- Email: [email protected]
Carcinoma of the lung
J THORAC CARDIOVASC SURG 1989;97:504-12
Carcinoma of the lung Evaluation of satellite nodules as a factor influencing prognosis after resection Satellite nodules are considered to be predictive of poor prognosis in breast cancer and in melanoma. In lung cancer, there is no information as to tbeir definition, prevalence, or implication as a prognosis factor of survival after resection. Over the past 18 years (1969 to 1987), 84 patients underwent pulmonary resection for primary lung cancer accompanied by satellite nodules. Tbese nodules were defined as well-circumscribed accessory carcinoma foci clearly separated from the main tumor but with identical histologic characteristics. All were smaller than the primary carcinoma and most were located within the same lobe. Survival rates of patients with satellite nodules were compared to those of 1021 patients without satellite nodules who underwent resection during the same time interval. The 1-,3-, and 5-year survival rates for all patients classified as having no satellite nodules were 78 %, 54 %, and 44 %, respectively, and the median survivalfor the entire group was 30 months. In patients with satellite nodules, these survival rates were 60.9%, 32.7%, and 21.6%, respectively, with a median survival of 15 months. The deleterious effect of satellite nodules was more significant in patients with stage I disease (p = 0.0008) than in patients with stage II (p = 0.0354) or stage rn (p = 0.0145) disease. Survival data obtained by comparison of satellite nodule status and histologic characteristics shows that 5-year survival figures are better for patients with no staellite nodules in both the squamoll'i and the nonsquamous groups. This study demonstrates that satellite nodules associated with lung cancer are indicative of locally advanced and/or premetastatic disease. Tbese patients should be included in the stage group rna of the TNM stage grouping classification.
Jean Deslauriers, MD," Jacques Brisson, Moe (by invitation), Raymond Cartier, MD" (by invitation), Marcien Fournier, MOb (by invitation), Denis Gagnon, BSc d (by invitation), Michel Piraux, MD" (by invitation), and Maurice Beaulieu, MD" (by invitation), Ste-Foy, Quebec, Canada
Satellite nodules are considered to be predictive of poor prognosis in malignant disease of the breast and in melanomas. In breast cancer' satellite skin nodules clearly indicate important local tumor spread and, according to the clinical staging nomenclature (TNM system), satellitosis confined to the same breast is a T4 descriptor (stage IV). The presence of satellite nodules
From the Division of Thoracic Surgery, Centre de Pneumologie,' and the Department of Pathology,' Hopital Laval; and the Departments of Epidemiology' and Statistics," Laval University, Ste-Foy, Quebec, Canada. Read at the Sixty-eighth Annual Meeting of The American Association for Thoracic Surgery, Los Angeles, Calif., April 18-20, 1988. Address for reprints: Dr. Jean Deslauriers, 2725, chemin Ste-Foy, Ste-Foy, Quebec G 1V 4G5, Canada.
504
within 2 em of any primary melanoma' is also considered to be a T4 descriptor. In lung cancer there is no specific information as to the definition of satellite pulmonary nodules, their prevalence, or even their implication as a prognostic factor of survival after resection. In addition, there is no provision for their coding in the new definitions for the TNM categories of primary tumors (T status).' Over the past 18 years, we have identified 84 patients having resection of a primary bronchogenic carcinoma in which the primary lesion was accompanied by malignant satellite nodules. This report presents a comparative survival analysis between those individuals and 1021 patients without satellite nodules who underwent resection during the same time interval. The results indicate that the presence of satellite nodules has a negative impact on survival and a proposal is made for
Volume 97 Number 4 April 1989
Carcinoma of lung
50 5
Fig. 1. Right lung with lower lobe containing squamou s carcinoma, pathologic stage T2 N2 , and one satellite nodule (arrow) central to the main tumor.
an addition to the currently accepted TNM nomenclature. Patients and methods Selection of patients: pathologic terminology. Several morphologic and histologic criteria were used to differentiate a single tumor with satellite nodules from independent synchronous carcinomas or from pulmonary metastasis arising from a tumor in another organ system . Satellite nodules were defined as well-circumscribed accessory carcinoma foci adjacent to but clearly separated from the main tumor by normal lung parenchyma (Fig. I). All were smaller than the primary carcinoma and most were located within the same lobe. On histologic examination, they were described as parenchymatous tumor deposits with the same histologic features as the primary carcinoma. Patients whose satellite lesion proved to be a lymphatic or a lymph node metasta sis were not included in this study. Also excluded were patients whose pulmonary lesions were by histologic examination highly indicative of being metastatic from an unknown primary site. All histologic material included in the series was initially
assessed by the consultant pathologist. This material was subsequently reviewed by one of the authors (M.F.) to ascertain the presence of nodules, revise diagnostic criteria, and determine the histologic grade of both primary tumor and satellite nodules. Information on patients designated as having definite or probable synchronous lung cancers was previously reported " and is not included in this report . These patients fulfilled the following two criteria: (I) Their tumors were physically separate and in some instances located in different lobes; (2) their tumors were of different histologic types or were shown to arise from different endobrochial lesions by bronchoscopic study or from separate foci of carcinoma in situ at pathologic examination. Differentiating satellite nodules from pulmonary metastases was done by recording of past a nd present medical history, investigation of other organ systems as possible sources of malignancy, anal ysis of pathologic characteristics of the neoplasm, careful observation during the follow-up period, and postmortem examination. Four patients in this series had malginant tumors of other organ systems considered to be unrelated to the lung tumor. Two individuals had laryngeal carcinomas 6 and 10 years before their lung cancer; neither
The Journal of Thoracic and Cardiovascular
5 0 6 Deslauriers et al.
Surgery
Table I. General characteristics of 1105 patients included in the comparison of survival by satellite nodule status (1969 to 1987)
Table II. Pathology and stage of 1105 patients included in the comparison of survival by satellite nodule status (1969 to 1987)
Satellite nodules Characteristics
Age (yr) Mean (±SD) Range Sex Male Female Staging mediastinoscopy Yes No Type of operation Limited resection Lobectomy Pneumonectomy Complete resection Yes No
Present (n = 84)
Absent (n= /02/)
58.3 (± 9.65) 37-76
60.3 (± 8.53) 28-83
68 (81%) 16 (19%)
884 (86.6%) 137 (13.4%)
79 (94%) 5 ( 6%)
918 (89.9%) 103 (10.1 %)
3 ( 3.6%) 35 (41.7%) 46 (54.8%)
26 (2.5%) 587 (57.5%) 408 (40.0%)
80 (95.2%) 4 (4.8%)
1021 (100%) 0
had clinical evidence of local recurrence at the time of pulmonary resection. A third patient had a sarcoma of the lower extremity removed 3 months before being operated on for lung cancer, and the fourth underwent resection of a carcinoma in situ of the cecum 8 years after pulmonary resection. General characteristics of the study population. Between January 1969 and December 1986, 1105 patients underwent elective pulmonary resection as primary treatment for bronchogenic carcinoma. Eighty-four individuals had a carcinoma accompanied by one or more satellite nodules as defined earlier, and the other 1021 patients had no satellite nodules: Age and sex distributions were relatively similar in the two groups (Table I). Mediastinoscopy, with random node sampling at three different levels, was done in almost every case for pretreatment staging of the superior mediastinum. Only 20 patients (1.8%) were operated on when nodal metastasis was identified by mediastinoscopy. Limited resections (segmental/wedge) were performed when the patient's pulmonary reserve was thought to be compromised, but these were infrequently done in either group. Patients undergoing lobectomy include all individuals with simple lobectomies, bilobectornies, sleeve lobectomies, or extended lobectomies. Similarly, all patients with standard, intrapericardial, or extended pneumonectomies are included in the pneumonectomy group. Pneumonectomy was done more often in patients with satellite nodules (54.5%), whereas lobectomy was more frequently done in patients without satellite nodules (57.5%). On the basis of removal of all gross carcinoma, disease-free resection margin, and highest mediastinal node free of tumor, all procedures were considered to be complete in the no satellite nodule group. Resection was incomplete in four patients (4.8%) of the satellite nodule group because the final pathology report identified microscopic residual cancer at the
Satellite nodules Characteristics
Histologic type Squamous cell carcinoma Adenocarcinoma Bronchoalveolar carcinoma Large cell carcinoma Small cell carcinoma Mixed tumors Stage Stage I TI NO T2 NO T1 NI Stage II T2 NI Stage III T3 NO T3 NI T3 N2 T2 N2 T3 N2
Present (n = 84)
Absent (n = /02/)
35 (41.7%)
589 (57.7%)
21 (25.0%) 4 (4.8%)
171 (16.7%) 41 (4.0%)
20 (23.8%)
158 (15.5%)
I (1.2%)
22 (2.2%)
3 (3.6%)
40 (3.9%)
43 (51.2%) 12
566 (55.4%) 216 325 25 236 (23.1%)
27 4
16 (19.0%) 25 (29.8%) 4
219 (21.5%)
11
84 8 42 13
I 6 3
72
bronchial resection margin (n = 3) or in the highest mediastinal node (n = I). Histopathology and stage of disease. Pathologic diagnosis was obtained by the study of surgical specimens in all patients. The majority of resected neoplasms were squamous carcinomas, but their prevalence was proportionally higher in the no satellite nodule group (57.7%) than in the satellite nodule group (41.7%) (Table II). In contrast, adenocarcinomas and large cell carcinomas were more common in the satellite nodule group. Few patients in either group had bronchoalveolar cell carcinoma or small cell carcinoma. In the survival analysis, the squamous group comprised those having pure squamous disease, whereas the nonsquamous group contained all other patients including some with a mixture of cell types. All tumors were pathologically staged in accordance with the TNM terminology as outlined by the American Committee on Cancer Staging.' For the first 9 years of the study (1969 to 1978), the staging was done retrospectively with bronchoscopy diagrams, chest films, and operative and pathology reports. For the remaining years (1978 to 1987), staging was based on complete clinical and pathologic analysis of the primary tumor (T status) and surgical sampling in at least the bronchopulmonary, hilar, and mediastinal nodal areas (N status). For any given patient, the presence of satellite nodules was not taken into consideration when formulating a final stage of disease. Stage distribution for patients included in this
Volume 97 Number 4
Carcinoma of lung
April 1989
Table m. Characteristics of the main tumor and satellite nodules in 84 patients with satellite nodules Primary tumor Location Right upper lobe Right middle lobe Right lower lobe Left upper lobe Left lower lobe Lung zone Central Mid-lung Peripheral Size 0-3 ern
3.1-6 em >6cm Histologic grade Well differentiated (grade I) Moderately well differentiated (grade II) Poorly differentiated and anaplastic (grade III) Satellite nodules No. of nodules I
90 80
70
No. of patients (%)
Characteristics
0/0
26 (31%) 3 17 24 (28.6%) 14
507
60 50 40
,~54.8%
30 20 10
~45.2%
YEAR 28 28 28 21 3 60 (71.4%)
16 26 42 (50%)
53 (63.1%)
2 >3
4 19 Unknown 8 Degree of differentiation versus primary tumor 8 Better differentiation Same differentiation 61 (72.6%) Worse differentiation 12 Unknown 3 Seen on roentgenogram 13 (15.5%) Yes 71 (84.5%) No
series shows that it is nearly identical in the two groups (Table II). Tumor characteristics in patients with satellite nodules. In patients with satellite nodules, the primary carcinomas were predominantly located in the upper lobes but evenly distributed among the various lung zones (Table III). Most tumors were large (>6 em) and 50% had a poor histologic grade. On the basis of their anatomic location, four subsets of nodules were identified: (1) nodules peripheral to the main tumor (n = 38), (2) nodules central to the main tumor (n = 12), (3) nodules around the main tumor (n = 18), and (4) nodules in a different lobe (n = 16). The majority of patients (63%) had a single satellite nodule and most were of the same histologic grade as the primary carcinoma. All preoperative chest films were reviewed, and in only 15.5% of patients could the satellite nodule be identified. Follow-up. Patients were followed up from the date of operation to the time of death, loss to follow-up, or end of observation (December 1987), whichever came first. Only five
# At risk ... •
1021
791
574
429
335
251
84
49
27
23
15
11
k---Jt.
NO SATELLITE NODULE
.........
SATELLITE NODULE
Fig. 2. Life-table analysis showing the percentage of all patients remaining alive after resection of bronchogenic carcinoma with or without satellite nodules.
patients (0.5%) were lost to follow-up. The potential duration of follow-up was at least 12 months for all patients under review. The mean follow-up times were 28 months and 43 months, respectively, in the satellite nodule group and in patients without satellite nodules. Patients alive at the end of the follow-up period were classified as being (1) disease free or having (2) evidence of local recurrence, (3) second primary tumors, or (4) systemic metastasis. A local recurrence was defined as tumor growth within the ipsilateral hemithorax or mediastinum, or both. For patients who died, the cause of death was determined from clinical records (73% of patients were followed up at the hospital), death certificates, or from information provided by family physicians and by relatives of the patient. Each death was classified as being due to (I) lung cancer, including second primary lesions and operative deaths, (2) causes unrelated to lung cance, or (3) unknown causes. Operative mortality was defined as a death occurring within 30 days of the operation or a death directly related to the procedure even if it occurred more than 30 days postoperatively. Minor complications were defined as events that had little impact on the postoperative course or duration of hospital stay, and major complications were nonfatal but sometimes lifethreatening events. If a patient had both minor and major complications, he was included in the major complications group only. Statistical analysis. Results are presented as means (± standard deviation) for continuous variables and as percentages for categorical data. The analysis of survival was based on deaths from all causes. Survival was analyzed according to the Kaplan-Meier method and differences between survival curves of patients with and without satellite nodules were compared by the log-rank test. Multivariate analysis of failure times made up use of Cox proportional hazard model. In this analysis, age was included as a continuous variable and other factors were entered as one or two dichotomous indicator variables: presence of nodules (absent, present), sex (male, female), stage (I, II, III), histology (squamous, nonsquamous),
The Journal of
5 0 8 Deslauriers et al.
0/0
Thoracic and Cardiovascular Surgery
100
100
80
80
60
0/0
40
60 40 2')
20 2
2
YEAR
YEAR
3
YEAR
# AI risk
..
5C6
•
483371
43
20
1 CJ
282
2~J
1"7
~72
236
11
179125
16
_
k----IL NO SATELLITE NODULE
97
78
57
10
21912978 25
10
50
39
23
4
SATELLITE NODULE
Fig. 3. Life-table analysis by stage showing the percentage of all patients remaining alive after resection of bronchogenic carcinoma with or without satellite nodules.
SQUAMOUS
100
p
80
0/0
NON SQUAMOUS
100
= 0.0001
80
60
%
40 20
60 40 20
a
4
3
2
a
5
2
YEAR
3
5
YEAR
# At risk
A •
589 478 35
A---A
17
366
270
219
166
432
313
208
159
116
87
8
8
7
7
49
32
17
14
8
6
NO SATELLITE NODULE
. . . - SATELLITE NODULE
Fig. 4. Life-table analysis by histologic type showing the percentage of all patients remaining alive after resection of bronchogenic carcinoma with or without satellite nodules.
type of operation (lobectomy and limited resections, pneumonectomy), and complications of operation (none, minor, and major including operative deaths). The crude effect of each of these factors was assessed by entering only the variable of interest in the model. Then, after all other variables were taken into account, the effect of each factor was estimated on the basis of a model that included all factors simultaneously. Because the effect of satellite nodules was slightly affected by this adjustment, only adjusted results are presented. These results are described in terms of the coefficient of each indicator variable with its standard error, hazard ratios with their 95% confidence interval, and the p values. Hazard ratios for categorical variables can be interpreted as the ratio of the rate of mortality in one category of patients to the rate in the reference group. The p values were obtained from the X computed as the ratio of the coefficient to its standard error.
Results Operative morbidity/mortality. There was a close similarity of results between the two groups in the
3Q-day operative morbidity and mortality. Approximately two thirds of patients in either group had an uneventful postoperative period. Thirty-six patients died (3.5% operative mortality rate) in the no satellite nodule group and four in the satellite nodule group (4.8%). Most deaths occurred in patients who had had a pneumonectomy and were due to respiratory failure, pleuropulmonary infections, or bronchopleural fistulas. Comparison of overall survival and of survival by stage and histologic type. The observed survival rates for the entire group and each subset of the TNM classification are shown in Figs. 2 and 3. The 1-, 3-, and 5-year survival rates for all patients classified as having no satellite nodules were 78%, 54.3%, and 43.7%, respectively. For those patients with stage I disease, the 5-year survival rate was 54.4%. It was 40.4% for patients with stage II disease and 20.3% for those with stage III disease. Median survival for the entire group was 30 months. In patients with satellite nodules, the 1-, 3-, and 5-year survival rates for all patients were 60.9%,32.7%, and 21.6%, respectively. For those patients with stage I disease, the survival rate at 5 years was 32%, and it was 12.5% for patients with stage II disease and 5.6% for patients with stage III disease. Median survival was 15 months for all patients included in this group. The deleterious effect of satellite nodules on survival is observable in all patients involved in the comparison analysis but is more significant in patients with stage I disease (p = 0.6008) than in those individuals with stage II (p = 0.0354) or stage III (p = 0.0145) disease. Survival data obtained by comparison of satellite nodule status and histologic type (Fig. 4) shows that 5-year survival figures are better for patients with no satellite nodules in both the squamous (p = 0.0001) and the nonsquamous and mixed histology groups (p = 0.0475).
Volume 97 Number 4
Carcinoma of lung
April 1989
509
Table IV. Cox regression analysis of survival Factor Nodules Absent* Present Stage
It II III
Coefficient (5£)*
Hazard ratio
95% CI
0.4591 (0.1401)
1.0 1.6
1.2-2.1
0.00096
0.3200 (0.1198) 0.8768 (0.1141)
1.0 1.4 2.4
1.1-1.7 1.9-3.0
0.0076 <0.00005
0.5289 (0.0890)
1.0 1.7
1.4-2.0
<0.00005
0.5070 (0.1010)
1.0 1.7
1.4-2.0
<0.00005
0.1106 (0.1178) 0.4705 (0.1074)
1.0 1.1 1.6
0.9-1.4 1.3-2.0
0.35 <0.00006
p Value
Histologic type
Squarnoust Nonsquamous Operation Lobectomy and limited resectiont Pneumonectomy Complications of operation
Nonet Minor Major
CI. Confidence intervals. 'The effect of each factor is adjusted for age at diagnosis, sex and all other factors shown in the table. tReference group.
Multivariate analysis. In the crude analysis, the hazard ratio (95% confidence interval) for patients with satellite nodules was 1.9 (1.5 to 2.5) compared to the ratio for patients without nodules. After adjustment for age at diagnosis, sex, stage of disease, histologic type of the primary tumor, and type and complications of operation, the deleterious effect of satellite nodules on survival was still clear and statistically significant (Table IV). After such adjustment, the presence of nodules was associated with a 1.6-fold increase in the rate of mortality (hazard ratio: 1.6; 95% confidence interval: 1.2 to 2.1; p values: 0.00096). Site of first recurrence and cause of death. From the beginning of the study, 579 patients died in the no satellite nodule group and 61 in the satellite nodule group. The analysis of the probable cause of death in those patients (Fig. 5) shows that a higher proportion of individuals died of cancer recurrence in the satellite nodule group (87%) than in the no satellite nodule group (63%). The distribution of sites of first recurrences (Table V) is not statistically significantly different in the respective groups of patients, although the recurrence rate is much higher in patients with satellite nodules (71%). Approximately 50% of patients with satellite nodules and who were alive at last follow-up had evidence of recurrent disease. Discussion The problem of defining two or more malignant masses appearing simultaneously in the same lung is
difficult. A number of factors must be considered in diagnosing these lesions as independent primary tumors, metastases from a nonpulmonary neoplasm, multicentric carcinomas, or one being a central primary tumor accompanied by satellite nodules. Distinguishing among those entities not only is of academic interest but also has serious implications in planning treatment and evaluating chances of prolonged disease-free survival. The anatomic location of each lesion and its histologic appearance may provide absolute criteria. Martini and Melamed" noted that when the two lesions are physically distinct and have different histopathologic appearances, they represent dual primary malignant tumors. In patients with tumors of similar histologic appearances, it is necessary that they arise from separate foci of carcinoma in situ, that there is no lymphatic involvement, and that no extra pulmonary metastases are present at the time of diagnosis. Survival after adequate resection of synchronous lung cancers is similar to that of patients with isolated lesions.v" Determining whether two or more pulmonary masses are metastatic from an unknown primary site is even more difficult. Payne, Clagett, and Harrison' described factors that favor metastatic disease in the lung: (1) highly suspected primary tumor elsewhere in the body, (2) absence of tissue communication between tumor and bronchus, (3) absence of metastatic involvement of hilar or medistinal nodes, and (4) a pulmonary lesion, which has by histologic, electron microscopic, and immunocytochemistry techniques a nonbronchogenic cell type. Median survival for patients with lung metastases from
The Journal of
5 10
Thoracic and Cardiovascular Surgery
Deslauriers et al.
o o c
8.2°4
•
CANC ER SECOND PRIMARY OTHERS UNKNOWN
1.9%
26.1°4
8.6%
SATELLITE NODULES
NO SATELLITE NODULES
Fig. 5. Cause of death in 640 patients who died and are included in the comparison by satellite nodule status.
Table V. Site offirst recurrence in 526 patients with tumor recurrence included in the comparison by satellite nodule status Satellite nodules Present First recurrences No. of patients with first recurrences Site of first recurrence Local Distant Local and distant Second primary Unknown
(n = 84)
Absent (n
= 1021)
60 (71.4%)
466 (45.6%)
16 (26.7%) 29 (48.3%) 5
131 (28.1%) 214 (45.9%) 26
4 6
70 25
an occult primary source is in the vicinity of 3 to 4 months" and is unchanged by any treatment modality. Satellite nodules have not been previously described, although multinodularity is known to be a growth pattern of bronchoalveolar cell carcinomas. This pattern is, however, rarely seen in patients with resectable disease, who generally have a single solitary nodule," Multinodularity is most often associated with poorly differentiated and unresectable tumors, as it nearly always indicates an advanced stage of disease with a grave prognosis.'?" Most satellite nodules are diagnosed once the surgical specimen is sliced and carefully examined. In contrast to the multinodularity of bronchoalveolar cell carcinomas, satellite nodules are clearly associated with a larger central tumor. They are histologically identical to the main tumor and most are located within the same lobe as the central mass.
The cause of malignant satellite nodules is not well established, but they should be considered as locally advanced disease rather than metastases, which are defined as the transfer of cancer from one organ to another not connected with it. 13, 14 The metastatic process starts, however, by local infiltration of host tissues followed by penetration of blood and lymph vessels with subsequent dissemination. IS Satellite nodules are probably, in some cases, the end result of peripheral emboli after tumor invasion of the pulmonary artery. The fact that in this series 63% (53/84) of all nodules were located peripherally or around the main tumor, usually within the same pulmonary artery distribution as the central mass, gives credence to the theory of pulmonary artery invasion. The association of bronchogenic carcinoma with centrally located satellite nodules is less common (14.3%) and probably reflects lymphatic or more likely pulmonary vein embolization. Blood vessel invasion has indeed been long recognized l 6• 17 as a poor prognostic factor for survival in resectable lung cancer. Collier and associates" even suggested that blood vessel invasion was perhaps more significant in terms of prognosis than lymph node metastases. Similarly, the finding of pulmonary vein invasion was included in Salzer's original classification" of carcinoma of the bronchus. Hinson and Nohl" showed that some degree of pulmonary vein involvement was present in every patient with lung cancer, but that infiltration of the larger veins exerted a negative influence on the chance of survival." In contrast, Shields" noted that blood vessel invasion, when identified by routine histologic examination, had little impact on long-term prognosis. Nodules located in a different lobe probably represent blood-borne metastasis, although it is possible that they
Volume 97 Number 4
Carcinoma of lung
April 1989
are the end result of air-borne metastasis as described in multicentric bronchoalveolar cell carcinomas. Preoperative detection of satellite nodules accompanying lung cancer is rare, although it is likely that the frequency of their preoperative discovery will rise with the increasing use of computed tomographic scanning in the staging of lung cancer. Their finding preoperatively or during resection should not act as a deterrent to curative operation even if they are located in a different lobe. Their presence should, however, be taken into consideration when staging the disease, estimating the patient's chances of long-term survival, and/or entering subjects in surgical adjuvant clinical trials. The survival data presented in this study clearly indicate that satellite nodules associated with lung cancer act as independent indicators of locally advanced or premetastatic disease. These data also indicate that these tumors have a biology similar to that of resectable stage III lesions (Figs. 2 and 3). Most patients included in this category have extra pulmonary extension of their tumor (T3) but are still candidates for complete resection. In the recently modified TNM stage grouping,' patients with the T3 descriptor and without N3 disease or Ml disease are classified as having stage IlIa disease. As reported by Mountain,' 13% of patients in this stage group are expected to survive 5 years, a figure that is comparable to that of patients with satellite nodules undergoing resection (21%), as shown in this study. On the basis of this information, we recommend that the T3 descriptor of the currently accepted TNM staging system be modified to read as follows: T3-A tumor of any size with direct extension into the chest wall (including superior sulcus tumors), diaphragm, or the mediastinal pleura or pericardium without involving the heart, great vessels, trachea, esophagus, or vertebral body, or a tumor in the main bronchus within 2 em of the carina without involving the carina, or a tumor accompanied by satellite nodules. As a result, patients with resectable bronchogenic carcinoma and satellite nodules will be included in the stage group IlIa.
1.
2.
3. 4.
REFERENCES Harris HR, Hellman S, Camellos GP, Fisher B. Cancer ofthe breast. In: Devita VT Jr, Hillman S, Rosenberg SA, eds. Cancer, principles and practice of oncology. Philadelphia: JB Lippincott, 1985:1114-77. Mastrangelo MJ, Baker AR, Katz HR. Cutaneous melanoma. In: Devita VT Jr, Hillman S, Rosenberg SA, eds. Cancer, principles and practice of oncology. Philadelphia, JB Lippincott, 1985:1371-1422. Mountain CF. A new international staging system for lung cancer. Chest 1986;89(suppl):225s-33s. Ferguson MK, Demeester TR, Deslauriers J, Little AG,
5. 6. 7.
8. 9.
10. 11. 12. 13.
14. 15. 16. 17.
18.
19.
20.
21.
22.
511
Piraux M, Golomb H. Diagnosis and management of synchronous lung cancers. J THoRAc CARDIOVASC SURG 1985;89:378-85. Carr DT. Staging of lung cancer. Am Rev Respir Dis 1978;117:819-23. Martini N, Melamed MR. Multiple primary lung cancers. J THORAC CARDIOVASC SURG 1975;70:606-12. Payne WS, Clagett OT, Harrison EG. Surgical management of bilateral malignant lesions of the lung. J THoRAc CARDIOVASC SURG 1962;43:279-90. Holmes FF, Fouts TL. Metastatic cancer of unknown primary site. Cancer 1979;26:816-20. Creco RJ, Steiner RM, Goldman S, Cotler H, Patchefsky A, Cohn HE. Bronchoalveolar cell carcinoma of the lung. Ann Thorac Surg 1986;41:652-6. Donaldson JC, Kaminsky DB, Elliott nc. Bronchiolar carcinoma. Cancer 1978;41:250-8. Marc Q, Galy P. Bronchiolo-alveolar carcinoma. Am Rev Respir Dis 1973;107:621-9. Tao LC, Delarue NC, Saunders D, Weisbrod G. Bronchiolo-alveolar carcinoma. Cancer 1978;42:2759-67. Fidler 11, Hart IR. Principles of cancer biology: cancer metastases. In: Devita VT Jr, Hillman S, Rosenberg SA, eds. Cancer, principles and practice of oncology. Philadelphia: 18 Lippincott, 1985:113-24. Fisher ER, Fisher B. Recent observations on concepts of metastasis. Arch Pathol 1967;83:321-4. Poste G, Fidler 11. The pathogenesis of cancer metastasis. Nature 1980;283:139-46. Spjut HJ, Roper CL, Butcher HR Jr. Pulmonary cancer and its prognosis. Cancer 1961;14:1251-8. Rienhoff WF, Talbert JL, Wood S Jr. Bronchogenic carcinoma: a study of cases treated at Johns Hopkins Hospital from 1935 to 1958. Ann Surg 1965;161:67484. Collier FC, Blakemore WS, Kyle RH, et al. Carcinoma of the lung: factors which influence five year survival with special reference to blood vessel invasion. Ann Surg 1957;146:417-23. Salzer G. A proposition on the classification of bronchus carcinomas according to there patho-anatomical clinical aspects. Wien Med Wochenschr 1951;101:102-3. Hinson KFW, Nohl He. Involvement of the pulmonary veins by bronchogenic carcinoma. Br J Dis Chest 1960; 54:54-8. Nohl HL. A three year follow-up of classified cases of bronchogenic carcinoma after resection. Thorax 1960; 15:11-6. Shields TW. Prognostic significance of parenchymal lymphatic vessel and blood vessel invasion in carcinoma of the lung. Surg Gynecol Obstet 1983;157:185-90.
Discussion Dr. Thomas W. Shields (Chicago. Ill.). We generally do not see as many satellite nodules proportionately as the Quebec group has reported. We do not see as many squamous cell carcinomas, either, and that may be one of the factors. However, when such a nodule is encountered, the questions are
The Journal of
5 12
Thoracic and Cardiovascular Surgery
Deslauriers et al.
how to classify the disease and what kind of prognosis it denotes. Should the nodule be ignored, or should it be classified as MI disease? Certainly the data presented in this report support Dr. Deslauriers' contention that this should be considered IlIA disease and that it is a poor prognostic finding. I do have one disagreement with Dr. Deslauriers, though. I do not accept that blood vessel invasion per se, as demonstrated on histologic examination of a resected specimen, is of any prognostic value. A great deal of literature says that it is. However, some of the more recent work disputes that theory. In the work that we did with the Veterans Administration Surgical Oncology Group, which was published in Surgery, Gynecology & Obstetrics a couple of years ago, we found by multivariate analysis that blood vessel invasion per se was of no prognostic significance whatsoever when it was determined by the standard routine histologic examination. In contrast, we did find that lymphatic invasion or permeation in the lung tissue, even in the absence of lymph node metastasis, was a prognostic feature. Furthermore, in the absence of lymph node metastases, the patient who had lymphatic invasion within the pulmonary parenchyma had the same prognosis as did the patient who had NI disease. This was particularly true in patients with squamous cell than in patients with the other cell types. Dr. Martin F. McKneally (Albany. N.y.) In human breast cancer and colon cancer, and in many experimental cancers, multiple cancers in the same field are very common. I wonder if it is necessary to postulate metastasis as the mechanism? Second, wouldn't exclusion of the incomplete resections be appropriate to make the groups more balanced? Dr. Deslauriers (Closing). Multicentric colon carcinomas are fairly common, especially in diseases such as ulcerative colitis. In such cases, each tumor has a separate epithelial origin and, from a pathologic point of view, each one has to be considered different from the other. Although it is difficult to
be absolutely certain that satellite nodules, such as defined in this paper, are not in fact multicentric carcinomas, it seems unlikely that they are, since they have none of the currently accepted diagnostic criteria for synchronous carcinomas. Furthermore, patients with multiple primaries may have a favorable outcome after complete resection, whereas the presence of a bronchogenic carcinoma accompanied by satellite nodules nearly always indicates an advanced stage of disease with a grave prognosis. The issue of microscopic blood vessel invasion as a factor of prognosis is still controversial in the literature. Nohl' showed that some degree of pulmonary vein involvement was present in every patient with lung cancer, but that infiltration of the larger veins exerted a negative influence on the chances of survival. Shields' paper- was the first one to demonstrate that blood vessel invasion when identified by routine histologic examination provided little, if any, additional predictive information. The Lung Cancer Study Group has similar data. There is nevertheless much evidence that, in general, the most important route of tumor spread involves invasion and penetration of tumor cells into blood vessels.' Satellite nodules associated with bronchogenic carcinoma are probably more common than reported, and it is important for the surgeon and/or pathologist to search for them in resected lung cancer specimens. REFERENCES I. Nohl HL. A three year follow-up of classified cases of bronchogenic carcinoma after resection. Thorax 1960; 15:11-6. 2. Shields TW. Prognostic significance of parenchymal lymphatic vessel and blood vessel invasion in carcinoma of the lung. Surg Gynecol Obstet 1983;157:185-90. 3. Poste G, Fidler 11.The pathogenesis of cancer metastasis. Nature 1980;283:139-46.
Carcinoma of the lung Evaluation of satellite nodules as a factor influencing prognosis after resection Satellite nodules are considered to be predictive of poor prognosis in breast cancer and in melanoma. In lung cancer, there is no information as to tbeir definition, prevalence, or implication as a prognosis factor of survival after resection. Over the past 18 years (1969 to 1987), 84 patients underwent pulmonary resection for primary lung cancer accompanied by satellite nodules. Tbese nodules were defined as well-circumscribed accessory carcinoma foci clearly separated from the main tumor but with identical histologic characteristics. All were smaller than the primary carcinoma and most were located within the same lobe. Survival rates of patients with satellite nodules were compared to those of 1021 patients without satellite nodules who underwent resection during the same time interval. The 1-,3-, and 5-year survival rates for all patients classified as having no satellite nodules were 78 %, 54 %, and 44 %, respectively, and the median survivalfor the entire group was 30 months. In patients with satellite nodules, these survival rates were 60.9%, 32.7%, and 21.6%, respectively, with a median survival of 15 months. The deleterious effect of satellite nodules was more significant in patients with stage I disease (p = 0.0008) than in patients with stage II (p = 0.0354) or stage rn (p = 0.0145) disease. Survival data obtained by comparison of satellite nodule status and histologic characteristics shows that 5-year survival figures are better for patients with no staellite nodules in both the squamoll'i and the nonsquamous groups. This study demonstrates that satellite nodules associated with lung cancer are indicative of locally advanced and/or premetastatic disease. Tbese patients should be included in the stage group rna of the TNM stage grouping classification.
Jean Deslauriers, MD," Jacques Brisson, Moe (by invitation), Raymond Cartier, MD" (by invitation), Marcien Fournier, MOb (by invitation), Denis Gagnon, BSc d (by invitation), Michel Piraux, MD" (by invitation), and Maurice Beaulieu, MD" (by invitation), Ste-Foy, Quebec, Canada
Satellite nodules are considered to be predictive of poor prognosis in malignant disease of the breast and in melanomas. In breast cancer' satellite skin nodules clearly indicate important local tumor spread and, according to the clinical staging nomenclature (TNM system), satellitosis confined to the same breast is a T4 descriptor (stage IV). The presence of satellite nodules
From the Division of Thoracic Surgery, Centre de Pneumologie,' and the Department of Pathology,' Hopital Laval; and the Departments of Epidemiology' and Statistics," Laval University, Ste-Foy, Quebec, Canada. Read at the Sixty-eighth Annual Meeting of The American Association for Thoracic Surgery, Los Angeles, Calif., April 18-20, 1988. Address for reprints: Dr. Jean Deslauriers, 2725, chemin Ste-Foy, Ste-Foy, Quebec G 1V 4G5, Canada.
504
within 2 em of any primary melanoma' is also considered to be a T4 descriptor. In lung cancer there is no specific information as to the definition of satellite pulmonary nodules, their prevalence, or even their implication as a prognostic factor of survival after resection. In addition, there is no provision for their coding in the new definitions for the TNM categories of primary tumors (T status).' Over the past 18 years, we have identified 84 patients having resection of a primary bronchogenic carcinoma in which the primary lesion was accompanied by malignant satellite nodules. This report presents a comparative survival analysis between those individuals and 1021 patients without satellite nodules who underwent resection during the same time interval. The results indicate that the presence of satellite nodules has a negative impact on survival and a proposal is made for
Volume 97 Number 4 April 1989
Carcinoma of lung
50 5
Fig. 1. Right lung with lower lobe containing squamou s carcinoma, pathologic stage T2 N2 , and one satellite nodule (arrow) central to the main tumor.
an addition to the currently accepted TNM nomenclature. Patients and methods Selection of patients: pathologic terminology. Several morphologic and histologic criteria were used to differentiate a single tumor with satellite nodules from independent synchronous carcinomas or from pulmonary metastasis arising from a tumor in another organ system . Satellite nodules were defined as well-circumscribed accessory carcinoma foci adjacent to but clearly separated from the main tumor by normal lung parenchyma (Fig. I). All were smaller than the primary carcinoma and most were located within the same lobe. On histologic examination, they were described as parenchymatous tumor deposits with the same histologic features as the primary carcinoma. Patients whose satellite lesion proved to be a lymphatic or a lymph node metasta sis were not included in this study. Also excluded were patients whose pulmonary lesions were by histologic examination highly indicative of being metastatic from an unknown primary site. All histologic material included in the series was initially
assessed by the consultant pathologist. This material was subsequently reviewed by one of the authors (M.F.) to ascertain the presence of nodules, revise diagnostic criteria, and determine the histologic grade of both primary tumor and satellite nodules. Information on patients designated as having definite or probable synchronous lung cancers was previously reported " and is not included in this report . These patients fulfilled the following two criteria: (I) Their tumors were physically separate and in some instances located in different lobes; (2) their tumors were of different histologic types or were shown to arise from different endobrochial lesions by bronchoscopic study or from separate foci of carcinoma in situ at pathologic examination. Differentiating satellite nodules from pulmonary metastases was done by recording of past a nd present medical history, investigation of other organ systems as possible sources of malignancy, anal ysis of pathologic characteristics of the neoplasm, careful observation during the follow-up period, and postmortem examination. Four patients in this series had malginant tumors of other organ systems considered to be unrelated to the lung tumor. Two individuals had laryngeal carcinomas 6 and 10 years before their lung cancer; neither
The Journal of Thoracic and Cardiovascular
5 0 6 Deslauriers et al.
Surgery
Table I. General characteristics of 1105 patients included in the comparison of survival by satellite nodule status (1969 to 1987)
Table II. Pathology and stage of 1105 patients included in the comparison of survival by satellite nodule status (1969 to 1987)
Satellite nodules Characteristics
Age (yr) Mean (±SD) Range Sex Male Female Staging mediastinoscopy Yes No Type of operation Limited resection Lobectomy Pneumonectomy Complete resection Yes No
Present (n = 84)
Absent (n= /02/)
58.3 (± 9.65) 37-76
60.3 (± 8.53) 28-83
68 (81%) 16 (19%)
884 (86.6%) 137 (13.4%)
79 (94%) 5 ( 6%)
918 (89.9%) 103 (10.1 %)
3 ( 3.6%) 35 (41.7%) 46 (54.8%)
26 (2.5%) 587 (57.5%) 408 (40.0%)
80 (95.2%) 4 (4.8%)
1021 (100%) 0
had clinical evidence of local recurrence at the time of pulmonary resection. A third patient had a sarcoma of the lower extremity removed 3 months before being operated on for lung cancer, and the fourth underwent resection of a carcinoma in situ of the cecum 8 years after pulmonary resection. General characteristics of the study population. Between January 1969 and December 1986, 1105 patients underwent elective pulmonary resection as primary treatment for bronchogenic carcinoma. Eighty-four individuals had a carcinoma accompanied by one or more satellite nodules as defined earlier, and the other 1021 patients had no satellite nodules: Age and sex distributions were relatively similar in the two groups (Table I). Mediastinoscopy, with random node sampling at three different levels, was done in almost every case for pretreatment staging of the superior mediastinum. Only 20 patients (1.8%) were operated on when nodal metastasis was identified by mediastinoscopy. Limited resections (segmental/wedge) were performed when the patient's pulmonary reserve was thought to be compromised, but these were infrequently done in either group. Patients undergoing lobectomy include all individuals with simple lobectomies, bilobectornies, sleeve lobectomies, or extended lobectomies. Similarly, all patients with standard, intrapericardial, or extended pneumonectomies are included in the pneumonectomy group. Pneumonectomy was done more often in patients with satellite nodules (54.5%), whereas lobectomy was more frequently done in patients without satellite nodules (57.5%). On the basis of removal of all gross carcinoma, disease-free resection margin, and highest mediastinal node free of tumor, all procedures were considered to be complete in the no satellite nodule group. Resection was incomplete in four patients (4.8%) of the satellite nodule group because the final pathology report identified microscopic residual cancer at the
Satellite nodules Characteristics
Histologic type Squamous cell carcinoma Adenocarcinoma Bronchoalveolar carcinoma Large cell carcinoma Small cell carcinoma Mixed tumors Stage Stage I TI NO T2 NO T1 NI Stage II T2 NI Stage III T3 NO T3 NI T3 N2 T2 N2 T3 N2
Present (n = 84)
Absent (n = /02/)
35 (41.7%)
589 (57.7%)
21 (25.0%) 4 (4.8%)
171 (16.7%) 41 (4.0%)
20 (23.8%)
158 (15.5%)
I (1.2%)
22 (2.2%)
3 (3.6%)
40 (3.9%)
43 (51.2%) 12
566 (55.4%) 216 325 25 236 (23.1%)
27 4
16 (19.0%) 25 (29.8%) 4
219 (21.5%)
11
84 8 42 13
I 6 3
72
bronchial resection margin (n = 3) or in the highest mediastinal node (n = I). Histopathology and stage of disease. Pathologic diagnosis was obtained by the study of surgical specimens in all patients. The majority of resected neoplasms were squamous carcinomas, but their prevalence was proportionally higher in the no satellite nodule group (57.7%) than in the satellite nodule group (41.7%) (Table II). In contrast, adenocarcinomas and large cell carcinomas were more common in the satellite nodule group. Few patients in either group had bronchoalveolar cell carcinoma or small cell carcinoma. In the survival analysis, the squamous group comprised those having pure squamous disease, whereas the nonsquamous group contained all other patients including some with a mixture of cell types. All tumors were pathologically staged in accordance with the TNM terminology as outlined by the American Committee on Cancer Staging.' For the first 9 years of the study (1969 to 1978), the staging was done retrospectively with bronchoscopy diagrams, chest films, and operative and pathology reports. For the remaining years (1978 to 1987), staging was based on complete clinical and pathologic analysis of the primary tumor (T status) and surgical sampling in at least the bronchopulmonary, hilar, and mediastinal nodal areas (N status). For any given patient, the presence of satellite nodules was not taken into consideration when formulating a final stage of disease. Stage distribution for patients included in this
Volume 97 Number 4
Carcinoma of lung
April 1989
Table m. Characteristics of the main tumor and satellite nodules in 84 patients with satellite nodules Primary tumor Location Right upper lobe Right middle lobe Right lower lobe Left upper lobe Left lower lobe Lung zone Central Mid-lung Peripheral Size 0-3 ern
3.1-6 em >6cm Histologic grade Well differentiated (grade I) Moderately well differentiated (grade II) Poorly differentiated and anaplastic (grade III) Satellite nodules No. of nodules I
90 80
70
No. of patients (%)
Characteristics
0/0
26 (31%) 3 17 24 (28.6%) 14
507
60 50 40
,~54.8%
30 20 10
~45.2%
YEAR 28 28 28 21 3 60 (71.4%)
16 26 42 (50%)
53 (63.1%)
2 >3
4 19 Unknown 8 Degree of differentiation versus primary tumor 8 Better differentiation Same differentiation 61 (72.6%) Worse differentiation 12 Unknown 3 Seen on roentgenogram 13 (15.5%) Yes 71 (84.5%) No
series shows that it is nearly identical in the two groups (Table II). Tumor characteristics in patients with satellite nodules. In patients with satellite nodules, the primary carcinomas were predominantly located in the upper lobes but evenly distributed among the various lung zones (Table III). Most tumors were large (>6 em) and 50% had a poor histologic grade. On the basis of their anatomic location, four subsets of nodules were identified: (1) nodules peripheral to the main tumor (n = 38), (2) nodules central to the main tumor (n = 12), (3) nodules around the main tumor (n = 18), and (4) nodules in a different lobe (n = 16). The majority of patients (63%) had a single satellite nodule and most were of the same histologic grade as the primary carcinoma. All preoperative chest films were reviewed, and in only 15.5% of patients could the satellite nodule be identified. Follow-up. Patients were followed up from the date of operation to the time of death, loss to follow-up, or end of observation (December 1987), whichever came first. Only five
# At risk ... •
1021
791
574
429
335
251
84
49
27
23
15
11
k---Jt.
NO SATELLITE NODULE
.........
SATELLITE NODULE
Fig. 2. Life-table analysis showing the percentage of all patients remaining alive after resection of bronchogenic carcinoma with or without satellite nodules.
patients (0.5%) were lost to follow-up. The potential duration of follow-up was at least 12 months for all patients under review. The mean follow-up times were 28 months and 43 months, respectively, in the satellite nodule group and in patients without satellite nodules. Patients alive at the end of the follow-up period were classified as being (1) disease free or having (2) evidence of local recurrence, (3) second primary tumors, or (4) systemic metastasis. A local recurrence was defined as tumor growth within the ipsilateral hemithorax or mediastinum, or both. For patients who died, the cause of death was determined from clinical records (73% of patients were followed up at the hospital), death certificates, or from information provided by family physicians and by relatives of the patient. Each death was classified as being due to (I) lung cancer, including second primary lesions and operative deaths, (2) causes unrelated to lung cance, or (3) unknown causes. Operative mortality was defined as a death occurring within 30 days of the operation or a death directly related to the procedure even if it occurred more than 30 days postoperatively. Minor complications were defined as events that had little impact on the postoperative course or duration of hospital stay, and major complications were nonfatal but sometimes lifethreatening events. If a patient had both minor and major complications, he was included in the major complications group only. Statistical analysis. Results are presented as means (± standard deviation) for continuous variables and as percentages for categorical data. The analysis of survival was based on deaths from all causes. Survival was analyzed according to the Kaplan-Meier method and differences between survival curves of patients with and without satellite nodules were compared by the log-rank test. Multivariate analysis of failure times made up use of Cox proportional hazard model. In this analysis, age was included as a continuous variable and other factors were entered as one or two dichotomous indicator variables: presence of nodules (absent, present), sex (male, female), stage (I, II, III), histology (squamous, nonsquamous),
The Journal of
5 0 8 Deslauriers et al.
0/0
Thoracic and Cardiovascular Surgery
100
100
80
80
60
0/0
40
60 40 2')
20 2
2
YEAR
YEAR
3
YEAR
# AI risk
..
5C6
•
483371
43
20
1 CJ
282
2~J
1"7
~72
236
11
179125
16
_
k----IL NO SATELLITE NODULE
97
78
57
10
21912978 25
10
50
39
23
4
SATELLITE NODULE
Fig. 3. Life-table analysis by stage showing the percentage of all patients remaining alive after resection of bronchogenic carcinoma with or without satellite nodules.
SQUAMOUS
100
p
80
0/0
NON SQUAMOUS
100
= 0.0001
80
60
%
40 20
60 40 20
a
4
3
2
a
5
2
YEAR
3
5
YEAR
# At risk
A •
589 478 35
A---A
17
366
270
219
166
432
313
208
159
116
87
8
8
7
7
49
32
17
14
8
6
NO SATELLITE NODULE
. . . - SATELLITE NODULE
Fig. 4. Life-table analysis by histologic type showing the percentage of all patients remaining alive after resection of bronchogenic carcinoma with or without satellite nodules.
type of operation (lobectomy and limited resections, pneumonectomy), and complications of operation (none, minor, and major including operative deaths). The crude effect of each of these factors was assessed by entering only the variable of interest in the model. Then, after all other variables were taken into account, the effect of each factor was estimated on the basis of a model that included all factors simultaneously. Because the effect of satellite nodules was slightly affected by this adjustment, only adjusted results are presented. These results are described in terms of the coefficient of each indicator variable with its standard error, hazard ratios with their 95% confidence interval, and the p values. Hazard ratios for categorical variables can be interpreted as the ratio of the rate of mortality in one category of patients to the rate in the reference group. The p values were obtained from the X computed as the ratio of the coefficient to its standard error.
Results Operative morbidity/mortality. There was a close similarity of results between the two groups in the
3Q-day operative morbidity and mortality. Approximately two thirds of patients in either group had an uneventful postoperative period. Thirty-six patients died (3.5% operative mortality rate) in the no satellite nodule group and four in the satellite nodule group (4.8%). Most deaths occurred in patients who had had a pneumonectomy and were due to respiratory failure, pleuropulmonary infections, or bronchopleural fistulas. Comparison of overall survival and of survival by stage and histologic type. The observed survival rates for the entire group and each subset of the TNM classification are shown in Figs. 2 and 3. The 1-, 3-, and 5-year survival rates for all patients classified as having no satellite nodules were 78%, 54.3%, and 43.7%, respectively. For those patients with stage I disease, the 5-year survival rate was 54.4%. It was 40.4% for patients with stage II disease and 20.3% for those with stage III disease. Median survival for the entire group was 30 months. In patients with satellite nodules, the 1-, 3-, and 5-year survival rates for all patients were 60.9%,32.7%, and 21.6%, respectively. For those patients with stage I disease, the survival rate at 5 years was 32%, and it was 12.5% for patients with stage II disease and 5.6% for patients with stage III disease. Median survival was 15 months for all patients included in this group. The deleterious effect of satellite nodules on survival is observable in all patients involved in the comparison analysis but is more significant in patients with stage I disease (p = 0.6008) than in those individuals with stage II (p = 0.0354) or stage III (p = 0.0145) disease. Survival data obtained by comparison of satellite nodule status and histologic type (Fig. 4) shows that 5-year survival figures are better for patients with no satellite nodules in both the squamous (p = 0.0001) and the nonsquamous and mixed histology groups (p = 0.0475).
Volume 97 Number 4
Carcinoma of lung
April 1989
509
Table IV. Cox regression analysis of survival Factor Nodules Absent* Present Stage
It II III
Coefficient (5£)*
Hazard ratio
95% CI
0.4591 (0.1401)
1.0 1.6
1.2-2.1
0.00096
0.3200 (0.1198) 0.8768 (0.1141)
1.0 1.4 2.4
1.1-1.7 1.9-3.0
0.0076 <0.00005
0.5289 (0.0890)
1.0 1.7
1.4-2.0
<0.00005
0.5070 (0.1010)
1.0 1.7
1.4-2.0
<0.00005
0.1106 (0.1178) 0.4705 (0.1074)
1.0 1.1 1.6
0.9-1.4 1.3-2.0
0.35 <0.00006
p Value
Histologic type
Squarnoust Nonsquamous Operation Lobectomy and limited resectiont Pneumonectomy Complications of operation
Nonet Minor Major
CI. Confidence intervals. 'The effect of each factor is adjusted for age at diagnosis, sex and all other factors shown in the table. tReference group.
Multivariate analysis. In the crude analysis, the hazard ratio (95% confidence interval) for patients with satellite nodules was 1.9 (1.5 to 2.5) compared to the ratio for patients without nodules. After adjustment for age at diagnosis, sex, stage of disease, histologic type of the primary tumor, and type and complications of operation, the deleterious effect of satellite nodules on survival was still clear and statistically significant (Table IV). After such adjustment, the presence of nodules was associated with a 1.6-fold increase in the rate of mortality (hazard ratio: 1.6; 95% confidence interval: 1.2 to 2.1; p values: 0.00096). Site of first recurrence and cause of death. From the beginning of the study, 579 patients died in the no satellite nodule group and 61 in the satellite nodule group. The analysis of the probable cause of death in those patients (Fig. 5) shows that a higher proportion of individuals died of cancer recurrence in the satellite nodule group (87%) than in the no satellite nodule group (63%). The distribution of sites of first recurrences (Table V) is not statistically significantly different in the respective groups of patients, although the recurrence rate is much higher in patients with satellite nodules (71%). Approximately 50% of patients with satellite nodules and who were alive at last follow-up had evidence of recurrent disease. Discussion The problem of defining two or more malignant masses appearing simultaneously in the same lung is
difficult. A number of factors must be considered in diagnosing these lesions as independent primary tumors, metastases from a nonpulmonary neoplasm, multicentric carcinomas, or one being a central primary tumor accompanied by satellite nodules. Distinguishing among those entities not only is of academic interest but also has serious implications in planning treatment and evaluating chances of prolonged disease-free survival. The anatomic location of each lesion and its histologic appearance may provide absolute criteria. Martini and Melamed" noted that when the two lesions are physically distinct and have different histopathologic appearances, they represent dual primary malignant tumors. In patients with tumors of similar histologic appearances, it is necessary that they arise from separate foci of carcinoma in situ, that there is no lymphatic involvement, and that no extra pulmonary metastases are present at the time of diagnosis. Survival after adequate resection of synchronous lung cancers is similar to that of patients with isolated lesions.v" Determining whether two or more pulmonary masses are metastatic from an unknown primary site is even more difficult. Payne, Clagett, and Harrison' described factors that favor metastatic disease in the lung: (1) highly suspected primary tumor elsewhere in the body, (2) absence of tissue communication between tumor and bronchus, (3) absence of metastatic involvement of hilar or medistinal nodes, and (4) a pulmonary lesion, which has by histologic, electron microscopic, and immunocytochemistry techniques a nonbronchogenic cell type. Median survival for patients with lung metastases from
The Journal of
5 10
Thoracic and Cardiovascular Surgery
Deslauriers et al.
o o c
8.2°4
•
CANC ER SECOND PRIMARY OTHERS UNKNOWN
1.9%
26.1°4
8.6%
SATELLITE NODULES
NO SATELLITE NODULES
Fig. 5. Cause of death in 640 patients who died and are included in the comparison by satellite nodule status.
Table V. Site offirst recurrence in 526 patients with tumor recurrence included in the comparison by satellite nodule status Satellite nodules Present First recurrences No. of patients with first recurrences Site of first recurrence Local Distant Local and distant Second primary Unknown
(n = 84)
Absent (n
= 1021)
60 (71.4%)
466 (45.6%)
16 (26.7%) 29 (48.3%) 5
131 (28.1%) 214 (45.9%) 26
4 6
70 25
an occult primary source is in the vicinity of 3 to 4 months" and is unchanged by any treatment modality. Satellite nodules have not been previously described, although multinodularity is known to be a growth pattern of bronchoalveolar cell carcinomas. This pattern is, however, rarely seen in patients with resectable disease, who generally have a single solitary nodule," Multinodularity is most often associated with poorly differentiated and unresectable tumors, as it nearly always indicates an advanced stage of disease with a grave prognosis.'?" Most satellite nodules are diagnosed once the surgical specimen is sliced and carefully examined. In contrast to the multinodularity of bronchoalveolar cell carcinomas, satellite nodules are clearly associated with a larger central tumor. They are histologically identical to the main tumor and most are located within the same lobe as the central mass.
The cause of malignant satellite nodules is not well established, but they should be considered as locally advanced disease rather than metastases, which are defined as the transfer of cancer from one organ to another not connected with it. 13, 14 The metastatic process starts, however, by local infiltration of host tissues followed by penetration of blood and lymph vessels with subsequent dissemination. IS Satellite nodules are probably, in some cases, the end result of peripheral emboli after tumor invasion of the pulmonary artery. The fact that in this series 63% (53/84) of all nodules were located peripherally or around the main tumor, usually within the same pulmonary artery distribution as the central mass, gives credence to the theory of pulmonary artery invasion. The association of bronchogenic carcinoma with centrally located satellite nodules is less common (14.3%) and probably reflects lymphatic or more likely pulmonary vein embolization. Blood vessel invasion has indeed been long recognized l 6• 17 as a poor prognostic factor for survival in resectable lung cancer. Collier and associates" even suggested that blood vessel invasion was perhaps more significant in terms of prognosis than lymph node metastases. Similarly, the finding of pulmonary vein invasion was included in Salzer's original classification" of carcinoma of the bronchus. Hinson and Nohl" showed that some degree of pulmonary vein involvement was present in every patient with lung cancer, but that infiltration of the larger veins exerted a negative influence on the chance of survival." In contrast, Shields" noted that blood vessel invasion, when identified by routine histologic examination, had little impact on long-term prognosis. Nodules located in a different lobe probably represent blood-borne metastasis, although it is possible that they
Volume 97 Number 4
Carcinoma of lung
April 1989
are the end result of air-borne metastasis as described in multicentric bronchoalveolar cell carcinomas. Preoperative detection of satellite nodules accompanying lung cancer is rare, although it is likely that the frequency of their preoperative discovery will rise with the increasing use of computed tomographic scanning in the staging of lung cancer. Their finding preoperatively or during resection should not act as a deterrent to curative operation even if they are located in a different lobe. Their presence should, however, be taken into consideration when staging the disease, estimating the patient's chances of long-term survival, and/or entering subjects in surgical adjuvant clinical trials. The survival data presented in this study clearly indicate that satellite nodules associated with lung cancer act as independent indicators of locally advanced or premetastatic disease. These data also indicate that these tumors have a biology similar to that of resectable stage III lesions (Figs. 2 and 3). Most patients included in this category have extra pulmonary extension of their tumor (T3) but are still candidates for complete resection. In the recently modified TNM stage grouping,' patients with the T3 descriptor and without N3 disease or Ml disease are classified as having stage IlIa disease. As reported by Mountain,' 13% of patients in this stage group are expected to survive 5 years, a figure that is comparable to that of patients with satellite nodules undergoing resection (21%), as shown in this study. On the basis of this information, we recommend that the T3 descriptor of the currently accepted TNM staging system be modified to read as follows: T3-A tumor of any size with direct extension into the chest wall (including superior sulcus tumors), diaphragm, or the mediastinal pleura or pericardium without involving the heart, great vessels, trachea, esophagus, or vertebral body, or a tumor in the main bronchus within 2 em of the carina without involving the carina, or a tumor accompanied by satellite nodules. As a result, patients with resectable bronchogenic carcinoma and satellite nodules will be included in the stage group IlIa.
1.
2.
3. 4.
REFERENCES Harris HR, Hellman S, Camellos GP, Fisher B. Cancer ofthe breast. In: Devita VT Jr, Hillman S, Rosenberg SA, eds. Cancer, principles and practice of oncology. Philadelphia: JB Lippincott, 1985:1114-77. Mastrangelo MJ, Baker AR, Katz HR. Cutaneous melanoma. In: Devita VT Jr, Hillman S, Rosenberg SA, eds. Cancer, principles and practice of oncology. Philadelphia, JB Lippincott, 1985:1371-1422. Mountain CF. A new international staging system for lung cancer. Chest 1986;89(suppl):225s-33s. Ferguson MK, Demeester TR, Deslauriers J, Little AG,
5. 6. 7.
8. 9.
10. 11. 12. 13.
14. 15. 16. 17.
18.
19.
20.
21.
22.
511
Piraux M, Golomb H. Diagnosis and management of synchronous lung cancers. J THoRAc CARDIOVASC SURG 1985;89:378-85. Carr DT. Staging of lung cancer. Am Rev Respir Dis 1978;117:819-23. Martini N, Melamed MR. Multiple primary lung cancers. J THORAC CARDIOVASC SURG 1975;70:606-12. Payne WS, Clagett OT, Harrison EG. Surgical management of bilateral malignant lesions of the lung. J THoRAc CARDIOVASC SURG 1962;43:279-90. Holmes FF, Fouts TL. Metastatic cancer of unknown primary site. Cancer 1979;26:816-20. Creco RJ, Steiner RM, Goldman S, Cotler H, Patchefsky A, Cohn HE. Bronchoalveolar cell carcinoma of the lung. Ann Thorac Surg 1986;41:652-6. Donaldson JC, Kaminsky DB, Elliott nc. Bronchiolar carcinoma. Cancer 1978;41:250-8. Marc Q, Galy P. Bronchiolo-alveolar carcinoma. Am Rev Respir Dis 1973;107:621-9. Tao LC, Delarue NC, Saunders D, Weisbrod G. Bronchiolo-alveolar carcinoma. Cancer 1978;42:2759-67. Fidler 11, Hart IR. Principles of cancer biology: cancer metastases. In: Devita VT Jr, Hillman S, Rosenberg SA, eds. Cancer, principles and practice of oncology. Philadelphia: 18 Lippincott, 1985:113-24. Fisher ER, Fisher B. Recent observations on concepts of metastasis. Arch Pathol 1967;83:321-4. Poste G, Fidler 11. The pathogenesis of cancer metastasis. Nature 1980;283:139-46. Spjut HJ, Roper CL, Butcher HR Jr. Pulmonary cancer and its prognosis. Cancer 1961;14:1251-8. Rienhoff WF, Talbert JL, Wood S Jr. Bronchogenic carcinoma: a study of cases treated at Johns Hopkins Hospital from 1935 to 1958. Ann Surg 1965;161:67484. Collier FC, Blakemore WS, Kyle RH, et al. Carcinoma of the lung: factors which influence five year survival with special reference to blood vessel invasion. Ann Surg 1957;146:417-23. Salzer G. A proposition on the classification of bronchus carcinomas according to there patho-anatomical clinical aspects. Wien Med Wochenschr 1951;101:102-3. Hinson KFW, Nohl He. Involvement of the pulmonary veins by bronchogenic carcinoma. Br J Dis Chest 1960; 54:54-8. Nohl HL. A three year follow-up of classified cases of bronchogenic carcinoma after resection. Thorax 1960; 15:11-6. Shields TW. Prognostic significance of parenchymal lymphatic vessel and blood vessel invasion in carcinoma of the lung. Surg Gynecol Obstet 1983;157:185-90.
Discussion Dr. Thomas W. Shields (Chicago. Ill.). We generally do not see as many satellite nodules proportionately as the Quebec group has reported. We do not see as many squamous cell carcinomas, either, and that may be one of the factors. However, when such a nodule is encountered, the questions are
The Journal of
5 12
Thoracic and Cardiovascular Surgery
Deslauriers et al.
how to classify the disease and what kind of prognosis it denotes. Should the nodule be ignored, or should it be classified as MI disease? Certainly the data presented in this report support Dr. Deslauriers' contention that this should be considered IlIA disease and that it is a poor prognostic finding. I do have one disagreement with Dr. Deslauriers, though. I do not accept that blood vessel invasion per se, as demonstrated on histologic examination of a resected specimen, is of any prognostic value. A great deal of literature says that it is. However, some of the more recent work disputes that theory. In the work that we did with the Veterans Administration Surgical Oncology Group, which was published in Surgery, Gynecology & Obstetrics a couple of years ago, we found by multivariate analysis that blood vessel invasion per se was of no prognostic significance whatsoever when it was determined by the standard routine histologic examination. In contrast, we did find that lymphatic invasion or permeation in the lung tissue, even in the absence of lymph node metastasis, was a prognostic feature. Furthermore, in the absence of lymph node metastases, the patient who had lymphatic invasion within the pulmonary parenchyma had the same prognosis as did the patient who had NI disease. This was particularly true in patients with squamous cell than in patients with the other cell types. Dr. Martin F. McKneally (Albany. N.y.) In human breast cancer and colon cancer, and in many experimental cancers, multiple cancers in the same field are very common. I wonder if it is necessary to postulate metastasis as the mechanism? Second, wouldn't exclusion of the incomplete resections be appropriate to make the groups more balanced? Dr. Deslauriers (Closing). Multicentric colon carcinomas are fairly common, especially in diseases such as ulcerative colitis. In such cases, each tumor has a separate epithelial origin and, from a pathologic point of view, each one has to be considered different from the other. Although it is difficult to
be absolutely certain that satellite nodules, such as defined in this paper, are not in fact multicentric carcinomas, it seems unlikely that they are, since they have none of the currently accepted diagnostic criteria for synchronous carcinomas. Furthermore, patients with multiple primaries may have a favorable outcome after complete resection, whereas the presence of a bronchogenic carcinoma accompanied by satellite nodules nearly always indicates an advanced stage of disease with a grave prognosis. The issue of microscopic blood vessel invasion as a factor of prognosis is still controversial in the literature. Nohl' showed that some degree of pulmonary vein involvement was present in every patient with lung cancer, but that infiltration of the larger veins exerted a negative influence on the chances of survival. Shields' paper- was the first one to demonstrate that blood vessel invasion when identified by routine histologic examination provided little, if any, additional predictive information. The Lung Cancer Study Group has similar data. There is nevertheless much evidence that, in general, the most important route of tumor spread involves invasion and penetration of tumor cells into blood vessels.' Satellite nodules associated with bronchogenic carcinoma are probably more common than reported, and it is important for the surgeon and/or pathologist to search for them in resected lung cancer specimens. REFERENCES I. Nohl HL. A three year follow-up of classified cases of bronchogenic carcinoma after resection. Thorax 1960; 15:11-6. 2. Shields TW. Prognostic significance of parenchymal lymphatic vessel and blood vessel invasion in carcinoma of the lung. Surg Gynecol Obstet 1983;157:185-90. 3. Poste G, Fidler 11.The pathogenesis of cancer metastasis. Nature 1980;283:139-46.