Blood vessel invasion by tumor cells predicts recurrence in completely resected T1 N0 M0 non-small-cell lung cancer

Blood vessel invasion by tumor cells predicts recurrence in completely resected Tl NO MO non-small-celllung cancer The prognostic significance of traditional and newer tumor cell-related biologic parameters, like deoxyribonucleic acid ploidy (flow cytometry), proliferative activity (expression of proliferating cell nuclear antigen by immunohistochemistry), mitotic count, and intratumoral or peritumoral (or both) blood or lymphatic vessel invasion by tumor cells was investigated in 95 consecutive patients who had T1 NO MO non-small-cell lung cancer and who had operation alone between 1975 and 1985. The median follow-up for the entire group is now 8.3 years, and overall 5-, 10-, and 15-year-survivals were 75 %, 69%, and 61 %, respectively. Twenty-two patients died of either local (n = 3) or systemic (n = 19) recurrent non-small-cell lung cancer, 5 of non-eancer-related causes, 2 of new primary lung cancer, and 1 of an extrathoracic cancer. By multivariate analysis, blood vessel invasion by tumor cells (p = 0.0001) and mitotic count (p = 0.016) were independent predictors of survival; by contrast, the disease-free survival was influenced only by blood vessel invasion (p = 0.0004). The relative risk of death of recurrent non-small-cell lung cancer for low-risk patients (n = 79) was 13.3 (95 % confidence interval, 6.1 to 28.7) times lower than that of high-risk patients (n = 16) (p < 0.0001). The relative risk of manifesting recurrent disease as distant metastasis for high-risk patients was 25.64 (95 % confidence intervals, 8.4 to 77.6) times higher than that of their low-risk counterparts (p < 0.0001). These results provide a rationale for effective systemic adjuvant treatment in completely resected T1 NO MO non-small-cell lung cancer tailored to the individual patients' risk of development of recurrent nonsmall-cell lung cancer. (J THORAC CARDIOVASC SURe 1993;106:80-9)

Paolo Macchiarini, MD,a Gabriella Fontanini, MD,b Michael J. Hardin, PhD,c Hsu Chuanchieh, Phl)," Daniela Bigini, MD,b Silvana Vignati, MD,b Raffaele Pingitore, MD,b and C. Alberto Angeletti, MD,a Pisa, Italy, and Birmingham, Ala.

Operation remains the mainstay of treatment for resectable or marginally resectable non-small-cell lung cancer (NSCLC). Unfortunately, less than 40% fit operFrom the Service of Thoracic Surgery," Institute of Pathological Anatomy and Histology," University of Pisa, Pisa, Italy, and the Department of Biostatistics," University of Alabama at Birmingham, Birmingham, Ala. Supported by grants of the Italian Ministry of Education (60%) and by the Bracali Ferdinando Award. Received for publication Jan. 21, 1992. Accepted for publication July 2, 1992. Address for reprints: Paolo Macchiarini, MD, Department of Thoracic and Vascular Surgery and Heart-Lung Transplantation, Hopital Marie-Lannelongue (Paris-Sud University), 133 Avenue de la Resistance, 92350 Plessis Robinson, Paris, France. Copyright @ 1993 by Mosby-Year Book, Inc. 0022-5223/93 $\.00 + .10

80

12/1/40783

ation at time of initial diagnosis, and their outcome is largely dependent on the pathologic primary .umor and lymph node status.' Patients with T I NOMOdisease have lesions highly curable by operation alone, with 5-year survival in excess of 70%.2,3 Because of this favorable outcome and its high cost-effectiveness, adjuvant therapy is not recommended for Tl NO MO NSCLC. 4 Traditionally, the failure or success of operation for this early-staged disease has been estimated in survival, disease-free intervals, or both, often restricted to the 5-year period after initial operation.f 3, 5, 6 These endpoint observations, however, have a high chance of unintentionally underestimating or overestimating the therapeutic index of operation because they are influenced by low recurrence! and increased prevalence of non-cancerrelated deaths after 5 years after operation.' As a result of these studies.s 3, 5-6 a number of traditional parame-

The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 1

ters, for example, tumors size, histology, and location, have been identified as prognostic determinants of the natural history of the disease. Newer prognostic factors, including deoxyribonucleic acid (DNA) content': 8 and tumor proliferative index," have been and are being evaluated 10 in the research setting, but their prognostic significance in T1 NO MO disease has been studied less. Therefore controversy continues as to whether they provide more, if any, useful clinical information than more traditional prognostic parameters. 1I Based on these backgrounds, we investigated traditional as well as newer tumor cell-related biologic (DNA ploidy, proliferative activity, mitotic count, blood or lymphatic vessel invasion by tumor cells) parameters in a groupof95 consecutiveTl NO MOpatientswith NSCLC who had operation alone. These factors were subsequently used to predict which patients are destined to die of recurrent NSCLC and those who are not. Patients and methods Patient selection. The demographic data, hospital charts, pathologic slides, and tumor registry records of 100 consecutive patients treated between January 1975 and December 1985 for TI NO MO NSCLC were reviewed. All patients had operation alone as treatment and received neither postoperative adjuvant chemotherapy nor thoracic radiation. Preoperative evaluation included a detailed history and physical examination, biochemical profile, chest x-ray examination, lung tomography, bronchoscopy, computed tomography of the chest and upper portion of the abdomen, and abdominal ultrasound scan (after 1982), as well as cardiopulmonary tests. Patients' cardiac risk was graded according to the method of Goldman and colleagues.'? Neither preoperative computed tomography scan of the brain nor routine cervical or anterior mediastinoscopy was performed in the absence of clinical evidence of brain involvement and in the presence of a negative mediastinal window shown by tomographic examination, respectively.I': 14 Surgical procedures. The extent of operation was governed by the location of the primary tumor .13 According to our policy, 14 wedge resection was preferred to standard lobectomy for peripheral and superficially seated lesions crossing or not crossing an intersegmental plane. Careful intraoperative staging at the time of surgical resection was done by dissecting intrapulmonary and hilar nodes and sampling mediastinal lymph node stations. The primary tumor and lymph node status was classified according to the TNM staging system.l ' There were no operative deaths or major perioperative complications. Tumor immunohistochemical studies. All surgical specimens were reviewed. Conventional pathologic features were noted and recorded and included tumor size, histology, and histologic grade as described by the World Health Organization criteria.!" Sections of intrapulmonary, hilar, and mediastinal lymph nodes were reevaluated to confirm the absence of metastases. All paraffin blocks containing tumor tissue from each patient were analyzed. Consecutive sections were stained with hematoxylin and eosin and by the avidin-biotin peroxidase complex method'? for the localization of factor VIII-related antigen.l"

Macchiarini et al. 8 1

A mouse monoclonal Igfij antibody (Dako Corp., Santa Barbara, Calif.), was used to localize the factor VIII-related antigen. The presence of tumor emboli in endothelial-lined channels and their location in the specimen were assessed and designated as follows: (I) intratumorallymphatic vessel invasion or intratumoral blood vessel invasion (BVI), or both, reflecting the presence of tumor emboli within the primary lesion; (2) peritumoral lymphatic vessel invasion or peritumoral BVl, or both, which encompassed lymphatic vessel invasion and BVI peripheral to or at the advancing edge of the primary tumor." Lymphatic vessel invasion was distinguished from BVI based on previously reported morphologic characteristics. I9. 20 The pathologic material was reviewed without knowledge of the patient's outcome. All specimens were evaluated by one pathologist (G. F.), and positive findings were reviewed in a blind fashion by another pathologist (R. P.) for agreement and for consensus on equivocal findings. In the presence of a disagreement, results were recorded as negative for invasion. Tumor cell kinetic study. The proliferative activity ofT I NO MO NSCLC was determined by estimating the expression of the proliferating cell nuclear antigen (PCNA), also known as cyctin" or auxiliary protein for DNA polymerase.F It is a highly conserved 36 kD acid nuclear protein that defines and maps the proliferative compartment of proliferative cells in formalin-fixed, paraffin-embedded archival tissue of node-negative NSCLC. 7 Its synthesis increases in the late G\ phase, reaches its maximum in the S phase.i! declines during the G 2 phase, and is absent during the Go and mitotic phases of the cell cycle." Its expression was studied according to our method,' with the monoclonal antibody PC 10, a new immunoglobulin G 2a (IgG 2a ) 24, 2S obtained from Labometrics S.p.A., Milan, Italy. Briefly, sections (5 ~m) were cut, dewaxed, passed through alcohol, and then immersed for 10 minutes in methanol with 1% hydrogen peroxidase to block endogenous peroxidase activity. Sections were subsequently immersed in phosphate-buffered saline, and PCNA immunostaining was performed using the avidin-biotin-peroxidase complex technique (Vectastain ABC kit, Vector Laboratories, Burlingame, Calif.) with the aid of primary PClO monoclonal antibody (dilution of I :200) overnight. Diaminobenzidine hydrogen peroxidase was employed as a chromogen, and a light hematoxylin counterstain was used." PCNA immunolocalization, including its spatial distribution within cells, was assessed. Absolute counts of PCNA immunoreactivity were made by scoring a minimum of five high-power fields; the percentage of positive cells was determined, and PCNA immunoreactivity was graded as follows: low (I % to 25%), intermediate (26% to 75%), and high (76% to 100%). The mitotic count was assessed at X250 objective magnification and defined as the number of mitoses per ten high-power fields. 14 Flow cytometric analysis. DNA flow cytometry was performed on cell suspensions prepared from 50 ~m sections cut from formalin-fixed, paraffin-embedded tissue from the primary tumor as described previously.' All samples were analyzed by a FACScan flow cytometer (Becton Dickinson Imrnunocytometry Systems, Mountain View, Calif.) coupled with a HewlettPackard computer. At least 35,000 events were collected for each sample. A simple gating by side-forward scatter and red fluorescence signals was performed to cut off the debris, without electronic subtraction of the background. The cellular DN A content (ploidy) was defined as diploid or aneuploid in the pres-

The Journal of Thoracic and Cardiovascular Surgery July 1993

82Macchiarini et al.

Table I. Operative procedures Features

Side of operation Right vs. left Site of operation Upper lobe Middle lobe Lower lobe Extent of operation Pneumonectomy Lobectomy W·~-15e

resection/segmentectomy

Table II. Tumor characteristics Numbers

51/44 64 4 27 8

65 22/0

ence of a single (DNA index of I) or two distinct (DNA index different from I) Go/G] peaks, respectively. Multiploid tumors were included within the group of aneuploid tumors. Histograms were considered interpretable if the coefficient of variation was less than or equal to 8%. Statistical analysis. Patient characteristics were compared with the Wilcoxon rank sum test for continuous variables and the x 2 test for discrete variables. Overall survival was calculated from the date of operation until death or the date of last follow-up (censored). The duration of disease-free survival was measured from the date of operation until first evidence of recurrence (local or metastatic) from primary NSCLC or last date of follow-up for patients who remained alive and disease free (censored); eight patients who died of a cause other than primary NSCLC without evidence of disease were censored at death. The recurrenceofNSCLC, supported by radiographic or laboratory evidence and, whenever possible, by pathologic confirmation, was considered (I) local (if in the ipsilateral thorax) or (2) systemic (contralateral lung or extrathoracic sitej.!" A new primary malignancy either in the lung or in other sites was established by characteristic pathologic findings.' Survival and disease-free survival were estimated by the product-limit method,27and differences in their distribution were evaluated via the log-rank test,28 for univariate analysis and Cox's proportional hazards stepwise model.i? for multivariate analysis. Standard errors associated with the Kaplan-Meier estimator were based on Greenwood's forrnula.t" For those variables determined to be significant by Cox's model, the relative risk and 95% confidence intervals-l-V for survival and disease-free survival were computed according to the methods of Levine and colleagues'? for the 5-, 10-, and IS-year points. Numbers are expressed either as median or as mean ± standard deviation of n number of observations; differences between medians were evaluated via the Mann- Whitney U test. The a priori level of significance was set at p < 0.05; all tests were two sided.

Results Of the 100 patients analyzed, the pathologic material was technically inadequate for pathologic evaluation in five instances. The surgical and tumor characteristics of the 95 patients included in the study are illustrated in Tables I and II, respectively. Sixteen ( 17%) patients had tumors harboring either intratumoral BVI (n = 15) or peritumoral BVI (n = 1) by tumor cells, and the major-

Features

Tumor size (em) Median (range) « vs. ~2 em) Tumor histology Squamous vs. nonsquamous Tumor differentiation Well vs. poorlydifferentiated Tumor proliferation PCNA, %: median (range) Low vs. intermediate* BVIt Absent vs. present Lymphatic vessel invasion* Absent vs. present Mitotic count Median (range) «vs.~13)

DNA ploidy Aneuploid vs. diploid tumours

Numbers

2 em (0.2-3 em) 28/67 62/33 71/24 20 (0-70%) 55/40 79/16 95/0 13 (0-78) 45/50 64/31

'Low, 0% to 25%; intermediate, 26% to 75% peNA reactivity. tOf the 16 tumors displaying BV!, tumor cells were located within the substance of the tumor (n ; 15) or peritumorally (n ; I).

ity of them were aneuploid tumors (n = 14; 88%); none of the pathologic specimens examined expressed either intratumoral or peritumoral lymphatic vessel invasion. The overall proliferative activity ranged from 0% to 70% and was significantly (p < 0.0001) higher for tumors with BVI than for those without BVI by tumor cells (Fig. 1). Tumors with BVI also showed a significantly (p < 0.0001) higher mitotic count (median 33, range 5 to 78) as opposed to those without (median 11, range 0 to 47). The median follow-up for the entire group is now 8.3 years, and the overall 5-, 10- and 15-year survivals were 75%, 69%, and 61%, respectively; 65 patients are currently alive and disease free at 63 to 186 months (Fig. 2). Among the 30 deaths, 5 occurred for noncancer-related causes and 3 for new primary lung (n = 2) or extrathoracic (n = 1, colon) cancers, and the remaining 22 patients died from primary tumor because of either local (n = 3) or systemic recurrence to the brain (n = 6), bone (n = 5), liver (n = 4), adrenal glands (n = 2), skin (n = 1), or contralateral lung (n = 1); their median time to progression was 40 months (range 2 t095 months). The overall disease-free survival is depicted in Fig. 3. By univariate analysis, type of resection (lobectomy and pneumonectomy versus wedge resection), tumor size « versus 2: 2 em), BVI by tumor cells (without BVI versus with BVI), tumor-proliferative activity (low versus intermediate peNA immunoreactivity), and mitotic

The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 1

M acchiarini et al.

80

...Eo-~ ...

•••• 60

~

1= o-( >ol

40

~

-e

•• • •• • • •

••

z

u~

•

20

•

• •

•••• • ••• • •••••••••

••

• •••• •••••• ••••••• ••• •••••• ••

•••••••••••••

•••••••••••••

0

BVI +

1.00

> I:::i

iii

~

0.75

o

a:

D.

0.50

..J

~

sa: ~

0.25

Patients at risk

(70)

0.00

BVI·

Fig. 1. Comparison of the proliferative activitydetermined by PCNA reactivity in TI NO MO NSCLC with (+, n = 16) or without (-, n = 79) either intratumoral (n = 15) or peritumoral (n = I) BYI by tumor cells. Median PCNA reactivityscore was significantly higher in NSCLC positive for BYI by tumor cells than in NSCLC negative for BYI by tumor cells (50% versus 15%; Mann-Whitney U test; p < 0.0001).

(\4)

(2)

+--~--....--~--.--~--.--

a

60

120

180

TIME

Fig. 2. Overall survival(n dard errors.

..J

<

s>a: ::J

(/)

count « versus 2: 13) had statistically significant influence on survival; disease-free survival was influenced by type of resection, BVI by tumor cells, and tumor proliferative activity (Table III). No other clinical or biologic variable influenced these outcomes significantly. By multivariate survival analysis (Table IV), BVI by tumor cells and mitotic count were found to be independent predictors, and the relative risk of surviving less than 5, 10, and 15 years for a patient with BVI tumors and a mitotic count greater than or equal to 13 was, respectively, 15.2, 1.25, and 1.07 times higher than that for a patient without BVI tumors and a mitotic count less than 13. By multivariate disease-free survival analysis, only BVI maintained its independent and significant prognostic value; the relative risk of shortened disease-free survival for patients with BVI was more pronounced in the first 5 years after operation (Table V). The relative risk of development of recurrent NSCLC for an individual patient whose tumor showed BVI by tumor cells (high risk) was 13.3 (95% confidence intervals, 6.1 to 28.7) times higher than that of a patient whose tumor did not show BVI by tumor cells (low risk). There were 79 lowrisk and 16 high-risk patients, and the difference in their disease-free survival is reported in Fig. 4. Since the cumulative risk of recurrence for low-risk patients was 8.4%, the estimated cumulative risk of recurrent NSCLC for high-risk patients was in the order of 111.7% (13.3 X 8.4).

83

(months)

= 95); vertical bars indicate stan-

1.00

0.75

w

w a: u. I

w

0.50

(/)

< w

(/)

o

0.25

Patients at risk

0.00

(76)

(\4)

(2)

....-

+----..--.----..--.--~--

a

60

180

120

TIME

Fig. 3. Overall disease-free survival (n indicate standard errors.

=

(months)

95); vertical bars

The relative risk of manifesting recurrent disease as distant metastasis for high-risk patients was 25.64 (95% confidence limits, 8.4 to 77.6) times higher than that of low-risk patients (p < 0.00001). Discussion Although the majority of completely resected Tl NO MO NSCLC provides a normal life expectancy after operation, in a substantial minority of patients, for example, 20% to 30%, recurrent disease develops within the first 5 years after operation. Unfortunately, only one study published by the Lung Cancer Study Group has been prospectively focused on the prognostic significance of traditional parameters on cancer recurrence in Tl NO

The Journal of Thoracic and Cardiovascular Surgery

84Macchiarini et al.

July 1993

1.00 BVI-

-l

oct

s>a:

0.75

w w

0.50

::J

en

a:

Features

u,

W en oct w

en

Table III. Univariate survival and disease-free survival analysis

0.25

i5 BVI +

0.00

+--~----r--~--r-'--~-----r-­

a

60

120

180

TIME (months) Fig. 4. Comparison ofthe disease-free survival of patients who had complete resection of postsurgical TI NO MO NSCLC at low risk (n = 79) or high risk (n = 16) of death from recurrent NSCLC (log-rank test; p < 0.0000I). The risk status was conferred by the presence (high risk, +) or absence (low-risk, -) of BYI by tumor cells. Vertical bars indicate standard errors.

MO NSCLC,3 and as a consequence we were unable to learn about the absolute and relative recurrence rates of patients displaying whatever pool of prognostic factors. To answer this enigma and to increase the inherent predictive ability of traditional and potential newer prognostic parameters, we adopted an approach recently suggested by Levine and colleagues.l" Using this approach, we have identified subgroups of patients at low and high risk for recurrent NSCLC, comprising 83% and 17%, respectively, of the total number of patients. The overall survival reported in this study is in line with the very recent and largest studies reported in the surgical Iiterature.f 3.6 Moreover, the median follow-up time (8.3 years) is sufficiently mature and complete for drawing interesting conclusions. Extent of operation, tumor size, proliferative activity, mitotic count, and BYI by tumor emboli were found to predict survival in univariate analysis. Patients treated with wedge resection did significantly better than those treated by simple lobectomy or pneumonectomy, and this finding may be related to the different anatomic and biologic properties of neoplasms. Wedge resection was performed for superficially seated and peripheral tumors located within or crossing major intersegmental planes.l" Analyzing the characteristics of these tumors, we found they had a lower tumor size (1.8 versus 2.1 em), PCNA reactivity (20.9 versus 25.9%), mitotic count (9.9 versus 19.1), and prevalence of BYI (4.5 versus 21 %) compared with tumors undergoing major resection. Although this finding may confirm a

Extent of operation Wedge resection vs. lobectomyjpneumonectomy BY! (absent vs. present) PCNA reactivity* Low vs. intermediate Tumor size < vs. 2: median, 2 em Mitotic count < vs. 2: median, 13

Survival (p value)

Disease-free survival (p value)

0.0185

0.0224

0.00001

0.00001

0.00001

0.00004

0.0361

NS

0.0105

NS

NS. Not statistically significant; no other discrete and continuous variable met the 0.05 level of significance. 'Low. 0% to 25%; intermediate. 26% to 75%.

lower metastatic behavior of peripheral compared with more centrally located tumors.s 14 the failure of type of operation to hold significance in multivariate analysis outweighs very recent observations by Ginsberg and Rubinstein* contraindicating conservative operation as the treatment of choice for Tl NO MO NSCLC. It has been generally accepted that, within tumor categories, tumor diameter does not influence outcome significantly.35-37 Data from the current study are similar to our previous experience.!" however, and also are similar to recent observations made by Read and colleagues- 38 and others 39, 40 demonstrating a significant survival advantage for patients whose tumors measured less than 2 em compared with lesions measuring greater than 2 cm. Despite the fact that this is supportive of a strict correlation between tumor growth and metastatic potential," the failure of tumor size to retain prognostic significance in multivariate analysis suggests caution before introducing a different categorization of Tl NO MO lesions into TlI2 «2 em) and Tl (>2 em) stressed by several authors.v 39, 40 Two parameters assessing different compartments of the cell cycle, mitotic count and PCNA immunoreactivity, predicted survival significantly. The prognostic value of mitotic count in peripheral NSCLC has been documented by Weiss, Boucot, and Cooper'? and James and Davey':' and, more recently, by Takise and colleagues.'? We also demonstrated a significant inverse relationship between mitotic activity and survival. 14 Our data further confirm that increased mitotic activity is a peculiarity of *Ginsberg RJ, Rubinstein L. Patients with TI NO non-small cell lung cancer. 6th World Conference on Lung Cancer, Melbourne, Australia, Nov. 10-14, 1991 [Abstract No. 304].

The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 1

M acchiarini et al.

85

Table IV. Multivariate survival analysis Features

Coefficient

SD

p Value

RR-S 5-yr* (95% Cl)t

RR-S 10-yr* (95% Cl)t

RR-S 15-yr* (95% CI)t

Absent vs. present Mitotic count < vs. 2: median. 13

2.5244

0.4298

0.0001

5.34 (3.02-9.47)

1.139 (0.96-1.34)

1.026 (0.9-1.06)

0.0226

0.0089

0.016

2.85 ( 1.24-6.50)

1.1 (0.92-1.31 )

1.04 (0.99-1.11 )

(3

BYI

SD. Standard deviation; BVI. blood vessel invasion.

'Relative risk of survival calculated according to the methods of Levine ct al. 34 confidence interval; no other variable met the 0.05 significance level.

t95%

Table V. Multivariate disease-free survival analysis Features

(3 Coefficient

SD

p Value

RR-DFS 5-yr* (95% Cl)t

RR-DFS 10-yr* (95% Cl)t

RR-DFS 15-yr* (95% Cl)t

Absent vs, present

3.1853

0.785

0.00004

11.66 (5.3-25.64)

1.12 (0.95-1.29)

1.03 (0.99-1.06)

BY! SD. Standard deviation; BVI. blood vessel invasion.

'Relative risk of disease-free survival calculated according to the methods of Levine et al 3 4 t95% confidence interval; no other variable met the 0.05 significance level.

more aggressive tumors." To evaluate cell proliferation better, we investigated the expression of PCNA with a new monoclonal antibody.P This immunohistologic method allows a reliable estimation of the number of tumor cells in the DNA synthetic (S) phase of the cell cycle;' which is the most representative phase when one evaluates tumor proliferatiorr'"; moreover, PCNA expression seems to have some prognostic usefulness in peripheral, node-negative NSCLC. 7 In this study, patients whose tumor showed a low ( <25%) proliferative activity had a significantly better survival and disease-free survival than those whose tumors showed an intermediate proliferative (26% to 70%) activity, and this is in keeping with several studies demonstrating that increased proliferative rates are associated with a worsened prognosis in many of the commonest adult solid malignancies, including NSCLC. 44 It has been postulated that BYI by tumor cells represents one of the first steps in the metastatic process of a variety of neoplasms,18-20 including node-negative NSCLC. 14 There is now sufficient evidence that the metastatic process follows a cascade of two-stage events; the prevascular and the vascular phases.f The prevascular phase is characterized by the local invasion by malignant cells of host stroma within or surrounding the primary tumor; it is generally long lasting, associated with limited tumor growth, a paucity of metastasis, and is common of carcinomas of the cervix." bladder.f and some subtypes of breast cancer.f By contrast, the vascular phase begins with the penetration of tumor cells into blood or lymphatic

vessels, or both, is short lasting, and is associated with rapid tumor growth and an increased prevalence of metastasis. 18. 45 In this study the majority of the tumors examined were negative for BYI (83%) and thus in the prevascular stage; they had low proliferative activity (58%) and recurrent NSCLC rate (8.4%). By contrast, 17% of tumors were positive for BYI and therefore in the vascular phase; their more aggressive metastatic behavior is supported not only by their significantly higher number of cycling (PCNA reactivity) and mitotic cells as opposed to tumors negative for BYI, but also by the fact that 100% of patients who had tumors positive for BYI died from recurrent NSCLC. Moreover, because we did not observe lymphatic vessel invasion and there was a significant tendency of tumors positive for BYI to manifest systemic recurrence as opposed to those negative for BYI, it is realistic to speculate that Tl NO MO cells disseminate preferentially via the blood vessels. Since experimenta149. 50 and clinical'v- 51 data suggest that without angiogenesis tumor cells can rarely switch to the vascular phase, our data indirectly suggest that NSCLCs are angiogenic-dependent neoplasms.52 Taken together, these data may explain the highly predictive survival and disease-free survival ability of BYI observed in this study. Considering the previously published studies investigating the prognostic significance of DNA content in NSCLC (for a review, see Merkel and McGuire),44 the absence of correlation between DNA ploidy status and treatment outcome observed in stage I NSCLC is somewhat surprising but in line with our previous work? and

86Macchiarini et al.

recent observations of Ten Yelde 53 and Cibas'' and their colleagues. Several authors- 6,54 have noted significant differences between cell types in resected NSCLC. The fact that these findings differ considerably from those in the present study is interesting, but the reasonts) are not clear at present; one possible explanation might be that, although not significant, we observed a trend (p < 0.08) favoring squamous histology; in a larger study, squamous tumors may fare significantly better. Although the prevalence of new primary lung cancer noted in this study was 2.1 %, it is in line with the 5.2% reported by the largest series ofTl NO MO NSCLC ever published.' Looking at the recent development in the areas of monoclonal antibodies,55 these patients with NSCLC who have had resection may have a unique opportunity to benefit from early detection. By multivariate analysis, BYI and mitotic count were the only significant independent predictors of survival, and the relative risk of shorter survival for patients with both risk factors was remarkably pronounced in the first 5 years after operation; it then reached a plateau during the following years. Unfortunately, patient survival is not a sensitive estimate for Tl NO MO NSCLC. 3 We therefore looked with particular anxiety to the results of the multivariate disease-free survival analysis, where BYI turned out to be the only significant (p < 0.00004) and independent predictor of disease-free survival, particularly in the first 5-year period. The relative risk of recurrence for a patient with a tumor harboring BYI by tumor cells was 13.3 (95% confidence limits, 6.1 to 28.7) times higher that for a patient whose tumor was negative for BY!. There were 16 patients with such a high-risk factor present, and their cumulative recurrence risk was 111.7% higher than the low-risk population. While in stage I breast cancer, adjuvant chemotherapy reduces the recurrence rate by approximately 30% within 4 years after operation in node-negative tumors,56-58 the question arises whether the magnitude of the relative risk of recurrence for tumors positive for BYI warrants the morbidity of adjuvant therapy. The subject is hotly debated since there is no evidence in recent literature" that adjuvant chemotherapy has a dramatic impact on survival nor that it reduces recurrence in NSCLC; at a minimum, it has a biologic effect.i- 59 We share the opinion of Holmes and associates," however, that the absence of positive adjuvant trials in NSCLC reflects the ineffectiveness of drug regimens and the inappropriateness of patient selection. Thereby, with new available drugs it is not unrealistic to anticipate a more therapeutic index of adjuvant cytotoxic chemotherapy. If such adjuvant systemic therapy is given, however, it should be done in a properly controlled trial.

The Journal of Thoracic and Cardiovascular Surgery July 1993

In conclusion, the results presented demonstrate that within completely resected Tl NO MO NSCLC, the absence or presence of intratumoral or peritumoral BYI by neoplastic cells categorizes a particular patient into low risk or high risk, respectively. The good-prognosis group might be considered for strict surveillance without adjuvant therapy, because the vast majority of these patients are curable by operation alone. The group of patients at high risk of recurrent NSCLC have a prognosis similar to that of patients with involved mediastinal lymph nodes 6o for which adjuvant therapy seems to have a beneficial effect,4.59 Since patients positive for BYI have a significant tendency to relapse systemically in the first 5-year period, they would appear suitable for either adjuvant systemic chemotherapy or new therapeutic interventions.v' We are grateful to Marco Yalesini, MD, and Yannuzzi Ascolo for technical assistance. REFERENCES I. Shields TW. Carcinoma of the lung. In: Shields TW, ed. General thoracic surgery. 3rd ed. Philadelphia: Lea & Febiger, 1989:890-934. 2. Read RC, Yoder G, Schaeffer RC. Survival after conservative resection for TI NOMOnon-smallcell lung cancer. Ann Thorac Surg 1990;49:391-400. 3. Thomas P, Rubinstein L, and the Lung Cancer Study Group. Cancer recurrence after resection: T I NOnon-small cell lung cancer. Ann Thorac Surg 1990;49:242-7. 4. Holmes EC, BleehenNM, Le Chevalier T, et al. Postoperative adjuvant treatments for non-small cell lung cancers: a consensus report. Lung Cancer 1991 ;7:1-13. 5. Gail MH, Eagan RT, Feld R, et al. Prognostic factors in patients with resected stage I non-small cell lung cancer. Cancer 1984;54: 1802-13. 6. Thomas PA, Piantadosi S. PostoperativeTI NOnon-small cell lung cancer. J THORAC CARDIOVASC SURG 1987; 94:349-54. 7. Fontanini G, Macchiarini P, Pepe S, et al. The expression of proliferating cell nuclear antigen in paraffin sectionsof peripheral, node-negativenon-smallcell lung cancer. Cancer [In press]. 8. Cibas ES, Melamed MR, Zaman MB, Kimmel M. The effect of tumour size and tumour cell DNA content on the survivalof patients with stage I adenocarcinomaof the lung. Cancer 1989;63: 1552-6. 9. Silvestrini R, Muscolino G, Costa A, et al. Could cell kinetics be a predictor of prognosis in non-small cell lung cancer? Lung Cancer 1991;7:165-70. 10. Fontanini G, Pingitore R, Bigini D, et al. Growth fraction estimation in non-small cell lung cancer determined by proliferating cell nuclear antigen and comparison with the Ki-67 labeling and flow cytometry data. Am J Pathol [In press].

The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 1

II. Maki E, Feld R. Prognostic factors in patients with non-small cell lung cancer. A critique of the world literature. Lung Cancer 1991;7:27-34. 12. Goldman L, Caldera DL, Nussbaum SR, et al. Multifactorial risk of cardiac risk in noncardiac surgical procedures. N Engl J Med 1977;297:845-50. 13. Macchiarini P, Janni A, Silvano G, Mussi A, Chella A, Angeletti CA. Results of treatments and lessons learned from pathologically staged T4 non-small cell lung cancer. J Surg Oncol 1991;47:209-14. 14. Macchiarini P, Fontanini G, Hardin JM, Pingitore R, Angeletti CA. Most peripheral, node-negative non-smallcell lung cancers have low proliferative rates and no intratumoral and peritumoral blood and lymphatic vessel invasion: rationale for treatment with wedge resection alone. J THORAC CARDIOV ASC SURG 1992;I04:892-9. 15. Mountain CF. A new international staging system for lung cancer. Chest 1986;89:225S-33S. 16. World Health Organization. The World Health Organization.Histological typing of lung tumors. Am J Clinic Pathol 1982;77: 123-36. 17. Hsu SM, Raine L. The use of avidin-biotin peroxidase complex (ABC) in diagnostic and research pathology. In: DeLessis RA, ed. Advances in immunohistochemistry. New York: Masson, 1984:31-42. 18. LeeAKC, DeLellis RA,Silverman ML, HeatleyGl, Wolfe HJ. Prognostic significance of peritumoral lymphatic and blood vessel invasion in node-negative carcinoma of the breast. J Clin Oncol 1990;8:1457-65. 19. Rosen PP. Tumor emboli in intramammary lymphatics in breast cancer: pathological criteria for diagnosis and clinical significance. Pathol Annu 1983;18:215-32(part 2). 20. Weigand RA, Isenberg WM, Russo J, et al. Blood vessel invasion and axillary lymph node involvement as prognostic indicators for human breast cancer. Cancer 1982; 50:962-9. 21. Bravo R, Celis JE. A search for differential polypeptide synthesis throughout the cell cycle of HeLa cells. J Cell BioI 1980;84:795-802. 22. Bravo R, Frank R, Blundell PA, MacDonald-Bravo H. CyclinjPCNA is the auxiliary protein of DNA polymerase alpha. Nature 1987;326:515-7. 23. Kurki P, Ogata K, Tan EM. Monoclonal antibodies to proliferating cellnuclear antigen (PCN A) j cyclin as probes for proliferating cells by immunofluorescence microscopy andflow cytometry.J Immunol Methods 1988;I09:49-59. 24. HallPA, Levison DA, Woods AL, et al. Proliferating cell nuclear antigen (PCNA) immunolocalization in paraffin sections: an index of cell proliferation with evidence of deregulated expression in some neoplasms. J Pathol 1990; 162:285-94. 25. Waseem NH, Lane DP. Monoclonal antibody analysis of the proliferating cell nuclear antigen (PCNA). Structural conservation and the detection of a nucleolar form. J Cell Sci 1990;96:121-9. 26. Quinn eM, Wright NA. The clinical assessment of proliferation and growth in human tumors: evaluation of meth-

Macchiarini et al. 8 7

27. 28.

29. 30. 31. 32.

33. 34.

35.

36.

37.

38.

39.

40.

41. 42.

43. 44. 45.

46.

47.

ods and applications as prognostic variables. J Pathol 1990;160:93-102. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457-81. Peto R, Pike MC, Armitage P, Breslow NE, Cox DR, Howard SY. Design and analysis of randomized clinical trials requiring prolonged observations of each patient: II. Analysis and examples. Br J Cancer 1977;35:1-39. Cox DR. Regression models in life tables. J R Stat Assoc [B] 1972;34:187-220. Kalbfleish JD, Prentice RL. The statistical analysis of failure time data. New York: Wiley, 1980. Kahn HA, Sempos CT. Statistical methods in epidemiology. New York: Oxford University, 1989:45-71. Sackett DL, Haynes RB, Tugwell P. Clinical epidemiology: a basic science for clinical medicine. Boston: Little, Brown, 1985:47-138. Rosner B. Fundamentals in biostatistics. 3rd ed. Boston: PWS-Kent Publisher, 1990:42-70. Levine MN, Browman GP, Roberts R, Goodyear M. When is a prognostic factor useful? A guide for the perplexed. J Clin Oncol 1991;9:348-56. Hoffmann TH, Ransdell HT. Comparison of lobectomy and wedge resection for carcinoma of the lung. J THORAC CARDIOVASC SURG 1980;79:211-7. Williams DE, Pairolero PC, Davis CS, et al. Survival of patients surgically treated for stage I lung cancer. J THoRAC CARDIOVASC SURG 1981;82:70-6. Feld R, Rubenstein LY, Weisenberger TH. Sites of recurrence in resected stage l-non-small cell lung cancer: a guide for future studies. J Clin Oncol 1984;2:1352-8. Read RC, Schaefer R, Walls R. Diameter, cell type and survival in stage I primary non-small cell lung cancer. Arch Surg 1988;123:446-9. Takise A, Kodama T, Shimosato Y, Watanabe S, Suemasu K. Histologic prognostic factors in adenocarcinomas of the peripheral lung less than 2 em in diameter. Cancer 1988;61:2083-8. Roeslin N, Chalkiadaakis G, Dumont P, Witz JP. A better prognostic value from a modification oflung cancer staging. J THoRAc CARDIOVASC SURG 1987;94:504-9. Folkman J. What is the evidence that tumours are angiogenesis dependent? J Natl Cancer Inst 1990;82:2-4. Weiss W, Boucot KR, Cooper DA. Survival of men with peripheral lung cancer in relation to histologic characteristics and growth rate. Am Rev Respir Dis 1968;98:75-86. James RD, Davey AT. Mitotic index in carcinoma of the lung. Lancet 1979;20:852. Merkel DE, McGuire WL. Ploidy, proliferative activity and prognosis. Cancer 1990;65: 1194-1205. Weidner N, SempleJP, William RW, FolkmanJ. Tumor angiogenesis and metastasis-correlation in invasive breast carcinoma. N Engl J Med 1991;324:1-8. Sillman F, Boyce J, Fruchter R. The significance of atypical vessels and neovascularization in cervical neoplasia. Am J Obstet Gynecol 1981;139:154-159. Chodak GW, Haudenschild C, Gittes RF, Folkman J.

The Journal of Thoracic and

88Macchiarini et at.

48.

49. 50.

51.

52.

53.

54.

55.

56.

Angiogenic activity as a marker of neoplastic and of preneoplastic lesions of the human bladder. Ann Surg 1980;192:762-71. Jensen HM, Chen I, deVault MR, Lewis AR. Angiogenesis induced by "normal" human breast tissue: a probable marker for precancer. Science 1982;218:293-5. Fidler IJ, Gersten OM, Hart IR. The biology of cancer invasion and metastasis. Adv Cancer Res 1978;28:149-250. Liotta LA, Saidel G, Kleinerman J. The significance of hematogenous tumour cell clumps in the metastatic process. Cancer Res 1976;36:889-94. Srivastava A, Laidler P, Davies RP, Horgan K, Hughes LE. The prognostic significance of tumour vascularity in intermediate-thickness (0.76-4.0 mm thick) skin melanoma: a quantitative histologic study. Am J Pathol 1988; 133:419-23. Liotta L, Kleinerman J, Saidel G. Quantitative relationships of intravascular tumour cells, tumour vessels, and pulmonary metastases following tumour implantation. Cancer Res 1974;34:997-1004. Ten Velde GPM, Schutte B, Vermeulen A, Volovics A, Reynders MMJ, Blijham GH. Flow cytometric analysis of DNA ploidy level in paraffin-embedded tissue of non-small cell lung cancer. Eur J Cancer Clin Oncol 1988;24:455-60. Mountain CF, Lukeman JM, Hammer SP, et al. Lung cancer classification: the relationship of disease extent and cell type to survival in a clinical trials population. J Surg Oncol 1987;35:147-56. Tockman MS, Gupta PK, Myers JD, et al. Sensitive and specific monoclonal recognition of human lung cancer antigen on preserved sputum cells: a new approach to early lung cancer detection. J Clin Oncol 1988;6:1685-93. Fisher B, Redmond C, Dimitrov NV, et al. A randomized

Cardiovascular Surgery July 1993

57.

58.

59.

60.

61.

clinical trial evaluating sequential methotrexate and fluorouracil in the treatment of patients with node-negative breast cancer who have estrogen receptor negative tumours. N Engl J Med 1989;320:473-8. Mansour EG, Gray R, Shatila AH, et al. Efficacy of adjuvant chemotherapy in high-risk node-negative breast cancer. N Engl J Med 1989;320:485-90. The Ludwig Breast Cancer Study Group. Prolonged disease-free survival after one course of perioperative adjuvant chemotherapy for node-negative breast cancer. N Engl J Med 1989;320:491-6. Macchiarini P, Hardin M, Angeletti CA. Long-term evaluation of intrapleural bacillus Calmette-Guerin with or without adjuvant chemotherapy in completely resected stages II and III non-small cell lung cancer. Am J Clin OncoI1991;14:291-7. Shields TW. The significance of ipsilateral mediastinal lymph node metastasis (N2 disease) in non-small cell carcinoma of the lung. J THORAC CARDIOVASC SURG 1990; 99:48-53. Denekamp J. Vascular attack as a therapeutic strategy for cancer. Cancer Metastasis Rev 1990;9:267-82.

Appendix To test whether there was a relationship between tumor size and growth fraction (PCNA imunoreactivity and mitotic count), we further analyzed interrelations between these two parameters. The size of the tumor correlated significantly with PCNA reactivity and mitotic count, and there was a significant correlation between PCNA immunoreactivity and mitotic count (Appendix Fig. I). These data denote a strict parallelism between tumor size and growth and might provide evidence that larger NSCLCs display a higher fraction of proliferating tumor cells that ultimately enhance their metastatic behavior.