Status of Surgical Margins and Prognosis in Adult Soft Tissue Sarcomas of the Extremities: A Series of Patients Treated at a Single Institut
http://www.100md.com
《临床肿瘤学》
the Department of Surgery, Department of Cancer Medicine, Department of Biostatistics, Department of Pathology, and Department of Diagnostic Imaging and Radiotherapy, Istituto Nazionale per lo Studio e la Cura dei Tumori, Milano, Italy
ABSTRACT
PATIENTS AND METHODS: We analyzed 911 consecutive patients surgically treated throughout a 20-year span at a single referral center. Six hundred forty-two were first seen with a primary tumor, and 269, with a locally recurrent tumor. All patients underwent macroscopically complete resection. Microscopic marginal status was negative (tumor size > 1 mm) in 748 patients and positive (≤ 1 mm) in 163 patients. Median follow-up was 107 months.
RESULTS: Patients with primary disease had a lower disease-specific mortality in comparison to those first examined for recurrence (25% v 37%, respectively, at 10 years). Size, malignancy grade, depth, histotype, and local recurrence had a statistically significant prognostic effect at multivariable analysis, while microscopically positive surgical margins had not, though a trend in favor of negative margins was observed. However, an extra risk was observed for patients with positive margins after 3 to 5 years (hazard ratio, 1.8 after 5 years v 0.8 before 5 years). In patients treated for a local recurrence, the prognostic impact of positive margins was higher (hazard ratio, 1.6).
CONCLUSION: Positive surgical margins had a weak adverse prognostic effect, which was more pronounced for those patients escaping an early relapse. This would seem to justify a policy of surgical adequacy in adult soft tissue sarcomas, though clinical decision making in borderline presentations for conservative surgery might be reasonably flexible and shared with the patient. Once a local relapse has occurred, the impact of local treatments seems more critical.
INTRODUCTION
A correct understanding of the prognostic effect of the quality of local treatment (that is, the status of resection margins) and local recurrence is essential. If local recurrence is merely a problem of local control that does not affect overall survival, the performance of closer surgical margins with a consequent lower loss of function may be advantageous. However, if local recurrence has by itself a substantial influence on survival, wider surgical margins would be indicated, and larger sequelae accepted.
In this study, we decided to retrospectively assess the impact of surgical margins on disease outcome in a consecutive case series of patients treated at our institution throughout a 20-year time span.
PATIENTS AND METHODS
this series, we excluded 11 patients who could not undergo a complete surgical excision and 91 patients affected by dermatofibrosarcoma protuberans, as this tumor rarely metastasizes. Thus, our sample includes 911 patients, whose main characteristics are summarized in Table 1 and detailed throughout this section.
Median age at the time of referral to our institution was 50 years (range, 16 to 90 years). The male-female ratio was approximately 1:1, with a slight male predominance. Median size of the primary tumor was 6 cm. (interquartile range, 4 to 10 cm.). Tumors were located in the lower extremities in 540 patients (393 to the thigh and 147 to the leg-foot), in the upper extremities in 155 patients (74 to the arm and 81 to the forearm-hand), and in the girdles in 216 patients (scapular girdle in 108 and pelvic girdle in 108). Thus, the lower limbs (including the girdle) accounted for approximately 70% of all the tumors. Six hundred forty-two patients presented with primary disease, while 269 had a locoregional recurrent tumor. One hundred sixty-five patients had superficial lesions (135 in primary and 30 in recurrences), while 746 had deeply located tumors (507 in primary and 239 in recurrences). As far as the histotype are concerned, during the study period, all tumors were reviewed by at least two experienced pathologists at our institution. For statistical purposes, histotypes were grouped as follows: liposarcomas, synovial sarcomas, leiomyosarcomas, malignant peripheral nerve sheat tumors (MPNSTs), malignant fibrous histiocytomas (MFH), vascular sarcomas, and other sarcomas. In grading assignment, the French Federation of Cancer Centers (FNCLCC) grading system3 was applied to the untreated primary tumors. In cases of recurrent tumor, the grading was performed on the slides from the primary untreated tumor seen in consultation. In this case series, all types mentioned above were represented, with a slight prevalence of liposarcomas (31%). Tumors were graded as grade 1 in 255 cases (28%), grade 2 in 226 cases (25%), and grade 3 in 430 cases (47%).
A total of 87 major exarticulations (30 major amputation, 30 disarticulation, and 27 forequarter amputations) were performed—50 at presentation at our center (33 in primary cases and 21 in recurrences) and 37 for further recurrences. All surgical resections were considered macroscopically complete, which was defined as being in the absence of gross residual disease after surgical excision of the tumor. Margins were defined at the time of the pathological assessment by a dedicated pathologist. The surgical specimen was always examined in the presence of the operating surgeon. The margins were inked and separately sampled. The closest margin was microscopically categorized as positive (tumor within 1 mm from the inked surface) or negative (absence of tumor within 1 mm from the inked surface). Therefore, resection margins were classified as negative in 748 cases (82%; 556 in primary tumors and 192 in recurrences) and positive in 163 cases (86 in primary tumors and 77 in recurrences).
Radiation therapy was delivered as an adjunct in 340 cases (37%; 37 among primary presentations and 103 among recurrences). The indication to radiation therapy was given by both the operating surgeon and the radiation oncologist when a higher risk of recurrence was supposed to exist on clinical grounds. However, no prospectively selected criteria were used to this end. In particular, radiation therapy was administered to 92 patients (56%) with positive tumor margins as compared with 248 patients (33%) with negative tumor margins. External beam radiation was used in all such cases, and doses ranged from 45 to 65 Gy (median, 57 Gy).
Chemotherapy was given to 179 patients (20%; 114 primary cases and 65 recurrences) at the discretion of the multidisciplinary Soft Tissue Sarcoma Group of our institution or as part of clinical trials. Antracycline-based regimens were used in most of the cases associated with ifosfamide.
As of June 2003, the median follow-up duration of the entire group was 107 months (interquartile range, 58 to 130). Forty-six patients (5%) were lost to follow-up before the 10th year.
The end points of this study were cause-specific mortality, local recurrence, and distant metastasis. Time to occurrence of any event was computed from the date of surgery at our institution to the date when the event was first recorded, or censored at the date of last follow-up assessment in event-free subjects. These end points were analyzed in the whole series and separately for primary or recurrent patients.
Crude cumulative incidence curves for each end point were calculated in a competing risks framework,21 and comparisons between curves in different subgroups were carried out by means of the Gray test.22 In the analysis of cause-specific mortality, deaths due to conditions unrelated with sarcoma were regarded as competing events. For local recurrence (distant metastasis) analysis, deaths without evidence of disease and distant metastasis (local recurrence), whichever occurred first, were regarded as competing events.
Multivariable analyses of each end point were based on cause-specific hazards and were therefore carried out using Cox multiple regression models. The analyses mainly focused on the prognostic effect of microscopic margin status. The following covariates, chosen before statistical analysis on the ground of literature information,4-7,10 were included in the models for the purpose of adjustment: age, presentation (primary, recurrent), size, depth (superficial, deep), FNCLCC grade (I, II, III), location (proximal, distal), histotype (liposarcoma, synovial sarcoma, leiomyosarcoma, MPNST, MFH, vascular sarcoma, other sarcoma), and radiotherapy (yes, no). An exploratory analysis showed a prognostic effect of tumor size only among deeply located tumors. Consequently, the interaction term size x depth, but not the main effect of size, was entered into the models. Age and tumor size were modeled as continuous variables by using four-knot restricted cubic splines,23 whereas the other covariates were modeled as categoric by using dummy (0/1) variables. To investigate the association between local relapse and cause-specific mortality, a Cox model was fitted with the inclusion of local relapse as an additional covariate. In such analysis, death-times were left-truncated at the time of local relapse occurrence. All of the above analyses were repeated with the inclusion of type of surgery, chemotherapy, and the year of surgery (to investigate possible period effects). As the estimates for the remaining covariates were negligibly affected, we only report the results of simpler models.
To check the proportional hazard assumption implied by the Cox model, we relied on the graphical analysis of scaled Schoenfeld residuals,24 and on testing time-dependent covariates when deviations from the assumption were suspected.
We used SAS software (SAS Institute Inc, Cary, NC) and the S-Plus (StatSci; MathSoft, Seattle, WA) Design and Cmprsk (Harvard University, Cambridge, MA) libraries to perform the modeling and statistical calculations.25 We considered as significant two-sided P values below the 5% conventional threshold.
RESULTS
Five and 10-year mortality estimates were 0.24 (95% CI, 0.21 to 0.27) and 0.29 (95% CI, 0.26 to 0.32) in the whole series, 0.22 (95% CI, 0.18 to 0.25) and 0.25 (95% CI, 0.21 to 0.29) in the primary cases, and 0.28 (95% CI, 0.23 to 0.34) and 0.37 (95% CI, 0.31 to 0.44) in recurrences. This difference in outcome according to phase (primary versus recurrence) at presentation was statistically significant (P < .001).
In Figure 1, crude cumulative mortality curves are represented according to microscopic margin status. For primary cases (Fig 1A) 10-year estimates were 0.31 (95% CI, 0.20 to 0.42) in positive-margin cases and 0.24 (95% CI, 0.20 to 0.28) in negative-margin cases (P = .338). The corresponding figures in recurrent cases (Fig 1B) were 0.47 (95% CI, 0.34 to 0.59) and 0.34 (95% CI, 0.26 to 0.41); P = .019.
Table 2 presents the results from the Cox multiple regression model for the whole case series. Significant prognostic factors for survival were tumor depth, size, presentation, FNCLCC grade, histotype, and radiotherapy. Notably, microscopic margin status failed to reach statistical significance (P = .237); the hazard ratio estimate for positive- versus negative-margin tumors was 1.2 (95% CI, 0.9 to 1.7). The Cox model, including local recurrence as a predictor yielded for the latter significant results (P < .001), with a hazard ratio estimate of 3.4 (95% CI, 2.6 to 4.6). In this model, the hazard ratio for microscopic margin status was closer to one than in the former analysis, namely 1.1 (95% CI, 0.8 to 1.6); P = .443.
The Cox models fitted in the subset of patients with primary or recurrent tumors (not shown in detail) yielded hazard ratios for the microscopic margin status of 0.8 (95% CI, 0.5 to 1.3; P = .423) and 1.6 (95% CI, 1.0 to 2.5; P = .034), respectively.
Local Recurrence
Two hundred twenty-six patients had local recurrence after surgery performed at our institute (129 with primary disease and 97 with local recurrence at presentation). One hundred fifty patients had only one local recurrence, while 76 had two or more local recurrences. Time to first local recurrence varied from 1 to 201 months, with the median time for those who recurred being 15 months. One hundred three patients had only local recurrence, while 123 had local recurrence and distant metastasis (the local recurrence anticipating the distant metastasis in 75 patients, being concurrent in 32 and posterior in 16).
One hundred seventy-eight local recurrences occurred as first events, of which there were 101 among primary tumors. Five-year and 10-year estimates were 0.17 (95% CI, 0.15 to 0.20) and 0.19 (95% CI, 0.17 to 0.22) in the whole series, 0.14 (95% CI, 0.11 to 0.17) and 0.15 (95% CI, 0.13 to 0.18) in the primary cases, and 0.25 (95% CI, 0.20 to 0.30) and 0.28 (95% CI, 0.23 to 0.34) in recurrences. This difference in local recurrence according to phase (primary v recurrence) at presentation was statistically significant (P < .001).
In Figure 2, crude cumulative incidence curves are represented according to microscopic margin status. For primary cases (Fig 2A), 10-year estimates were 0.29 (95% CI, 0.19 to 0.39) in positive-margin cases and 0.13 (95% CI, 0.10 to 0.16) in negative-margin cases (P < .001). The corresponding figures in recurrent cases (Fig 2B) were 0.34 (95% CI, 0.23 to 0.44) and 0.26 (95% CI, 0.20 to 0.32); P = .222.
Table 3 shows the results from the Cox multiple regression model on the whole case series. Significant prognostic factors for local relapse were tumor presentation, microscopic margin status, and radiotherapy, whereas borderline P values were obtained on age and depth. The hazard ratio for positive versus negative-margin tumors was 1.9 (95% CI, 1.4 to 2.7), roughly corresponding to a doubling in the risk of local recurrence for margin-positive tumors.
The Cox model fitted in the subset of patients with primary tumors (not shown in detail) roughly confirmed the above findings for clearly significant factors, while among age and depth, only the first became significant (P = .040). Hazard ratio for the microscopic margin status was 2.3 (95% CI, 1.4 to 3.8; P = .002). In the recurrent patient subset, only age was statistically significant (P = .046), while microscopic margin status achieved a borderline P value of .059, with a hazard ratio of 1.6 (95% CI, 1.0 to 2.6).
Distant Metastases
Two hundred seventy-nine patients had distant metastasis after surgery performed at our center (176 with primary disease and 103 with local recurrence at presentation). One hundred eighty-two patients had only pulmonary metastasis, 58 also developed extrapulmonary metastasis (11 patients before pulmonary metastasis, 28 concurrently to the lung, and 19 after), and 39 had only extrapulmonary metastasis. The site of extrapulmonary metastasis were the regional lymph nodes in 20 cases (14 of these died of disease), the skeleton in 18 cases, the soft tissues in 17 cases, the abdomen in 12 cases, the liver in nine cases, the brain in one case, and multiple organs in the remaining 20 cases. Time to distant metastasis varied from 1 to 213 months, with the median time for those who developed distant metastasis being 14 months. One hundred fifty-six patients had only distant metastasis, while 123 had also local recurrence, as reported above.
Two hundred four distant metastases occurred as first event (136 of these among primary tumors). Five and 10-year estimates were 0.21 (95% CI, 0.19 to 0.24) and 0.22 (95% CI, 0.20 to 0.25) in the whole series, 0.20 (95% CI, 0.17 to 0.23) and 0.21 (95% CI, 0.18 to 0.24) in the primary cases, and 0.25 (95% CI, 0.19 to 0.30) and 0.25 (95% CI, 0.20 to 0.30) in recurrences. This difference in distant metastasis according to phase (primary v recurrence) at presentation was not statistically significant (P = .174).
In Figure 3, crude cumulative incidence curves are represented according to microscopic margin status. For primary cases (Fig 3A), 10-year estimates were 0.23 (95% CI, 0.14 to 0.32) in positive-margin cases and 0.21 (95% CI, 0.17 to 0.24) in negative-margin cases (P = .450). The corresponding figures in recurrent cases (Fig 3B) were 0.31 (95% CI, 0.21 to 0.42) and 0.23 (95% CI, 0.17 to 0.29; P = .124).
Table 4 presents the results from the Cox multiple regression model for the whole case series. Significant prognostic factors for distant metastasis were tumor depth, size, FNCLCC grade, and histotype. The hazard ratio for positive- versus negative-margin tumors was 1.1 (95% CI, 0.8-1.6; P = .495), a figure quite similar to that obtained for mortality.
Results with the Cox model fitted in the subset of patients with primary and recurrent tumors (not shown in detail) overlapped those previously shown for the whole case series as regards the prognostic effect of covariates. In the primary tumor subset, the hazard ratio of microscopic margin status was 0.8 (95% CI, 0.5 to 1.3; P = .416); in the recurrent tumor subset, the hazard ratio was 1.5 (95% CI, 0.9 to 2.5; P = .136).
DISCUSSION
As far as primary tumors are concerned, patients with positive margins behave like those with negative margins in the first 3 to 5 years of follow-up, but in the subsequent interval, the difference in favor of negative margins becomes more striking (hazard ratio, 1.8 after 5 years, compared with 0.8 in the first 5 years), though still not statistically significant (P = .179). This implies a time-dependent relationship between quality of surgery and final outcome, by which poor surgical margins affect survival only after a time interval, as though only patients who escape a systemic risk may benefit, or suffer, from quality of surgery. A similar observation was made by Stojadinovic et al9,15 in a series from Memorial Sloan-Kettering Cancer Center. Indeed, this series included all sites, while our series is made up of extremity cases only. Obviously, one may believe that causes of systemic failure connected to the inherent aggressiveness of the disease prevail in the early years, while the way in which the disease has been treated locally becomes more important afterwards, when inherent aggressiveness has already produced its effects. If so, quality of surgery might be important for the proportion of patients who survive the inherent aggressiveness of disease. Indeed, patients with positive margins do not represent more than 20% of all surgically treated patients. Therefore, the lack of statistical significance in our series is likely to simply depend on the relatively small numbers. That said, quality of surgery might make the difference after 3 to 5 years for this limited subset of patients, being overshadowed earlier by more important prognostic factors. This may possibly create understanding as to why so often, the independent prognostic value of quality of surgical margins could not be objectively appreciated when analyzing large case series, though it nonetheless seems universally important to the sarcoma surgeon.
Overall, however, while positive surgical margins adversely affected the local outcome, as shown both at univariable and multivariable analyses, distant metastasis and cause-specific mortality were only slightly different between the two groups of patients resected for a primary tumor with negative and positive margins, and this difference was not statistically significant at multivariable analysis. In other words, positive surgical margins may not be an independent factor, simply reflecting other problems inherent to the disease (eg, size, depth). Therefore, the same adverse prognostic factors would affect quality of surgery, local recurrence, and distant metastases.
In published literature, most studies failed to find a strong correlation between quality of surgery and final outcome. In both a series of 559 patients treated within the Scandinavian Sarcoma Group centers,14 and another series of 848 patients treated at The University of Texas M.D. Anderson Cancer Center,16 what was relevant for distant metastasis, and therefore disease-specific survival, was the intrinsic biologic aggressiveness of the tumor, as defined by size, depth, malignancy grade, and histotype.4-7,11,17 In this sense, local recurrence might be well regarded as an additional biologic marker of aggressiveness (ie, a result rather than a cause). Under this interpretation, when a local recurrence is detected, the patient should be viewed as being at a higher risk of death due to factors inherent to his disease. Also, Lewis et al26, in a series of 911 patients treated at Memorial Sloan-Kettering Cancer Center, found local that recurrence correlated with the development of subsequent metastases and tumor mortality, but concluded that local recurrence is not a source of metastasis, but rather a biologic phenomenon on the continuum of recurrent sarcoma. Some other smaller series showed that improved local control rates observed in patients with adequate surgical margins failed to translate into a survival benefit.27-31 Furthermore, in the three published randomized trials18-20 that compared more and less extensive local treatment modalities (amputation v conservative surgery + radiation therapy, or conservative surgery v conservative surgery + adjuvant radiation), differences in local outcome did not translate into differences in survival. The power of these randomized studies, however, is limited by the small sample size; therefore, no definitive conclusions can be drawn.
An important point should be taken into account when interpreting the lack of correlation between quality of surgical margins and final outcome. A strict application of Enneking criteria32 is precluded when analyzing large retrospective series, and the only possibility is to codify margins as either positive or negative. Indeed, the positive or negative status of the closest surgical margin is clearly a rough indicator of quality of surgery. Actually, some marginal excisions may well be included in the negative-margins group, thus contributing to confounding conclusions. Of course, the anatomic heterogeneity of adult STSs adds to these difficulties. It is probable that innovative definitions of quality of surgical margins (eg, based on width and type of surrounding healthy tissues, type of tumor margins, histotype, and grade) may help settle this highly controversial issue, and prospective clinical studies aimed at working out new definitions of adequacy of surgery would be welcome at this stage.
By contrast, in patients with an already locally recurrent tumor, the effect of surgical margins was more striking in our series. A distinct impact on metastasis-free and overall survival, as well as local control, was observed. One may believe that this patient population is selected in such a way that the local control is more important for them and may even have a prognostic impact. Clearly, however, patients in whom negative margins were not reached will have had worse prognostic factors locally. Therefore, we cannot conclude that a more aggressive local policy would have changed the final outcome. At any rate, it was just in this patient subgroup that receiving radiation therapy or not had prognostic relevance (and the choice to do radiation therapy or not, in this retrospective series, may have been less dependant on the local aggressiveness of disease). In brief, the hypothesis is that local recurrence defines a subgroup of patients for whom more aggressive local treatment might be worthwhile, though final proof thereof cannot be provided by this retrospective analysis.
In conclusion, microscopic residual disease may have an independent effect on disease-specific mortality, but this effect is minor. Indeed, a small proportion of patients (ie, those escaping early relapse due to inherent aggressiveness of disease) may suffer from inadequate surgery more clearly. Therefore, there are still good reasons to continue to pursue appropriate surgery in adult STSs (ie, clean margins at the very least; more sophisticated classifications of adequacy of margins would be welcome). On the other hand, when the clinical presentation is borderline; for example, when obtaining clean margins would require an amputation or major functional impairments, decision making may be reasonably flexible and shared with the patient, considering that his or her survival will directly depend only to a limited degree on quality of surgery itself. This may not apply to patients who have already had a local relapse. These patients seem to be selected in such a way that indicates that quality of margins is more important for them. In this subgroup of patients, therefore, to our knowledge, an attempt to reach an adequate local control might be especially crucial.
Authors’ Disclosures of Potential Conflicts of Interest
NOTES
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Brennan M, Alektiar KM, Maki RG: Soft tissue sarcoma, in DeVita VT, Hellmann S, Rosenberg SA (eds): Cancer: Principles and Practice of Oncology. Philadelphia, PA, Lippincott Williams & Wilkins, 2001, pp 1841-1891
2. Greenlee RT, Hill-Harmon Mb, Murray T, et al: Cancer statistics, 2001. CA Cancer J Clin 51:15-36, 2001
3. Trojani M, Contesso G, Coindre JM, et al: Soft tissue sarcomas of adults: Study of pathological prognostic variables and definition of a histopathologic grading system. Int J Cancer 33:37-42, 1984
4. Gustafson P: Soft tissue sarcoma epidemiology and prognosis in 508 patients. Acta Orthop Scand 65:1-31, 1994 (suppl 259)
5. Singer S, Corson K, Gonin JM, et al: Prognostic factors predictive of survival and local recurrence for extremity soft tissue sarcoma. Ann Surg 219:165-173, 1994
6. Pisters PWT, Leung DHY, Woodruff J, et al: Analysis of prognostic factors in 1041 patients with localized soft tissue sarcoma of th extremities. J Clin Oncol 14:1679-1689, 1996
7. Coindre JM, Terrier P, Bui NB, et al: Prognostic factors in adult patients with locally controlled soft tissue sarcoma: A study on 546 patients from the French Federation of Cancer Centers Sarcoma Group. J Clin Oncol 14:869-877, 1996
8. Coindre JM, Terrier P, Guillou L, et al: Predictive value of grade for metastasis development in the main histologic types of adult soft tissue sarcomas: A study of 1240 patients from the French Federation of Cancer Centers Sarcoma Group. Cancer 91:1914-1926, 2001
9. Stojadinovic A, Leung DHY, Allen P, et al: Primary adult soft tissue sarcoma: Time-dependent influence of progostic variables. J Clin Oncol 20:4344-4352, 2002
10. Weitz J, Antonescu CR, Brennan MF: Localized extremity soft tissue sarcoma: Improved knowledge with unchanged survival over time. J Clin Oncol 21:2719-2725, 2003
11. Karakousis CP, Proimakis C, Rao U, et al: Local recurrence and survival in soft tissue sarcomas. Ann Surg Oncol 3:255-260, 1996
12. Brennan MF: The enigma of local recurrence. Ann Surg Oncol 4:1-12, 1997
13. Lewis JJ, Leung DHY, Casper ES, et al: Multifactorial analysis of long-term follow-up (more than 5 years) of primary extremit sarcoma. Arch Surg 134:190-194, 1999
14. Trovik CS, Bauer HCF, Alvegard TA, et al: Surgical margins, local recurrence and metastasis in soft tissue sarcomas: 559 surgically-treated patients from the Scandinavian Sarcoma Group register. Eur J Cancer 36:710-716, 2000
15. Stojadinovic A, Leung DHY, Hoos A, et al: Analysis of the prognostic significance of microscopic margins in 2084 localized primary adult soft tissue sarcomas. Ann Surg 235:424-433, 2002
16. Zagars GK, Ballo MT, Pisters PWT, et al: Prognostic factors for patients with localized soft tissue sarcoma treated with conservation surgery and radiation therapy: An analysis of 1225 patients. Cancer 97:2530-2543, 2003
17. Eilber FC, Rosen G, Nelson SD, et al: High grade extremity soft tissue sarcomas: Factor predictive of local recurrence and its effect on morbidity and mortality. Ann Surg 237:218-226, 2003
18. Rosenberg SA, Tepper J, Glatsein E, et al: The treatment of soft tissue sarcoma of the extremities: Prospective randomized evaluation of (1) limb-sparing surgery plus radiation therapy compared with amputation and (2) the role of adjuvant chemotherapy. Ann Surg 196:305-315, 1982
19. Pisters PWT, Harrison LB, Leung DHY, et al: Long-term results of a prospective randomized trial of adjuvant brachitherapy in soft tissue sarcoma. J Clin Oncol 14:859-868, 1996
20. Yang JC, Chang AE, Baker AR, et al: Randomized prospective study of the benefit of adjuvant radiation therapy in the treatment of soft tissue sarcomas of the extremity. J Clin Oncol 16:197-203, 1998
21. Marubini E, Valsecchi MG: Analysing survival data for clinical trials and observational studies. Chichester, UK, John Wiley and Sons, 1995
22. Gray RJ: A class of k-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 16:1141-1154, 1988
23. Durrleman S, Simon R: Flexible regression models with cubic splines. Stat Med 8:551-561, 1989
24. Schoenfeld D: Partial residuals for the proportional hazards regression model. Biometrika 69:239-241, 1982
25. SAS Institute Inc: SAS/STAT Software: Changes and Enhancements Through Release 6.11. Cary, NC, SAS Intitute Inc, 2000
26. Lewis JJ, Leung DHY, Heslin M, et al: Association of local recurrence with subsequent survival in extremity soft tissue sarcoma. J Clin Oncol 15:646-652, 1997
27. Ueda T, Yoshikawa H, Mori S, et al: Influence of local recurrence on the prognosis of soft tissue sarcomas. J Bone Joint Surg 79B:553-557, 1997
28. Tanabe KK, Pollock RE, Ellis LM, et al: Influence of surgical margins on outcome in patients with preoperatively irradiated extremity soft tissue sarcomas. Cancer 73:1652-1659, 1994
29. Li XQ, Parkekh SG, Rosenberg AE, et al: Assessing prognosis for high grade soft tissue sarcomas: Search for a marker. Ann Surg Oncol 3:550-557, 1996
30. Gustafson P, Dreinhofer KE, Rydholm A: Metastasis-free survival after local recurrence of soft tissue sarcoma. J Bone Joint Surg 75B:658-660, 1993
31. Sadoski C, Suit HD, Rosenberg A, et al: Preoperative radiation, surgical margins, and local control of extremity sarcomas of soft tissue. J Surg Oncol 52:223-230, 1993
32. Enneking WP, Spainer SS, Goodman MA: A system for the surgical staging of musculoskeletal sarcoma. Clin Orthop 153:106-120, 1980(A. Gronchi, P.G. Casali, )
ABSTRACT
PATIENTS AND METHODS: We analyzed 911 consecutive patients surgically treated throughout a 20-year span at a single referral center. Six hundred forty-two were first seen with a primary tumor, and 269, with a locally recurrent tumor. All patients underwent macroscopically complete resection. Microscopic marginal status was negative (tumor size > 1 mm) in 748 patients and positive (≤ 1 mm) in 163 patients. Median follow-up was 107 months.
RESULTS: Patients with primary disease had a lower disease-specific mortality in comparison to those first examined for recurrence (25% v 37%, respectively, at 10 years). Size, malignancy grade, depth, histotype, and local recurrence had a statistically significant prognostic effect at multivariable analysis, while microscopically positive surgical margins had not, though a trend in favor of negative margins was observed. However, an extra risk was observed for patients with positive margins after 3 to 5 years (hazard ratio, 1.8 after 5 years v 0.8 before 5 years). In patients treated for a local recurrence, the prognostic impact of positive margins was higher (hazard ratio, 1.6).
CONCLUSION: Positive surgical margins had a weak adverse prognostic effect, which was more pronounced for those patients escaping an early relapse. This would seem to justify a policy of surgical adequacy in adult soft tissue sarcomas, though clinical decision making in borderline presentations for conservative surgery might be reasonably flexible and shared with the patient. Once a local relapse has occurred, the impact of local treatments seems more critical.
INTRODUCTION
A correct understanding of the prognostic effect of the quality of local treatment (that is, the status of resection margins) and local recurrence is essential. If local recurrence is merely a problem of local control that does not affect overall survival, the performance of closer surgical margins with a consequent lower loss of function may be advantageous. However, if local recurrence has by itself a substantial influence on survival, wider surgical margins would be indicated, and larger sequelae accepted.
In this study, we decided to retrospectively assess the impact of surgical margins on disease outcome in a consecutive case series of patients treated at our institution throughout a 20-year time span.
PATIENTS AND METHODS
this series, we excluded 11 patients who could not undergo a complete surgical excision and 91 patients affected by dermatofibrosarcoma protuberans, as this tumor rarely metastasizes. Thus, our sample includes 911 patients, whose main characteristics are summarized in Table 1 and detailed throughout this section.
Median age at the time of referral to our institution was 50 years (range, 16 to 90 years). The male-female ratio was approximately 1:1, with a slight male predominance. Median size of the primary tumor was 6 cm. (interquartile range, 4 to 10 cm.). Tumors were located in the lower extremities in 540 patients (393 to the thigh and 147 to the leg-foot), in the upper extremities in 155 patients (74 to the arm and 81 to the forearm-hand), and in the girdles in 216 patients (scapular girdle in 108 and pelvic girdle in 108). Thus, the lower limbs (including the girdle) accounted for approximately 70% of all the tumors. Six hundred forty-two patients presented with primary disease, while 269 had a locoregional recurrent tumor. One hundred sixty-five patients had superficial lesions (135 in primary and 30 in recurrences), while 746 had deeply located tumors (507 in primary and 239 in recurrences). As far as the histotype are concerned, during the study period, all tumors were reviewed by at least two experienced pathologists at our institution. For statistical purposes, histotypes were grouped as follows: liposarcomas, synovial sarcomas, leiomyosarcomas, malignant peripheral nerve sheat tumors (MPNSTs), malignant fibrous histiocytomas (MFH), vascular sarcomas, and other sarcomas. In grading assignment, the French Federation of Cancer Centers (FNCLCC) grading system3 was applied to the untreated primary tumors. In cases of recurrent tumor, the grading was performed on the slides from the primary untreated tumor seen in consultation. In this case series, all types mentioned above were represented, with a slight prevalence of liposarcomas (31%). Tumors were graded as grade 1 in 255 cases (28%), grade 2 in 226 cases (25%), and grade 3 in 430 cases (47%).
A total of 87 major exarticulations (30 major amputation, 30 disarticulation, and 27 forequarter amputations) were performed—50 at presentation at our center (33 in primary cases and 21 in recurrences) and 37 for further recurrences. All surgical resections were considered macroscopically complete, which was defined as being in the absence of gross residual disease after surgical excision of the tumor. Margins were defined at the time of the pathological assessment by a dedicated pathologist. The surgical specimen was always examined in the presence of the operating surgeon. The margins were inked and separately sampled. The closest margin was microscopically categorized as positive (tumor within 1 mm from the inked surface) or negative (absence of tumor within 1 mm from the inked surface). Therefore, resection margins were classified as negative in 748 cases (82%; 556 in primary tumors and 192 in recurrences) and positive in 163 cases (86 in primary tumors and 77 in recurrences).
Radiation therapy was delivered as an adjunct in 340 cases (37%; 37 among primary presentations and 103 among recurrences). The indication to radiation therapy was given by both the operating surgeon and the radiation oncologist when a higher risk of recurrence was supposed to exist on clinical grounds. However, no prospectively selected criteria were used to this end. In particular, radiation therapy was administered to 92 patients (56%) with positive tumor margins as compared with 248 patients (33%) with negative tumor margins. External beam radiation was used in all such cases, and doses ranged from 45 to 65 Gy (median, 57 Gy).
Chemotherapy was given to 179 patients (20%; 114 primary cases and 65 recurrences) at the discretion of the multidisciplinary Soft Tissue Sarcoma Group of our institution or as part of clinical trials. Antracycline-based regimens were used in most of the cases associated with ifosfamide.
As of June 2003, the median follow-up duration of the entire group was 107 months (interquartile range, 58 to 130). Forty-six patients (5%) were lost to follow-up before the 10th year.
The end points of this study were cause-specific mortality, local recurrence, and distant metastasis. Time to occurrence of any event was computed from the date of surgery at our institution to the date when the event was first recorded, or censored at the date of last follow-up assessment in event-free subjects. These end points were analyzed in the whole series and separately for primary or recurrent patients.
Crude cumulative incidence curves for each end point were calculated in a competing risks framework,21 and comparisons between curves in different subgroups were carried out by means of the Gray test.22 In the analysis of cause-specific mortality, deaths due to conditions unrelated with sarcoma were regarded as competing events. For local recurrence (distant metastasis) analysis, deaths without evidence of disease and distant metastasis (local recurrence), whichever occurred first, were regarded as competing events.
Multivariable analyses of each end point were based on cause-specific hazards and were therefore carried out using Cox multiple regression models. The analyses mainly focused on the prognostic effect of microscopic margin status. The following covariates, chosen before statistical analysis on the ground of literature information,4-7,10 were included in the models for the purpose of adjustment: age, presentation (primary, recurrent), size, depth (superficial, deep), FNCLCC grade (I, II, III), location (proximal, distal), histotype (liposarcoma, synovial sarcoma, leiomyosarcoma, MPNST, MFH, vascular sarcoma, other sarcoma), and radiotherapy (yes, no). An exploratory analysis showed a prognostic effect of tumor size only among deeply located tumors. Consequently, the interaction term size x depth, but not the main effect of size, was entered into the models. Age and tumor size were modeled as continuous variables by using four-knot restricted cubic splines,23 whereas the other covariates were modeled as categoric by using dummy (0/1) variables. To investigate the association between local relapse and cause-specific mortality, a Cox model was fitted with the inclusion of local relapse as an additional covariate. In such analysis, death-times were left-truncated at the time of local relapse occurrence. All of the above analyses were repeated with the inclusion of type of surgery, chemotherapy, and the year of surgery (to investigate possible period effects). As the estimates for the remaining covariates were negligibly affected, we only report the results of simpler models.
To check the proportional hazard assumption implied by the Cox model, we relied on the graphical analysis of scaled Schoenfeld residuals,24 and on testing time-dependent covariates when deviations from the assumption were suspected.
We used SAS software (SAS Institute Inc, Cary, NC) and the S-Plus (StatSci; MathSoft, Seattle, WA) Design and Cmprsk (Harvard University, Cambridge, MA) libraries to perform the modeling and statistical calculations.25 We considered as significant two-sided P values below the 5% conventional threshold.
RESULTS
Five and 10-year mortality estimates were 0.24 (95% CI, 0.21 to 0.27) and 0.29 (95% CI, 0.26 to 0.32) in the whole series, 0.22 (95% CI, 0.18 to 0.25) and 0.25 (95% CI, 0.21 to 0.29) in the primary cases, and 0.28 (95% CI, 0.23 to 0.34) and 0.37 (95% CI, 0.31 to 0.44) in recurrences. This difference in outcome according to phase (primary versus recurrence) at presentation was statistically significant (P < .001).
In Figure 1, crude cumulative mortality curves are represented according to microscopic margin status. For primary cases (Fig 1A) 10-year estimates were 0.31 (95% CI, 0.20 to 0.42) in positive-margin cases and 0.24 (95% CI, 0.20 to 0.28) in negative-margin cases (P = .338). The corresponding figures in recurrent cases (Fig 1B) were 0.47 (95% CI, 0.34 to 0.59) and 0.34 (95% CI, 0.26 to 0.41); P = .019.
Table 2 presents the results from the Cox multiple regression model for the whole case series. Significant prognostic factors for survival were tumor depth, size, presentation, FNCLCC grade, histotype, and radiotherapy. Notably, microscopic margin status failed to reach statistical significance (P = .237); the hazard ratio estimate for positive- versus negative-margin tumors was 1.2 (95% CI, 0.9 to 1.7). The Cox model, including local recurrence as a predictor yielded for the latter significant results (P < .001), with a hazard ratio estimate of 3.4 (95% CI, 2.6 to 4.6). In this model, the hazard ratio for microscopic margin status was closer to one than in the former analysis, namely 1.1 (95% CI, 0.8 to 1.6); P = .443.
The Cox models fitted in the subset of patients with primary or recurrent tumors (not shown in detail) yielded hazard ratios for the microscopic margin status of 0.8 (95% CI, 0.5 to 1.3; P = .423) and 1.6 (95% CI, 1.0 to 2.5; P = .034), respectively.
Local Recurrence
Two hundred twenty-six patients had local recurrence after surgery performed at our institute (129 with primary disease and 97 with local recurrence at presentation). One hundred fifty patients had only one local recurrence, while 76 had two or more local recurrences. Time to first local recurrence varied from 1 to 201 months, with the median time for those who recurred being 15 months. One hundred three patients had only local recurrence, while 123 had local recurrence and distant metastasis (the local recurrence anticipating the distant metastasis in 75 patients, being concurrent in 32 and posterior in 16).
One hundred seventy-eight local recurrences occurred as first events, of which there were 101 among primary tumors. Five-year and 10-year estimates were 0.17 (95% CI, 0.15 to 0.20) and 0.19 (95% CI, 0.17 to 0.22) in the whole series, 0.14 (95% CI, 0.11 to 0.17) and 0.15 (95% CI, 0.13 to 0.18) in the primary cases, and 0.25 (95% CI, 0.20 to 0.30) and 0.28 (95% CI, 0.23 to 0.34) in recurrences. This difference in local recurrence according to phase (primary v recurrence) at presentation was statistically significant (P < .001).
In Figure 2, crude cumulative incidence curves are represented according to microscopic margin status. For primary cases (Fig 2A), 10-year estimates were 0.29 (95% CI, 0.19 to 0.39) in positive-margin cases and 0.13 (95% CI, 0.10 to 0.16) in negative-margin cases (P < .001). The corresponding figures in recurrent cases (Fig 2B) were 0.34 (95% CI, 0.23 to 0.44) and 0.26 (95% CI, 0.20 to 0.32); P = .222.
Table 3 shows the results from the Cox multiple regression model on the whole case series. Significant prognostic factors for local relapse were tumor presentation, microscopic margin status, and radiotherapy, whereas borderline P values were obtained on age and depth. The hazard ratio for positive versus negative-margin tumors was 1.9 (95% CI, 1.4 to 2.7), roughly corresponding to a doubling in the risk of local recurrence for margin-positive tumors.
The Cox model fitted in the subset of patients with primary tumors (not shown in detail) roughly confirmed the above findings for clearly significant factors, while among age and depth, only the first became significant (P = .040). Hazard ratio for the microscopic margin status was 2.3 (95% CI, 1.4 to 3.8; P = .002). In the recurrent patient subset, only age was statistically significant (P = .046), while microscopic margin status achieved a borderline P value of .059, with a hazard ratio of 1.6 (95% CI, 1.0 to 2.6).
Distant Metastases
Two hundred seventy-nine patients had distant metastasis after surgery performed at our center (176 with primary disease and 103 with local recurrence at presentation). One hundred eighty-two patients had only pulmonary metastasis, 58 also developed extrapulmonary metastasis (11 patients before pulmonary metastasis, 28 concurrently to the lung, and 19 after), and 39 had only extrapulmonary metastasis. The site of extrapulmonary metastasis were the regional lymph nodes in 20 cases (14 of these died of disease), the skeleton in 18 cases, the soft tissues in 17 cases, the abdomen in 12 cases, the liver in nine cases, the brain in one case, and multiple organs in the remaining 20 cases. Time to distant metastasis varied from 1 to 213 months, with the median time for those who developed distant metastasis being 14 months. One hundred fifty-six patients had only distant metastasis, while 123 had also local recurrence, as reported above.
Two hundred four distant metastases occurred as first event (136 of these among primary tumors). Five and 10-year estimates were 0.21 (95% CI, 0.19 to 0.24) and 0.22 (95% CI, 0.20 to 0.25) in the whole series, 0.20 (95% CI, 0.17 to 0.23) and 0.21 (95% CI, 0.18 to 0.24) in the primary cases, and 0.25 (95% CI, 0.19 to 0.30) and 0.25 (95% CI, 0.20 to 0.30) in recurrences. This difference in distant metastasis according to phase (primary v recurrence) at presentation was not statistically significant (P = .174).
In Figure 3, crude cumulative incidence curves are represented according to microscopic margin status. For primary cases (Fig 3A), 10-year estimates were 0.23 (95% CI, 0.14 to 0.32) in positive-margin cases and 0.21 (95% CI, 0.17 to 0.24) in negative-margin cases (P = .450). The corresponding figures in recurrent cases (Fig 3B) were 0.31 (95% CI, 0.21 to 0.42) and 0.23 (95% CI, 0.17 to 0.29; P = .124).
Table 4 presents the results from the Cox multiple regression model for the whole case series. Significant prognostic factors for distant metastasis were tumor depth, size, FNCLCC grade, and histotype. The hazard ratio for positive- versus negative-margin tumors was 1.1 (95% CI, 0.8-1.6; P = .495), a figure quite similar to that obtained for mortality.
Results with the Cox model fitted in the subset of patients with primary and recurrent tumors (not shown in detail) overlapped those previously shown for the whole case series as regards the prognostic effect of covariates. In the primary tumor subset, the hazard ratio of microscopic margin status was 0.8 (95% CI, 0.5 to 1.3; P = .416); in the recurrent tumor subset, the hazard ratio was 1.5 (95% CI, 0.9 to 2.5; P = .136).
DISCUSSION
As far as primary tumors are concerned, patients with positive margins behave like those with negative margins in the first 3 to 5 years of follow-up, but in the subsequent interval, the difference in favor of negative margins becomes more striking (hazard ratio, 1.8 after 5 years, compared with 0.8 in the first 5 years), though still not statistically significant (P = .179). This implies a time-dependent relationship between quality of surgery and final outcome, by which poor surgical margins affect survival only after a time interval, as though only patients who escape a systemic risk may benefit, or suffer, from quality of surgery. A similar observation was made by Stojadinovic et al9,15 in a series from Memorial Sloan-Kettering Cancer Center. Indeed, this series included all sites, while our series is made up of extremity cases only. Obviously, one may believe that causes of systemic failure connected to the inherent aggressiveness of the disease prevail in the early years, while the way in which the disease has been treated locally becomes more important afterwards, when inherent aggressiveness has already produced its effects. If so, quality of surgery might be important for the proportion of patients who survive the inherent aggressiveness of disease. Indeed, patients with positive margins do not represent more than 20% of all surgically treated patients. Therefore, the lack of statistical significance in our series is likely to simply depend on the relatively small numbers. That said, quality of surgery might make the difference after 3 to 5 years for this limited subset of patients, being overshadowed earlier by more important prognostic factors. This may possibly create understanding as to why so often, the independent prognostic value of quality of surgical margins could not be objectively appreciated when analyzing large case series, though it nonetheless seems universally important to the sarcoma surgeon.
Overall, however, while positive surgical margins adversely affected the local outcome, as shown both at univariable and multivariable analyses, distant metastasis and cause-specific mortality were only slightly different between the two groups of patients resected for a primary tumor with negative and positive margins, and this difference was not statistically significant at multivariable analysis. In other words, positive surgical margins may not be an independent factor, simply reflecting other problems inherent to the disease (eg, size, depth). Therefore, the same adverse prognostic factors would affect quality of surgery, local recurrence, and distant metastases.
In published literature, most studies failed to find a strong correlation between quality of surgery and final outcome. In both a series of 559 patients treated within the Scandinavian Sarcoma Group centers,14 and another series of 848 patients treated at The University of Texas M.D. Anderson Cancer Center,16 what was relevant for distant metastasis, and therefore disease-specific survival, was the intrinsic biologic aggressiveness of the tumor, as defined by size, depth, malignancy grade, and histotype.4-7,11,17 In this sense, local recurrence might be well regarded as an additional biologic marker of aggressiveness (ie, a result rather than a cause). Under this interpretation, when a local recurrence is detected, the patient should be viewed as being at a higher risk of death due to factors inherent to his disease. Also, Lewis et al26, in a series of 911 patients treated at Memorial Sloan-Kettering Cancer Center, found local that recurrence correlated with the development of subsequent metastases and tumor mortality, but concluded that local recurrence is not a source of metastasis, but rather a biologic phenomenon on the continuum of recurrent sarcoma. Some other smaller series showed that improved local control rates observed in patients with adequate surgical margins failed to translate into a survival benefit.27-31 Furthermore, in the three published randomized trials18-20 that compared more and less extensive local treatment modalities (amputation v conservative surgery + radiation therapy, or conservative surgery v conservative surgery + adjuvant radiation), differences in local outcome did not translate into differences in survival. The power of these randomized studies, however, is limited by the small sample size; therefore, no definitive conclusions can be drawn.
An important point should be taken into account when interpreting the lack of correlation between quality of surgical margins and final outcome. A strict application of Enneking criteria32 is precluded when analyzing large retrospective series, and the only possibility is to codify margins as either positive or negative. Indeed, the positive or negative status of the closest surgical margin is clearly a rough indicator of quality of surgery. Actually, some marginal excisions may well be included in the negative-margins group, thus contributing to confounding conclusions. Of course, the anatomic heterogeneity of adult STSs adds to these difficulties. It is probable that innovative definitions of quality of surgical margins (eg, based on width and type of surrounding healthy tissues, type of tumor margins, histotype, and grade) may help settle this highly controversial issue, and prospective clinical studies aimed at working out new definitions of adequacy of surgery would be welcome at this stage.
By contrast, in patients with an already locally recurrent tumor, the effect of surgical margins was more striking in our series. A distinct impact on metastasis-free and overall survival, as well as local control, was observed. One may believe that this patient population is selected in such a way that the local control is more important for them and may even have a prognostic impact. Clearly, however, patients in whom negative margins were not reached will have had worse prognostic factors locally. Therefore, we cannot conclude that a more aggressive local policy would have changed the final outcome. At any rate, it was just in this patient subgroup that receiving radiation therapy or not had prognostic relevance (and the choice to do radiation therapy or not, in this retrospective series, may have been less dependant on the local aggressiveness of disease). In brief, the hypothesis is that local recurrence defines a subgroup of patients for whom more aggressive local treatment might be worthwhile, though final proof thereof cannot be provided by this retrospective analysis.
In conclusion, microscopic residual disease may have an independent effect on disease-specific mortality, but this effect is minor. Indeed, a small proportion of patients (ie, those escaping early relapse due to inherent aggressiveness of disease) may suffer from inadequate surgery more clearly. Therefore, there are still good reasons to continue to pursue appropriate surgery in adult STSs (ie, clean margins at the very least; more sophisticated classifications of adequacy of margins would be welcome). On the other hand, when the clinical presentation is borderline; for example, when obtaining clean margins would require an amputation or major functional impairments, decision making may be reasonably flexible and shared with the patient, considering that his or her survival will directly depend only to a limited degree on quality of surgery itself. This may not apply to patients who have already had a local relapse. These patients seem to be selected in such a way that indicates that quality of margins is more important for them. In this subgroup of patients, therefore, to our knowledge, an attempt to reach an adequate local control might be especially crucial.
Authors’ Disclosures of Potential Conflicts of Interest
NOTES
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Brennan M, Alektiar KM, Maki RG: Soft tissue sarcoma, in DeVita VT, Hellmann S, Rosenberg SA (eds): Cancer: Principles and Practice of Oncology. Philadelphia, PA, Lippincott Williams & Wilkins, 2001, pp 1841-1891
2. Greenlee RT, Hill-Harmon Mb, Murray T, et al: Cancer statistics, 2001. CA Cancer J Clin 51:15-36, 2001
3. Trojani M, Contesso G, Coindre JM, et al: Soft tissue sarcomas of adults: Study of pathological prognostic variables and definition of a histopathologic grading system. Int J Cancer 33:37-42, 1984
4. Gustafson P: Soft tissue sarcoma epidemiology and prognosis in 508 patients. Acta Orthop Scand 65:1-31, 1994 (suppl 259)
5. Singer S, Corson K, Gonin JM, et al: Prognostic factors predictive of survival and local recurrence for extremity soft tissue sarcoma. Ann Surg 219:165-173, 1994
6. Pisters PWT, Leung DHY, Woodruff J, et al: Analysis of prognostic factors in 1041 patients with localized soft tissue sarcoma of th extremities. J Clin Oncol 14:1679-1689, 1996
7. Coindre JM, Terrier P, Bui NB, et al: Prognostic factors in adult patients with locally controlled soft tissue sarcoma: A study on 546 patients from the French Federation of Cancer Centers Sarcoma Group. J Clin Oncol 14:869-877, 1996
8. Coindre JM, Terrier P, Guillou L, et al: Predictive value of grade for metastasis development in the main histologic types of adult soft tissue sarcomas: A study of 1240 patients from the French Federation of Cancer Centers Sarcoma Group. Cancer 91:1914-1926, 2001
9. Stojadinovic A, Leung DHY, Allen P, et al: Primary adult soft tissue sarcoma: Time-dependent influence of progostic variables. J Clin Oncol 20:4344-4352, 2002
10. Weitz J, Antonescu CR, Brennan MF: Localized extremity soft tissue sarcoma: Improved knowledge with unchanged survival over time. J Clin Oncol 21:2719-2725, 2003
11. Karakousis CP, Proimakis C, Rao U, et al: Local recurrence and survival in soft tissue sarcomas. Ann Surg Oncol 3:255-260, 1996
12. Brennan MF: The enigma of local recurrence. Ann Surg Oncol 4:1-12, 1997
13. Lewis JJ, Leung DHY, Casper ES, et al: Multifactorial analysis of long-term follow-up (more than 5 years) of primary extremit sarcoma. Arch Surg 134:190-194, 1999
14. Trovik CS, Bauer HCF, Alvegard TA, et al: Surgical margins, local recurrence and metastasis in soft tissue sarcomas: 559 surgically-treated patients from the Scandinavian Sarcoma Group register. Eur J Cancer 36:710-716, 2000
15. Stojadinovic A, Leung DHY, Hoos A, et al: Analysis of the prognostic significance of microscopic margins in 2084 localized primary adult soft tissue sarcomas. Ann Surg 235:424-433, 2002
16. Zagars GK, Ballo MT, Pisters PWT, et al: Prognostic factors for patients with localized soft tissue sarcoma treated with conservation surgery and radiation therapy: An analysis of 1225 patients. Cancer 97:2530-2543, 2003
17. Eilber FC, Rosen G, Nelson SD, et al: High grade extremity soft tissue sarcomas: Factor predictive of local recurrence and its effect on morbidity and mortality. Ann Surg 237:218-226, 2003
18. Rosenberg SA, Tepper J, Glatsein E, et al: The treatment of soft tissue sarcoma of the extremities: Prospective randomized evaluation of (1) limb-sparing surgery plus radiation therapy compared with amputation and (2) the role of adjuvant chemotherapy. Ann Surg 196:305-315, 1982
19. Pisters PWT, Harrison LB, Leung DHY, et al: Long-term results of a prospective randomized trial of adjuvant brachitherapy in soft tissue sarcoma. J Clin Oncol 14:859-868, 1996
20. Yang JC, Chang AE, Baker AR, et al: Randomized prospective study of the benefit of adjuvant radiation therapy in the treatment of soft tissue sarcomas of the extremity. J Clin Oncol 16:197-203, 1998
21. Marubini E, Valsecchi MG: Analysing survival data for clinical trials and observational studies. Chichester, UK, John Wiley and Sons, 1995
22. Gray RJ: A class of k-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 16:1141-1154, 1988
23. Durrleman S, Simon R: Flexible regression models with cubic splines. Stat Med 8:551-561, 1989
24. Schoenfeld D: Partial residuals for the proportional hazards regression model. Biometrika 69:239-241, 1982
25. SAS Institute Inc: SAS/STAT Software: Changes and Enhancements Through Release 6.11. Cary, NC, SAS Intitute Inc, 2000
26. Lewis JJ, Leung DHY, Heslin M, et al: Association of local recurrence with subsequent survival in extremity soft tissue sarcoma. J Clin Oncol 15:646-652, 1997
27. Ueda T, Yoshikawa H, Mori S, et al: Influence of local recurrence on the prognosis of soft tissue sarcomas. J Bone Joint Surg 79B:553-557, 1997
28. Tanabe KK, Pollock RE, Ellis LM, et al: Influence of surgical margins on outcome in patients with preoperatively irradiated extremity soft tissue sarcomas. Cancer 73:1652-1659, 1994
29. Li XQ, Parkekh SG, Rosenberg AE, et al: Assessing prognosis for high grade soft tissue sarcomas: Search for a marker. Ann Surg Oncol 3:550-557, 1996
30. Gustafson P, Dreinhofer KE, Rydholm A: Metastasis-free survival after local recurrence of soft tissue sarcoma. J Bone Joint Surg 75B:658-660, 1993
31. Sadoski C, Suit HD, Rosenberg A, et al: Preoperative radiation, surgical margins, and local control of extremity sarcomas of soft tissue. J Surg Oncol 52:223-230, 1993
32. Enneking WP, Spainer SS, Goodman MA: A system for the surgical staging of musculoskeletal sarcoma. Clin Orthop 153:106-120, 1980(A. Gronchi, P.G. Casali, )