| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
1 Department of Surgery, Istituto Nazionale per lo studio e la cura dei Tumori, via Venezian, 1, 20133, Milano, Italy
2 Department of Biostatistics, Istituto Nazionale per lo studio e la cura dei Tumori, via Venezian, 1, 20133, Milano, Italy
3 Department of Pathology, Istituto Nazionale per lo studio e la cura dei Tumori, via Venezian, 1, 20133, Milano, Italy
4 Department of Diagnostic Imaging and Radiotherapy, Istituto Nazionale per lo studio e la cura dei Tumori, via Venezian, 1, 20133, Milano, Italy
5 Department of Cancer Medicine, Istituto Nazionale per lo studio e la cura dei Tumori, via Venezian, 1, 20133, Milano, Italy
Correspondence: Address correspondence and reprint requests to: Alessandro Gronchi, MD; E-mail: alessandro.gronchi{at}istitutotumori.mi.it
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: Out of a series of 1017 patients with extremity soft tissue sarcoma treated over 20 years, we picked a group of 238 patients operated on at our institution for their first local relapse: 88 after their primary operation performed at the same center and 150 elsewhere. At operation for relapse, margins were microscopically negative in 77% and 75% of patients, respectively. Median follow-up was 107 months.
Results: The 10-year mortality rate was 22% in the absence of local relapse, whereas in locally relapsing patients it was 54% and 43%, respectively, for patients first operated on at our institute and for those who were not. The hazard ratio of positive versus negative surgical margins was 1.7 for cause-specific death and 2.1 for distant metastases in patients first operated on at our institute, as opposed to 1.2 and 1.3 for the others.
Conclusions: Local relapse was an unfavorable prognostic factor. In the face of a consistent surgical policy for local relapse in a single-institution setting, patients relapsing after the first operation performed at our institution received rescue treatment less frequently than those previously operated on outside a referral center. This is likely due to an inherently higher tumor aggressiveness. In the presence of such a higher aggressiveness, the adequacy of surgical margins at operation for first relapse seemed more critical prognostically. This may have clinical and speculative implications.
Key Words: Sarcoma • Local recurrence • Limbs • Surgery • prognosis
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
To further investigate the prognostic meaning of local relapse and surgical margins, we retrospectively reanalyzed our institutional series by selecting all local relapses seen in patients surgically re-treated at our institution over a 20-year time span. Therefore, this article focuses on first local relapses. These have been split according to whether surgery for the first local relapse followed a previous one performed at our institute or elsewhere. In other words, we considered all patients surgically treated at our institution for an extremity soft tissue sarcoma in 20 years and picked up all patients surgically treated at our institution for their first local relapse, and then we split them according to whether their first operation had been performed at our institute or elsewhere. One may assume that providing rescue treatment for a local relapse after an operation performed at a referral center may have a different meaning from that for a previous operation performed elsewhere. The former local relapse may largely depend on the tumor’s inherent aggressiveness, whereas the second may also reflect a degree of surgical inadequacy.
This may be a strong assumption, although it is widely believed that surgery for extremity soft tissue sarcomas is exceedingly sensitive to institutional experience. Actually, making this assumption might help us to understand the direct contribution of local relapse, and surgical adequacy, to the prognosis of soft tissue sarcomas. Indeed, a querelle dating back many years has focused on whether local relapse in soft tissue sarcomas directly contributes to distant metastases and, thus, the eventual prognosis of the disease or whether it is just a marker of tumor aggressiveness.9–18 As of today, this querelle is far from being settled.
|
PATIENTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Seven hundred thirty-eight patients presented with primary disease, and 279 had a locally recurrent tumor. Among the 738 primary cases, 135 patients developed a local relapse. Of these patients, 12 were not operated on (they underwent radiotherapy alone in 1 case, chemotherapy in 4 cases, isolated limb perfusion in 3 cases, and no treatment in 4 cases as a result of patient refusal). Thirty-five patients developed local relapse after distant metastases. The remaining 88 patients were operated on with an eradicating intent. Among the 279 patients who presented at our center with a local relapse, 150 had only 1 operation before being referred to our institution.
We investigated the postrelapse course of disease in the 88 recurrent cases already treated at Istituto Nazionale dei Tumori for their primary disease and the 150 recurrent cases previously treated elsewhere (referred to as PR-P and PR-R, respectively, henceforth). None of the 150 recurrent cases previously treated elsewhere had undergone a re-excision at our center for their primary tumor and were otherwise included in the first group (PR-P).
Tumors were characterized as superficial or deep according to the involvement of the investing fascia. All tumors were reviewed by experienced pathologists at our institution. The FNCLCC grading system19 was applied to the untreated primary tumors. For recurrent tumors, grading was assigned by evaluating the primary untreated tumor seen on consultation and was then reassigned on relapse.
Surgical excisions were considered as macroscopically complete in the absence of gross residual disease. No macroscopically incomplete resection was performed on primary tumors or recurrences. Margins were evaluated by a dedicated pathologist. In general, the surgical specimen was examined in the presence of the operating surgeon; margins were inked and separately sampled. The closest margin was microscopically categorized as positive (tumor within 1 mm of the inked surface) or negative (absence of tumor within 1 mm of the inked surface).
The indication for radiotherapy was given by both the operating surgeon and the radiation oncologist when a higher risk of relapse was suggested on clinical grounds. However, no prospectively selected criteria were used to this end. External beam radiation was used in all such cases, and doses ranged from 45 to 70 Gy (median, 60 Gy).
Chemotherapy was given at the discretion of the multidisciplinary institutional sarcoma board or as part of ongoing clinical trials. Anthracycline-based regimens were used and in most cases were combined with ifosfamide.
We considered as end points for this analysis cause-specific death, local relapse, and distant metastasis (whichever occurred first between the two latter). Multivariable proportional hazard regression models were performed to investigate the prognostic effect of local relapse and margin status (negative or positive) on each end point while adjusting for the following covariates: tumor size (
5, 6–10, or >0 cm), depth (superficial or deep), histotype (liposarcomas, malignant fibrous histiocytoma/synovial/malignant peripheral nerve sheath tumor, or other), grade (1, 2, or 3), and chemotherapy and radiotherapy (yes or no). In particular, local relapse was analyzed as a time-dependent variable to compare the postrelapse phase versus the prerelapse phase of disease separately for PR-P and PR-R patients. Additional exploratory analyses were performed by including in the models interaction terms suitable for testing the prognostic effect of time to local relapse and whether local relapse modified the prognostic effect of other covariates.
The model results are summarized by reporting the hazard ratio (HR) estimates, corresponding 95% confidence intervals, and P values (Wald test). Additional descriptive analyses concerned the shape of cause-specific death, local relapse, and distant metastasis prerelapse hazard functions and 5- and 10-year average risk estimates in the prerelapse and postrelapse phases.
We used SAS (SAS Institute Inc., Cary, NC) and R software to perform the modeling and statistical calculations. We considered two-sided P values less than the conventional 5% threshold as significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
separately for patients with primary disease in the prerelapse phase, PR-P, and PR-R patients. Sex and age at diagnosis, as well as tumor site and histotype, were characteristics inherent to the patient or disease, whereas size, depth, grade, margin status, and treatments were assessed either on the primary tumor or on the locally recurring lesion. The three investigated sets of patients proved reasonably balanced for sex, age, site, and size. The most represented histotype was liposarcoma; however, malignant fibrous histiocytoma and malignant peripheral nerve sheath tumor were more frequently diagnosed among PR-P and PR-R patients, respectively. PR-P and PR-R patients also tended to show less favorable tumor characteristics in terms of depth, grade, and margin status. When comparing the primary tumor of the PR-P and the PR-R patients, no major differences in term of size, depth, and grade were observed. As regards treatment, the most notable differences were the greater number of major exarticulations, along with a less frequent use of chemotherapy, in the PR-P patients.
|
The median time from diagnosis to local relapse was slightly higher in PR-P (19.5 months; interquartile range, 10–37 months) than in PR-R patients (16 months; interquartile range, 7–37 months). As of June 2006, the median follow-up duration from the first operation at Istituto Nazionale dei Tumori in the overall series of 888 patients (738 primary cases and 150 cases previously treated elsewhere) was 107 months (interquartile range, 62–131 months). A small fraction of patients was lost to follow-up, amounting to 45 (5%) before the 10th year.
The number of events according to disease phase is shown in Table 2. The figures were not negligible even in the smaller PR-P and PR-R sets.
|
provides results of multivariable proportional hazard models. Phase of disease yielded significant results for all the investigated end points. The prerelapse hazard functions of cause-specific death, local relapse, and distant metastasis are shown in Fig. 1. The local relapse hazard tended to decrease with time, up to reaching a plateau at approximately 48 to 50 months. A similar pattern was observed for distant metastasis, but the hazard levels were initially higher than for local relapse. In contrast, the hazard of death tended to increase until approximately 24 months and to decrease thereafter up to stabilization after approximately 60 months. In any case, a clear finding is that the first 5 years after surgery were the most relevant from a prognostic viewpoint. The hazards of death, local relapse, and distant metastasis increased in the postrelapse phase, as shown by the adjusted HR estimates consistently >1 (Table 3). Such an effect was stronger in PR-P cases, for which the HRs were consistently higher than for PR-R cases. HRs for the comparison between PR-P and PR-R cases (not shown in Table 3) were 2.04 (1.32–3.13) for death, 1.47 (.86–2.50) for local relapse, and 2.22 (1.35–3.70) for distant metastasis. There was no evidence that local relapse could change its prognostic effect depending on the time of relapse occurrence, as reflected by the high P values for testing the phase of disease x time interaction with all the investigated end points (cause-specific death, P = .597; local relapse, P = .724; distant metastasis, P = .811). What changed over time was the absolute risk of developing distant metastasis and death, as reported in Fig. 1, so that the sooner the local recurrence the higher the subsequent risk, gained by the original risk, calculated on the characteristics of the original presentation,20–22 times the relative HR related to the type of local recurrence (PR-P vs. PR-R). In this series,16,18 the relative HRs of local recurrences remained constant over time. The 5- and 10-year average risk estimates according to phase of disease are shown in Table 4. For all events, the risk was clearly higher in PR-P patients, with absolute 10-year differences toward the prerelapse phase that ranged between 32% (from 22% to 54%) for cause-specific death and 41% (17% to 58%) for local relapse. PR-R patients were in between, and the absolute increase in risk was at most 29% (17% to 46%) for local relapse.
|
|
|
). The prognostic effect of margin status in the prerelapse or postrelapse phase is provided in Table 5. The strongest effect was generally observed in PR-P patients; however, the difference toward prerelapse or PR-R patients was not statistically significant at the test for interaction. For local relapse, in particular, the HR estimates were 3.31 in PR-P patients, compared with 2.78 and 1.97 for prerelapse and PR-R patients, respectively.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). However, patients who developed a local relapse after an operation performed at our institution were 3.7 times more likely to die of disease, as compared with 1.8 times for those operated on at our institution for a relapse after a previous operation performed elsewhere. There is little doubt in the literature that local relapse is associated with a worse prognosis. To quote just one of the main and most recent series, results similar to ours were reported by Eilber et al.,6 with a 3-fold increase in mortality rate after local relapse. The unsolved issue is whether local relapse is causative. The occurrence of local relapse per se might favor the systemic spread of disease and, therefore, directly affect survival, or it might simply be a marker of biological tumor aggressiveness. Theoretically, the two mechanisms might coexist and possibly interact, with both contributing to the outcome.6,9–11,13–18
In this regard, our observation may be meaningful. Some patients had a local relapse after surgery performed at nonreferral centers, whereas others had it after surgery performed at our institution. The assumption may well be that, for soft tissue sarcoma, surgery performed at a referral institution is likely to be more adequate, and thus a local relapse may depend more on the disease’s inherent aggressiveness.9 This is left to demonstrate, but in our series, the strongest adverse prognostic effect was exerted by local relapse in those patients previously operated on at our institute.
Clearly, the most interesting problem is not so much whether local relapse is causative, but whether some surgical inadequacy behind it is prognostically relevant.1–4 In soft tissue sarcoma, it is not easy to rate surgical adequacy objectively. The margin status is not necessarily the only marker of surgical adequacy. For example, anatomical margins may differ in their value, all anatomical sites are not equal, and so forth. However, whether margins were clear or not is an easy thing to record in retrospective series.1,4,5,8 In our series, positive margins exerted an adverse effect much more on local relapse than on distant metastasis, as already reported in our previous analysis,1 with an estimated HR of 2.60 (P < .001). More precisely, a weaker but consistent trend was observed in regard to distant metastasis and mortality, which failed to reach statistical significance. However, an estimated HR of 1.25 denotes a 25% relative increase in the mortality rate: to demonstrate such an increase as significant, with a two-sided test at the 5% significance level, 80% power, and 73% 10-year survival (the average for the 738 Istituto Nazionale dei Tumori primary cases), approximately 700 deaths should be recorded from some thousands of patients, with their precise number depending on the accrual and follow-up pattern within the cohort. By the way, similar considerations in terms of statistical power also apply to published studies comparing different local treatment strategies, which generally failed to show any effect on survival, despite a local relapse reduction with more aggressive treatments.23–25 Therefore, we cannot draw any definite conclusions, but in any case the effect of margins on distant metastasis seems to be weaker.
It is interesting to note, however, that a difference was found depending on where the previous operation had been performed when the prognostic effect of margins was investigated (Table 5). The adverse effect of positive margins was comparatively stronger when patients with recurrent disease were considered who had already been operated on at our institution for their primary disease. The HR for cause-specific death was 1.70, denoting a 70% relative increase in the mortality rate for patients with positive margins. Again, our interpretation may be that patients with local recurrence after being first operated on at our institution had a higher tumor aggressiveness. In the face of this, the unfavorable effect of positive margins was higher, as if the two factors—tumor aggressiveness and surgical adequacy—add to each other or as if surgical adequacy becomes more important when the disease is more aggressive.
In brief, in our series, (1) the prognostically adverse effect of local relapse was greater in patients in whom the aggressiveness of the disease could be its main explanation; (2) positive margins did not matter so much for mortality as for local relapse; and (3) when the tumor aggressiveness was high enough to have given a local relapse unexplained by previous surgical inadequacy, the adequacy of (salvage) surgery seemed to regain a prognostic value in terms of survival. One may conclude that, at the beginning, surgical adequacy is mainly important for local relapse and that tumor aggressiveness is important for both local and distant relapse. In the presence of a local relapse mainly favored by tumor aggressiveness, quality of margins may become more important. This could also be stated differently by saying that the quality of surgery is all the more important when the tumor is aggressive. This is somewhat contraintuitive, because systemic rather than local treatments may seem to be the focus when tumor aggressiveness is high. Indeed, there is no reason to rule out the idea that systemic treatments are important as well. One should simply acknowledge that quality of surgery is important in these patients, along with, possibly, any other treatment option. Neither should one believe that surgical adequacy is less important if the disease is less aggressive: the fact that prognosis may be relatively unaffected by local relapses does not mean that a local relapse is not a problem on its own in terms of costs, quality of life, and so on.
Aside from obvious academic implications, there may be a very practical consequence. One may speculate that more aggressive surgery might be justified in patients with local recurrence after initially adequate treatment, because in these patients the adverse effect of inadequate margins may add to the inherent tumor aggressiveness enough to actually affect the final outcome. Strategically, the sarcoma community might well go on with the conservative surgical policies pursued in soft tissue sarcoma over the last decades, but it may need to accept an amputation rate for local relapses after adequate primary operation. Of course, if tumor aggressiveness could be objectively assessed initially, a different surgical policy might be justified from the very beginning. This might be the subject of a prospective clinical trial, as difficult it may be, provided one believes that markers of inherent tumor aggressiveness are actually available in soft tissue sarcoma.
|
FOOTNOTES |
|---|
Received for publication October 28, 2006. Accepted for publication November 28, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. Gronchi, S. Lo Vullo, M. Fiore, C. Mussi, S. Stacchiotti, P. Collini, L. Lozza, E. Pennacchioli, L. Mariani, and P. G. Casali Aggressive Surgical Policies in a Retrospectively Reviewed Single-Institution Case Series of Retroperitoneal Soft Tissue Sarcoma Patients J. Clin. Oncol., January 1, 2009; 27(1): 24 - 30. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. M. Thomas, A. Gronchi, and B. O'Sullivan A Multidisciplinary Approach to Retroperitoneal Sarcomas: Current State-of-the-Art ASCO Educational Book, January 1, 2009; 2009(1): 701 - 708. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
