage I tumors, 65%; with stage II, 42.9%; with stage Illa, 22.5%; with stage I1Ib, 5.6%; and with stage IV, 7.5%. Thus, a 65% to 70% 5 year, disease-free survival rate after resection for patients with stage I tumors, and a 42% to 56% 5 year survival rate for stage II patients are now the accepted norms [12, 18]. The 18% unresectability rate during thoracotomies and the 26% rate of incomplete resections in our series reflected the advanced stage of patients with elevated preoperative CEA levels, the insufficiency of preoperative assessment of mediastinal or diffuse pleural involvement, and our aggressive surgical policy (18 of the 45 pneumonecties were extended resections). However, this policy has not led to a high mortality rate (3.9%). Presently, to better assess resectability we perform preoperative mediastinoscopy or video-thoracoscopic exploration in selected cases just before thoracotomy. Is there a critical level of preoperative CEA of prognostic significance? Vincent and coworkers [4] reported that all but 7 of their 49 patients with CEA concentrations greater than 15 ng/mL had locally advanced inoperable diseases or disseminated diseases. The 7 patients with values greater than 15 ng/mL and considered to have limited disease were operated upon and 6 were resected. All of these patients died within 1 year of surgery, with evidence of remote metastatic spread. Concannon and coworkers [3] indicated that a preoperative CEA level greater than 20 ng/mL reflected advanced disease because 16 of 21 patients (73%) with values in this range had stage IV diseases at the time of thoracotomy. However, these authors [2-4] were unable to demonstrate a critical level of CEA which would be of prognostic significance. Dent and coworkers [5] reported a 3 month median survival time in patients with CEA values greater than 20 ng/mL and only 3 of 24 such patients lived longer than 9 months postoperatively. Stokes and coworkers [6] reported that all of their 7 patients with preoperative CEA levels above 40.9 ng/mL (their normal upper limit was 20.9 ng/mL) had recurrence of metastases within 31/2 months after apparent curative resection, and none of these patients lived longer than 10 months. In our study, we have found a critical level of prognostic significance at 30 ng/mL. In effect, patients with preoperative CEA exceeding 30 ng/ ml., and particularly those who underwent resection of a stage I or stage II tumor, had significantly worsened survival than patients at equal stages with preoperative CEA levels less than 30 ng/mL. Only 4 patients in our series with preoperative CEA above 30 mg/mL were alive more than 3 years postoperatively. Marked elevation of CEA (>50 ng/mL) was associated with poor survival (all but 2 of our 25 patients with such values died within 2 years after operation). Consequently, such a rise in preoperative CEA concentrations casts great doubt on the probability of long-term survival, even if initial operative staging appeared limited. Eleven of 25

Ann

(44%) of our patients with CEA levels greater than 50 ng/mL were initially considered to have stage I or II diseases. Therefore, such patients with marked elevation of preoperative CEA are highly suspect of having metastases; consequently, thorough investigations are necessary before operation. Furthermore, in these patients, postoperative complementary chemotherapy must be discussed. In conclusion, determining preoperative CEA levels provides prognosis information which is supplemental to that available by staging.

References 1. Fletcher RH. Carcinoembryonic antigen. Ann Intern Med 1986;104:66-73. 2. Concannon JP, Dalbow MH, Liebler GA, Blake KE, Weil CS, Cooper JW. The carcinoembryonic antigen assay in bronchogenic carcinoma. Cancer 1974;34:184-92. 3. Concannon JP, Dalbow MH, Hodgson SE et al. Prognostic value of preoperative carcinoembryonic antigen (CEA)plasma levels in patients with bronchogenic carcinoma. Cancer 1978; 42:1477-83. 4. Vincent RG, Chu TM, Fergen TB, Ostrander M. Carcinoernbryonic antigen in 228 patients with carcinoma of the lung. Cancer 1975;36:2069-76. 5. Dent PB, McCulloch PB, Wesley-James 0, MacLaren R, Muirhead W, Dunnett CWo Measurement of carcinoembryonic antigen in patients with bronchogenic carcinoma. Cancer 1978;42:1484-91. 6. Stokes TC, Stevens JFS, Lockey E, Miller L. Preoperative carcinoembryonic antigen and survival after resection of lung cancer. Br J Dis Chest 1980;74:390-4. 7. Hansen JH, Snyder JJ, Miller E et al. Carcinoembryonic antigen (CEA) assay. A laboratory adjunct in the diagnosis and management of cancer. Hum Pathol 1974;5:139-47. 8. Alexander [C, Silverman NA, Chretien PB. Effect of age and cigarette smoking on carcinoembryonic antigen levels. JAMA 1976;235:1975-9. 9. Clarke C, Hine KR, Dykes PW, Whitehead TP, Whitefield AGW. Carcinoembryonic antigen and smoking. J R Coli Physicians Lond 1980;14:227-8. 10. Mountain CF. A new international staging system for lung cancer. Chest 1986;89(Suppl):225-33. 11. Kaplan EL, Meier P. Non parametric estimation from incomplete observations. J Am Stat Assoc 1958;53:499-505. 12. Martini N, Beattie EJ. Results of surgical treatment in stage I lung cancer. J Thorac Cardiovasc Surg 1977;74:499-505. 13. Martini N, Flehinger BI,Nagasaki F, Hart B. Prognosis significance of N1 disease in carcinoma of the lung. J Thorac Cardiovasc Surg 1983;86:646-53. 14. Martini N, Flehinger B.The role of surgery in N2 lung cancer. Surg Clin North Am 1987;67:1037-49. 15. Merlier M, Le Brigand H, Rojas-Miranda A et al. Etude retrospective de la survie eloignee apres exerese dun cancer bronchique primitif (cancer a petites cellules exclu). Chirurgie 1983;109:590-7. 16. Merlier M, Rojas-Miranda A, Gharbi N, Silbert D, RansonBitker B.The staging issue: unification of criteria. International trends in surgery. Philadelphia: Saunders, 1985;1:27-37. 17. Mountain CF. Assessment of the role of surgery for control of lung cancer. Ann Thorac Surg 1977;24:365-73. 18. Naruke T, Goya T, Tsuchiya R, Suemasu K. Prognosis and survival in resected lung carcinoma based on the new international staging system. J Thorac Cardiovasc Surg 1988;96: 440-7.