Metastasectomy for soft tissue sarcoma

J THORAC CARDIOVASC SURG 1989;97:695-705

Metastasectomy for soft tissue sarcoma Further evidence for efficacy and prognostic indicators Between 1982 and 1987,74 patients (46 men and 28 women) had exploration for presumed metastases from high-grade soft tissue sarcoma of the bead/neck, extremity, or trunk. Ages ranged from 11 to 75 years (median 38 years). Thirty (41%) had multiple procedures for recurrences (range two to six explorations). Median postthoracotomy time for the group of patients with histologically confirmed sarcoma (n = 63) was 20.3 months. Patients rendered free of disease at initial thoracotomy had significantly longer postthoracotomy survival times (26.8 months median) than those with unresectable The prognostic significance of age, sex, location of metastatic disease(9 months median); P2 < O.OOOI~ primary tumor, disease-free interval, number of nodules on preoperative computed tomograms or conventional linear tomograms, numberof metastases resected, and the use of postoperative chemotherapy wereanalyzed. In a univariate analysis, sex, age, and location of the primary tumor did not impact significantly on survival, nor did the use of postoperative chemotherapy. Initial disease-free intervalof 1 year or less was associated with a significantly shorter survival time, and patients with five nodules or fewer on preoperative computed tomographyhad significantly longersurvival times than patients with six nodules or more.Patients with three nodules or fewer on linear tomography had a longerpostthoracotomy survival time than patients with four nodules or more. In patients whose malignant disease could be completely resected, the numberof nodules resectedat thoracotomydid not impact on long-term survival According to proportional-hazards modeling, disease-free interval, sex, resectability, and truncal location were found to associate with length of survival after metastasis removal. We conclude that pulmonary metastasis resection in patients with soft tissue sarcoma is associated with long-term survival, and consistent indicators can define which patients may benefit from these interventions.

David Jablons, MD," Seth M. Steinberg, Phlr," Jack Roth, MD," Stefania Pittaluga, MD,c Steven A. Rosenberg, MD, PhD," and Harvey I. Pass, MD," Bethesda. Md.

Soft tissue sarcomas are malignant tumors arising from connective tissue, largely from mesodermal structures. Aggressive surgical resection of the primary lesion combined with postoperative adjuvant therapy has resulted in 5-year survival rates of 40% to 60%. J The lungs are the most common site of distant metastases, with recurrence usually within the first 24 months after From the Thoracic Oncology Section, Surgery Branch,' Surgical Pathology Department,' and Biostatistics and Data Management Section: National Cancer Institute, National Institutes of Health, Bethesda, Md. Read at the Fourteenth Annual Meeting of The Western Thoracic Surgical Association in Hawaii, June 22-25, 1988. Address for reprints: Harvey I. Pass, MD, Head, Thoracic Oncology Section, Surgery Branch, National Cancer Institute, National Institutes of Health, Building 10, Room 2B07, Bethesda, MD 20892.

management of the primary lesion.' Few series in the literature have documented the efficacy of removing pulmonary metastases in patients with recurrent soft tissue sarcoma when the patient could be rendered free of disease by the thoracotomy. Questions remain, however, regarding the appropriate roentgenographic surveillance of such patients. Moreover, no specific guidelines addressing the prognosis of these patients have been formulated concerning criteria that identify patients who will live the longest with this approach, specifically when preoperative or intraoperative parameters are examined. Also, because of the small number of patients available from any given institution, the possible efficacy of postthoracotomy adjuvant therapy remains unanswered. The Surgery Branch of the National Cancer Institute, National Institutes of Health, has investigated the 695

696

The Journal of Thoracic and Cardiovascular Surgery

Jablons et al.

Table I. Soft tissue sarcoma: Histologic type of

Table II. Current status of patients having one

the primary tumor

exploration or more for metastatic soft tissue sarcoma-1982 to 1987

Synovial sarcoma Malignant fibrous histiocytoma Leiomyosarcoma Fibrosarcoma Spindle cell sarcoma Liposarcoma Schwannoma Alveolar soft parts sarcoma Unclassified Hemangiopericytoma Neuroectodermal sarcoma Angiosarcoma Epithelioid sarcoma

19 15 11

7 4 4 4 3 2 1

I 2 I

management of high-grade soft tissue sarcomas since 1974 and has adopted a policy of aggressive resection of pulmonary metastases in these patients. This report analyzes a second series of patients having operative management of pulmonary metastases from soft tissue sarcomas since 1982 in an attempt to identify factors that may predict long-term survival after metastasis resection. This newer series also attempts to analyze the role of computed tomography (CT) to predict the suitability for resection and the impact of postoperative adjuvants in a subset of these patients. Patients and methods From 1982 to 1987, 74 patients had resection of presumed metastases from high-grade soft tissue sarcoma by the Surgery Branch of the National Cancer Institute. The primary soft tissue sarcoma in these patients had been resected between 1975 and 1986. Six of the 74 patients in the series had already had one thoracotomy or more at our insitution before 1982 and were included in the analysis to follow their long-term survival. In 63 patients thoracotomy disclosed one pulmonary metastasis or more that were histologically consistent with the primary soft tissue sarcoma. Follow-up data were available on all 74 patients, with a median potential follow-up of 43.0 months from the time of initial thoracotomy. The median survival time of the patients with documented pulmonary metastases who are alive (with or without disease [n = 34]) was 28.0 months. Ages ranged from 11 to 75 years with a median age of 38 years. There were 46 male and 28 female patients. Primary tumors were located in the extremity (n = 58), trunk (n = 12), or head/neck (n = 4). All histologic types were high grade and are listed in Table I. Preoperative evaluation. After excision of the primary tumor, patients under the sarcoma protocols were evaluated with thoracic CT scanning or linear tomography (LT), or both, every 3 months for 2 years and with bone scans every year. Patients with newly developing abnormalities consistent with pulmonary metastases were offered resection of the metastases if the primary tumor was controlled, if the presence of a new pulmonary lesion was consistent with a metastasis, if

Exploration

Current status

Exploration I

68 Patients II Negative thoracotomy II Never recurred 57 Positive thoracotomy 8 Unresectable disease 8DWD 49 Resectable disease 7 NED 2AWD 15 DWD 25 Recurred-to exploration 2 Exploration 2* 30 Patients 7 Unresectable disease IAWD 6DWD 23 Resectable disease 4 NED 3DWD 16 Recurred-to exploration 3 Exploration 3t 17 Patients 5 Unresectable disease 5DWD 12 Resectable disease 2 NED IAWD IDWD 8 Recurred-to exploration 4 Exploration 4 8 Patients 8 Resectable disease IDWD 7 Recurred-to exploration 5 Exploration 5 7 Patients 2 Unresectable disease 2AWD 5 Resectable disease I NED 4 Recurred-to exploration 6 Exploration 6 4 Patients 2 Unresectable disease 2DWD 2 Resectable disease I NED IAWD AWD, Alive with disease; DWD, dead with disease; NED, no evidence of recurrent disease. *Five other patients had had an initial exploration before 1982. tOne other patient had had two explorations before 1982.

extra pulmonary metastases were absent, and if the amount of lung tissue remaining would provide the patient with sufficient pulmonary parenchymal reserve. Operative procedure. Conduct of the operation was as previously described.' Briefly, the metastases were resected through a median sternotomy or lateral thoracotomy with

Volume 97 Number 5 May 1989

Pulmonary metastases from soft tissue sarcoma

Table m. Operative approach (n = 134)-1982 to 1987 Exploration No.

Median sternotomy

I

56

2 3 4

18 10 4

5 6

o o

88 (66%)

Thoracotomy

12 12 7 4 7 4 46 (34%)

double-lumen endotracheal anesthesia. The right and left sides were individually explored by bimanual palpation of the collapsed lung. The thoracic cavity, including the mediastinum, chest wall, and hilar nodes, was thoroughly explored. Palpable abnormalities that did not appear to be obvious granulomas or intrapulmonary lymph nodes were excised with automatic stapling devices. Patients with nodules not amenable to wedge or segmental resections had formal pulmonary resection if they could be rendered free of disease, if lymph node sampling did not reveal metastatic sarcoma, and if discontinuous pleural metastatic disease was not present. Initially, the patient's condition was monitored via an indwelling radial arterial line, but over the past 2 years on-line arterial digital oximetry has been used exclusively. Roentgenographic monitoring. A nodule was defined as a focus of increased density, spherical, based on either the pleural or parenchyma, and unassociated with linear densities. Newly developing nodules were considered highly suggestive of sarcoma metastases and, as described earlier, served as an absolute indication for exploration. Chest CT examinations were performed with a GE 8800 CT IC scanner (General Electric Company, Fairfield, Conn.) by scanning the entire thorax in transaxial slices at 10 mm thick intervals, and the images were displayed on a 256 by 256 matrix. Contiguous I cm chest linear tomograrns were obtained in the coronal plane throughout the entire depth of the lungs. Statistical analysis. Survival durations were computed from the date of the initial thoracotomy until the date of the last known follow-up or date of death. Probabilities of survival were computed actuarially by the Kaplan-Meier method." Although 74 patients are available for analysis, we have generally restricted survival analyses to the 63 patients with documented sarcoma on initial exploration. The statistical significance of the difference between survival curves was determined by the Mantel-Haenszel technique.' Demographic and clinical factors were analyzed for their independent effect on survival by means of the Cox proportional hazards model." All p values are two-sided.

Results From 1982 to 1987, the 74 patients had a total of 134 thoracic explorations for presumptive metastatic soft tissue sarcoma. Sixty-eight patients had an initial exploration, and 30 patients had two or more explorations (66 total reoperations) (Table II). Symptoms and signs were

697

Table IV. Resection characteristics Procedure

Wedge only (one or more) Wedge and lobectomy Wedge and pneumonectomy Lobectomy alone Pneumonectomy alone Concomitant CWR

No. 115

7 I

9 2 6

CWR. Chest wall resection.

rare on initial thoracic exploration and included spontaneous pneumothorax (n = 1), weight loss (n = 2), shortness of breath (n = 1), and chest pain (n = 1). Four patients having repeated explorations had chest wall pain or pleural effusion before their reexploration, and all four had unresectable disease. Preoperative roentgenographic studies. The 74 patients were evaluated with a total of 142 CTs or LTs (83 CTs, 59 LTs) before the 134 thoracic explorations. In 31 instances CT and LT were both performed before nodule excision. Of the 205 total nodules resected in those patients having concurrent CT and LT, 99 CT and 59 LT abnormalities were detected, which highlights the greater sensitivity of CT compared with LT. No comment on specificity could be made, however, as specific roentgenographic abnormalities were not correlated with individually resected metastases. Operative findings. The explorations performed are summarized in Table III. Of the 57 patients with documented metastases having initial thoracotomy during this time period, 49 (86%) were rendered free of disease. Median sternotomy was the most common approach (86 median sternotomies versus 46 lateral thoracotomies). The 11 patients having five explorations or more all had lateral thoracotomies. Extent of parenchymal resection at the second thoracotomy is depicted in Table IV, with the most common operation being one or more wedge resections of peripheral nodules. In 11 patients (16%) only benign disease was found at initial exploration:Granuloma wasseen in four, intrapulmonary lymph nodes in two, and asbestosis, hamartoma, pleural cyst, osteophyte, and atelectasis in one each. None of these patients subsequently had sarcoma. Operative mortality and morbidity. There were no operative or perioperative deaths, but there were 11 complications in the 134 explorations (8%): supraventricular arrhythmia (n = 4), pneumonitis (n = 2), prolonged intubation (n = 1), congestive heart failure (n = 1), chylothorax (n = 1), recurrent nerve injury (n = 1), and superficial wound infection (n = 1).

698

The Journal of Thoracic and Cardiovascular Surgery

Jablons et al.

100

_.-f-ft--+-+H

90 80 ..J

70

< ....> >

60

::J lJl

50

a::

I-

Z

W

U

a:: W 0-

40 30 20 10

a

---+----I-----+----f---+-----!

L---~----f---+I

12

24

* 0 METS

LQ9Qndi

36

48

60

TIME IN MONTHS

72

84

96

108

o 1+ METS

1111 fc11Qd

39/63 fc11ad

Fig. 1. Survival after initial exploration in all patients. Patients found to have benign disease had significantly longer survival than those with malignant disease (P2 = 0.01). METS, Metastases.

100 90 80 ..J

70

< > .... >

80

::J lJl

50

a::

I-

Z

40

a::

30

W U

w

a..

I

I

20 10

a 12 LQgQndl

*

RESECTABLE 31/55 fa11Qd

24

36

48

60

TIME IN MONTHS

72

84

96

108

o NON-RESECTABLE 8/8 fc11Qd

Fig. 2. Survival after initial exploration in patients found to have metastases. Patients who could be rendered free of disease had significantly longer survival than patients found to have unresectable disease (P2 < 0.0001).

Volume 97 Number 5

Pulmonary metastases from soft tissue sarcoma

May 1989

699

100 90

BO ...J

70

< ....> >

80

Ul

50

Z

40

0::

30

0:: :J

.....

UJ U

UJ

a..

20 10

o

I 12

Lgggnd,

* 1-3 LT NODULES 24/35 fa1lgd

84

24

98

lOB

o 4+ LT NODULES SIS fotlgd

Fig. 3. Preoperative roentgenographic prognostic indicators-LT and influence on survival. Patients with three nodules or fewer on preoperative LT had significantly longer survival times than those patients with four LT nodules or more (P2 = 0.03)

Actuarial survival after resection. The median postthoracotomy survival time for the group of patients with histologically confirmed sarcoma (n = 63) was 20.3 months (Fig. I). Patients who were rendered free of disease at initial thoracotomy had significantly longer postthoracotomy survival times (26.8 months median) than those with unresectable metastatic disease (9.0 months median; P2 < 0.0001 for the overall differences between the curves (Fig. 2). Univariate analyses. Univariate (Mantel-Haenszel) analyses in the 63 patients indicated that sex, location of the primary tumor, chemotherapy after primary resection (before thoracotomy), or. primary site irradiation did not significantly affect survival, although there were slight trends toward improved survival in female patients or those with truncal location. Forty patients with pulmonary metastatic sarcoma had LTs before their initial exploration. Patients with three or fewer nodules on LT had a longer postthoracotomy survival time than patients with four nodules or more (P2 = 0.03, Fig. 3). No patient with four nodules or more on preoperative tomography lived longer than 3 years. Forty-three patients had preoperative CT before their initial exploration. Patients with five nodules or fewer were found to

live significantly longer (23.8 months median) than those with six nodules or more (3.6 months median; P2 < 0.0001 for the overall difference between the curves) (Fig. 4). Among the 63 patients with metastases at initial thoracotomy, a disease-free interval (i.e., the time from initial resection of the primary tumor to first recurrence of pulmonary metastasis) of greater than 1 year was associated with a longer survival time (median 26.5 months) than was a disease-free interval less than or equal to I year (median 12.6 months; P2 = 0.03 for the overall difference between the curves) (Fig. 5). Among 55 patients in the series who could be rendered free of disease at their first thoracotomy (49 between 1982 and 1987 and six from 1979 to 1982), no difference in survival could be determined between patients having four or fewer metastases resected (median survival 23.2 months) (Fig. 6), compared with patients with five metastases or more (median survival 27.8 months). The maximum number of metastases resected at initial thoracotomy was 32 in a patient with a leiomyosarcoma, and he remains alive with stable disease 46 months after the initial thoracotomy. Four other patients had 20 metastases or more resected at the initial thoracotomy.

700

The Journal of Thoracic and Cardiovascular Surgery

Jablons et al.

100 90 80 ....J

70

< ....> > a::

eo

Ul

50

IZ

40

:::J

UJ

U

a::

UJ

a,

30 20 10 0

12 LQ9Qnd.

24

36

48

60

72

84

96

108

TIME IN MONTHS

* 1-5 CT NODULES

o 6+ CT NODULES

20/38 fa11Qd

5/5 fa11Qd

Fig. 4. Preoperative roentgenographic prognostic indicators-CT and influence on survival. Patients with five nodules or fewer on preoperative CT had significantly longer survival times than those patients with six CT nodules or more (P2 < 0.00(1).

Two are alive (29 and 21 months after resection) and two are dead (3 months and 13 months after resection). Finally, of the 32 patients who did not receive adjuvant chemotherapy after their primary sarcoma operation, 15 received systemic adjuvant therapy after their thoracotomy, with no improvement in survival (Fig. 7). Statistical modeling of prognostic factors. The Cox proportional hazards modeling technique was used to identify factors that could be important in determining length of survival. From among 63 patients with documented sarcoma on initial exploration, four independent factors were found to jointly affect survival in a statistically significant fashion: disease-free interval, sex, resectability, and truncal location. Details of the parameters and their associated relative risks may be found in Table V. As disease-free interval increases from less than 1 to 1-2 years, or from 1-2 to 2-3 years, or from 2-3 to 3 years or more, the risk of dying is diminished by approximately 50%. In female patients the risk of dying is approximately 40% that of male patients when data are adjusted for other factors in this model. Patients who have resectable disease have only about one-sixth the risk of dying compared with those whose disease is unresectable, and location in the trunk is associated with

approximately one-fourth the risk of death compared with other locations, after adjusting for other factors in the model. These findings indicate that there are characteristic which may be useful to provide some evidence as to the probabilities of survival of patients examined in this study. Discussion This report discusses the second series of patients undergoing removal of pulmonary metastases from a primary soft tissue sarcoma at the National Cancer Institute. It reinforces the view that, with proper patient selection, long-term survival with this approach is possible. In the first series of 67 patients with histologically proved resected metastases, the actuarial postthoracotomy survival rate was 25% at 3 years, with a 32% 3-year survival rate for patients who could be rendered disease free by the exploration.' The present series verifies these findings, with a 28% 3-year survival rate for all patients with metastatic disease and a 35% 3-year survival rate for those who could be rendered disease free. Patients with unresectable disease at initial thoracotomy had a median survival in this series of 9.0 months, and none of these eight patients survived beyond 17 months.

Volume 97 Number 5 May 1989

Pulmonary metastases from soft tissue sarcoma 7 0 1

100 90 80 ...J

70

>

60

l/l

50

< > .....

0:: :J ~

Z

..0

0::

30

W U W

l~-----l

a,

20 10 0

+

I 12

36

24

48

60

72

84

96

108

TIME IN MONTHS Legend.

• OF

<-

I

YEAR

16/20 foiled

o

OF

>

1 YEAR

23/43 fOiled

Fig. 5. A disease-free interval of I year or less was associated with a shorter survival time than disease-free intervals of 1 year or more (Pl = 0.03).

One of the purposes in the analysis of results of both series of patients was to investigate whether preoperative prognostic variables existed that would better define subsets of patients who would benefit from resection of such metastases. The present series confirms that patient age is not associated with length of survival, but that sex and location of the primary tumor are important, as determined by the Cox proportional hazards model analysis. Results in the newer series of patients imply that, in the case of histologically confirmed metastases, management of the primary tumor with local irradiation, or addition of adjuvant chemotherapy (doxorubicin, cyclophosphamide) after primary sarcoma removal, does not impact -on survival after initial pulmonary metastatic resection. R
Basic criteria for metastatic resection stipulate that the primary site is controlled and that the patient will tolerate exploration and pulmonary resection. The absence of operative deaths in both series, as well as the low morbidity rate (8% in the present series, 8.6% in the first series), attests to the safety of the operation in this population of patients, who are generally in good health otherwise. Moreover, double-lumen anesthetic techniques with the disposable tube no longer complicate the anesthetic management, especially when tubes are placed endoscopically if there is any question of proper placement. The choice of median sternotomy has obvious advantages for the exploration, as reported previously by our group, including the detection of occult metastases and the avoidance of staged thoracotomy in patients with bilateral disease.' In the present series, only approximately 50% of all nodules resected (benign and malignant) were detected by CT, a fact that emphasizes the need for a complete bilateral exploration. The sternotomy approach has been used safely at our institute for metastasis removal up to four times; however, because of the possible difficulty with adhesion lysis and subsequent pinpointing of small recurrent nodules, we recommend lateral thoracotomy for the fifth and subsequent

702

The Journal of Thoracic and Cardiovascular Surgery

Jablons et a1.

100 90

BO ...J

<

70

>

80

> .... IX:

:::J (J)

I-

Z lLJ U IX: lLJ

50 40

-l

30

a,

20

~

10 0 12

24

36

48

60

72

84

96

108

TIME IN MONTHS Legend.

* 1-4 METS 19/35 felled

o

5+ METS 12/20 felled

Fig. 6. In patients who could be rendered disease free by metastasis resection, the number of nodules resected was not associated with survival (P2 = 0.68).

resections, only on the side that has the new abnormality, or staged thoracotomies for bilateral disease. With regard to the aggressiveness of pulmonary resection, we prefer wedge resections of individual nodules, ample enough to ensure a disease-free margin without removal of excessive parenchyma. On occasion we will use the technique of precision cautery described by Cooper and colleagues.8 However, we have not performed laser-assisted resections." If the lesion is large and central, or in proximity to major bronchovascular structures, we will perform a lobectomy through the sternotomy incision, an approach described by Urschel and Razzuk.'? In obese patients with a lesion of the lower lobe of the left lung, we have combined median sternotomy with lateral thoracotomy at one operation when we believed access could not be safely afforded to the inferior pulmonary vein. Similarly, we have encountered no difficulty with either chest wall biopsy or formal resection from the sternotomy approach in these patients, with preservation of overlying musculature and skin making reconstruction unnecessary. The use of radiographic criteria to define which patient will benefit from resection must be qualified. We have analyzed the number of CT and LT nodules for all patients proved to have metastatic sarcoma, whether the

nodules were resectable or not. This is justified, because before operative intervention we believed that all of these patients could be rendered disease free by resection; however, this was not the case in all patients. The presence of four nodules or more on conventional LTs again was associated with an unfavorable outcome, as in the first series. Since the first series, however, a prospective serial evaluation of 19 patients with soft tissue and osteogenic sarcoma revealed greater CT specificity and sensitivity in a static and dynamic analysis. I I After the results of this study, we began to use CT selectively for initial workup and follow-up of our sarcoma population in the surgery branch. These patients then served as the data base for the analysis of number of CT nodules as a prognostic indicator. The data support the greater sensitivity of CT scanning in monitoring patients with soft tissue sarcoma, and this report discloses that the presence of six CT nodules or more is associated with a poor prognosis. The greater number of nodules detected by CT compared with LT is apparent in the analysis and suggests that CT nodule number may be a better prognostic indicator than the number of LT nodules. The most important question, however, is the management of patients with six CT nodules or more; specifically,should they be submitted to exploration? To

Volume 97 Number 5

Pulmonary metastases from soft tissue sarcoma

May 1989

70 3

100 90

eo -l

70

< > ...... >

60

::::l Ul

50

0::

I-

Z

W U 0:: W

a..

40 30 20 10

0 12

24

36

48

60

72

84

96

108

TI ME I N MONTHS Lgggnd.

* POST-OP CT

a NO POST-OP CT

10/15 failed

6/17 foiled

Fig. 7. In patients who had not received adjuvant systemic therapy (CT) after their primary sarcoma resection, there was no difference in long-term survival if they received or did not receive postthoracotomy adjuvant therapy (P2 = 0.48).

Table V. Multivariate analysis of prognostic factors Variable OFI (per group)* Sex (female) Resectability Truncal location

Parameter estimate -0.65 -0.90 -1.82

-1.34

p-value

PR

95% Cl for RR

0.0001 0.016 0.0001 0.035

0.52 0.41 0.16 0.26

0.38,0.72 0.19, 0.85 0.07,0.40 0.08,0.91

RR, Relative risk; Cl, confidence interval; OF!, disease-free interval. 'Increasing from
exclude them at this time would require the presence of better salvage regimens, either chemotherapy or immunotherapy, which presently do not exist for soft tissue sarcoma. This subset of patients, however, may merit consideration for enrollment in experimental trials of preoperative systemic therapy in anticipation of resection, because in this group objective response criteria could be recorded. Disease-free interval remains a strong prognostic indicator as it was in the first series of patients. Tumor doubling time was not examined in this series of patients. A few words of caution are in order at this point. This

was a retrospective study of all patients who had thoracic explorations since 1982 and whose primary disease was soft tissue sarcoma. Whenever we have indicated that a certain number of nodules or metastases were important in survival, or that the effects of a disease-free interval of greater than or less than a certain length of time was important, these determinations were made after we had conducted a very small number of repeated examinations of the data. Resulting significant levels that are small (P2 < 0.01) are likely to be indicative of important differences, but the findings with levels of significance between approximately 0.01 and 0.05 mayor may not apply when examined in other popula-

704

The Journal of Thoracic and Cardiovascular Surgery

Jab/ons et a/.

tions. These implicit multiple comparisons should be kept in mind when these results are interpreted. In addition, the Cox proportional hazards modeling technique can indicate reasonably well the joint effects of independent factors on survival, The results presented are derived from the 63 patients in our study, which is too small a series to provide infallible determinations. There may also have been other factors that were not examined and that may influence outcome as well. The identified factors were all reasonably statistically significant, though, and thus should be considered to be appropriately identified from this study group. Confirmation of these findings by similar examinations of other series, or prospectively, would strengthen the findings we have presented. The final important question to be addressed by this series concerns the intraoperative management of patients who can be rendered disease free by multiple resections. This has been addressed only superficially in the literature. Specifically, is there a threshold in the number of nodules to resect if the only question of resectability is nodule number detected at thoracotomy? Among patients whose disease could be completely resected (as contrasted with the CT and LT data that include patients with unresectable disease), we found no differences in long-term survival between patients with five nodules or more compared with those with fewer than five nodules. The limiting criteria, then, would only be the amount of functioning lung tissue preserved to ensure a satisfactory physiologic outcome for the patient. Such an aggressive approach, however, does imply that patients having repeated resections, after numerous nodule resections, should be evaluated preoperatively with pulmonary function testing and arterial blood gas parameters to ensure a satisfactory physiologic result after subsequent resections. The need for such an aggressive surgical approach is further bolstered by the inability of postoperative adjuvant chemotherapy to prolong survival in our series of patients whose malignant disease could be completely resected. REFERENCES 1. Rosenberg SA, Tepper J, Glatstein E, et al. Prospectiverandomized evaluation of adjuvant chemotherapy in adults with soft tissue sarcomas of the extremities. Cancer 1983;52:424-34. 2. Potter DA, Glenn J, Kinsella T, et al. Patterns of recurrence in patients with high grade soft-tissue sarcomas. J Clin Oncol 1985;3:353-66. 3. Putnam JB, Roth JA, Weslev MN, et al. Analysis of prognostic factors in patients undergoing resection of pulmonary metastases from soft tissue sarcomas. J THORAC CARDIOVASC SURG 87: 260. 1984;87:260-8.

4. Kaplan EL, Meier P. Non-parametric estimation from incomplete observations. J Am Stat Assoc 1958;53:45781. 5. Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chern Rep 1966;50:163-70. 6. Cox DR. Regression models and life-tables. J R Stat Soc B 1972;34:187-220. 7. Roth JA, Pass HI, Wesley MN, et al. Comparison of median sternotomy and thoracotomy for resections of pulmonary metastases in patients with soft tissue sarcomas. Ann Thorac Surg 1986;42:134-8. 8. Cooper JD, Perelman M, Todd TR, et al. Precision cautery excision of pulmonary lesions. Ann Thorac Surg 1986;41:51-3. 9. LoCicero J III, Hartz RS. Frederiksen JW, Michaelis LL. New applications of the laser in pulmonary surgery. Ann Thorac Surg 1985;40:546-50. 10. Urschel HC Jr, Razzuk MA. Median sternotomy as a standard approach for pulmonary resection. Ann Thorac Surg 1986;41:130-4. 11. Pass HI, Dwyer A, Makuch R, Roth JA. Detection of pulmonary metastases in patients with osteogenic and soft-tissue sarcoma: the superiority of CT scan compared with conventional linear tomograms using dynamic analyses. J Clin Oncol 1985;3:1261-5.

Discussion Dr. Robert L. Mitchell (Mountain View, Calif). The rationale for metastasectomy in this cohort of patients is a median survival of 26.8 months in patients rendered disease free versus 9 months in patients with unresectable lung metastasis. Is this sufficient evidence to advise routine metastasectomy? Proffered purpose for this paper is to identify factors that will predict long-term survival after metastasectomy. Selection of patients is not well defined except that the primary tumor was controlled, there was no known metastasis other than in the lung, and the patient could tolerate pulmonary resection. Were all the patients referred to the National Cancer Institute with soft tissue sarcoma and pulmonary metastasis accepted regardless of Karnofsky score, or is this a highly select group of patients? The paper does state that the patients were essentially free of symptoms. What have we learned from this paper? Metastasectomy with the previously mentioned caveats can be performed with low morbidity and mortality rates-8% and 0%, respectively, in this series. This has been confirmed in other series previously. Statistical analysis revealed that favorable predictors were disease-free interval of greater than 1 year, three or fewer on LT or five or less on CT, and ability to resect metastatic disease completely. Twice as many nodules were found at thoracotomy as defined by CT. Are we to assume that the surgeon's dexterity permitted palpation and resection of all metastatic nodules even after two to four operations, and the term "disease-free" is accurate? The authors conclude that the resection of pulmonary metastasis in patients with soft tissue sarcomas is associated

Volume 97 Number 5 May 1989

with long-term survival. I presume they are implying a causal relationship. However, this series is not controlled, and indeed no historical controls are mentioned. I am aware of no randomized controlled series in the literature. The difference in survival of 26.8 months (median) for disease-free patients after resection and 9 months (median) in patients with unresectable lesions could be a function of the biologic behaviorof the tumor, stage of the disease at thoracotomy, and patient selection. Finally, the authors state: 'The need for such an aggressive surgical approach is further bolstered by the inability of postoperative adjuvant chemotherapy to prolong survival ...." I would suggest that the mere fact that other treatment modalities are not efficacious does not, by itself, provide the surgeon a rationale to operate. I have two questions for the authors. Could you better define patient selection, and can you justify this report without either concurrent or historical controls? Dr. Jablons. As you point out, this was not a randomized prospective study and I cannot definitely state whether metastasectomy is of benefit. These patients were admitted to the Surgery Branch of the National Cancer Institute with the diagnosis, usually by incisional biopsy, of a primary sarcoma. Before they entered this study they had a complete work-up for metastatic disease, and they were free of detectable metastatic disease as far as could tell with CT, LT, bone scans, etc., when they underwent primary sarcoma excision. They then were followed up in clinic with routine appointments every 3 months and CT of the chest every 3 to 6 months for surveillance. Patients who were included in these studies, then, were those in whom pulmonary metastases initially developed. The conclusions we draw are from this cohort, not a true randomized group of patients. Dr. Mitchell. Certainly we see some patients, even with aggressive tumors and metastasis, who live in some sort of symbiotic relationship with their tumor. Can we justify this report without either concurrent controls or historical controls? Dr. Jablons. The important point is this: Is it of value to submit these patients to thoracotomy for what questionable gain? Certainly individuals in the community of general thoracic surgeons may have a difficult time accepting this. However, we believe that with aggressive surgical follow-up and management of these patients, we can significantly increase their life spans-significant in that the life span for historicalcontrol patients who have pulmonary metastases and are not treated is approximately 12 months. The median survival time was 26.8 months in our group of patients who were aggressively followed up and subjected to repeat thoracotomies as the disease necessitated. The patients in our study who could not be rendered free of disease or, in other words, who had unresectable disease at primary initial thoracotomy, had a median survival of 9 months. Thus the historical controls will tell you 9 to 12 months. We are also reporting a 35% 3-year survival rate as opposed to a 12-month survival rate (historical control) without pulmonary metastasectomy. My only other comment is that the morbidity is very low-less than 80/0--as you very aptly pointed out. The average stay in the hospital, and this probably reflects the younger age of these patients (median 38 years of age), was 5

Pulmonary metastases from soft tissue sarcoma

705

days. Patients are getting back to work after thoracotomy, and postoperative complications and pain have been minimal. We believe that their quality and extension of life for that, albeit short, 14-month difference is of benefit and is worthwhile. Dr. Richard Fosburg (La Jolla. Calif). Dr. Jablons, I have two questions for you. The first relates to whether or not your review of this information underscores the fact that observation at 3-month intervals has any bearing on outcome. Should we examine them more frequently. Dr. Jablons. That is a critical point. Our most sensitive surveillance technique at this time is CT of the chest, because in this select patient population with high-grade soft tissue sarcoma, the primary site of recurrence will be in the chest. We at the surgery branch examine patients every 3 months in clinic with 3- to 6-month interval CT scanning of the chest for the first 5 years. The other point that reinforces the fact that disease-free interval is important is that, in most patients whose sarcoma recurs, the disease will recur within the first 2 years. Thus this is the most critical time for enhanced surveillance for pulmonary metastases. Dr. Fosburg. My second question relates to the mode of death of those who died. Where was the preponderance of the tumor burden in these patients? Dr. Jablons. The majority of patients who ultimately died with their disease had massive pulmonary disease, including malignant pleural effusions, and for the most part died of pulmonary and respiratory insufficiency. Dr. Lucius D. Hill (Seattle. Wash.). The Lung Cancer Study Group did a study that showed the same thing you have shown here: that radiation alone for lung cancer did not extend life. Chemotherapy also did not extend life. However, preoperative radio-chemotherapy obliterated unresectable tumors to the point that 14% of the patients came to surgery with no microscopic residual. Have you tried combined radio-chemotherapy in these patients? We don't know what the life span will be because this approach is still too new, but we have never seen such eradication of what initially appeared to be inoperable tumors by the combination of radio-chemotherapy. Dr. Jablons. Again, to answer that question solidly would require a randomized prospective trial. We have not done that in the surgery branch, nor do we purport to report that here. For sarcoma the only effective chemotherapies are doxorubicin-containing regimens. These patients who were offered postthoracotomy chemotherapy, and who had never received chemotherapy because of their initial randomization after the primary sarcoma was excised, did not benefit in survival from posthoracotomy chemotherapy. The temptation was great to say this patient had never been given chemotherapy and might benefit from it. I believe this is actually one of the more significant findings in this retrospective review. It highlights the fact that, although the numbers are small (only 20 patients roughly in each arm), once a patient has a recurrence in the chest, subsequent chemotherapy is probably not going to be effective. Dr. Hill. But you haven't tried the combined radiochemotherapy? Dr. Jablons. That is correct. We have not. The radiation therapy mentioned in this paper pertains only to local control and is radiation solely to the site of the primary sarcoma.