The objective of this study was to investigate prognostic factors and clinical outcome of myxoid/round cell and pleomorphic liposarcoma.
The objective of this study was to investigate prognostic factors and clinical outcome of myxoid/round cell and pleomorphic liposarcoma.
Three hundred twenty-nine patients with localized myxoid/round cell or pleomorphic liposarcoma who underwent surgery at the Istituto Nazionale per lo Studio e la Cura dei Tumori (Milan, Italy) over 25 years were reviewed. The reates of local recurrence, distant metastases, and survival were studied.
Two hundred fourteen patients presented with primary disease, and 115 patients had locally recurrent tumors. The disease-specific survival rate was 75% at 10 years, and the local recurrence and distant metastases incidence were 25% and 15%, respectively. Presentation with recurrent disease, tumor size (>10 cm), tumor grade (French Federation of Cancer Centers grade II or III vs grade I), and positive surgical margins were independent predictors of death. Tumor site and radiation therapy also played a role, mostly related to their effect on local outcome. Pathologic grade and histologic subtype influenced distant metastases. Extrapulmonary metastases were associated with poorer postmetastatic disease-specific survival.
Myxoid/round cell liposarcomas shared similar prognostic factors with other soft tissue sarcomas and had a relatively good clinical outcome. The presence of >5% of round cell component singled out a group of patients at greater risk of metastases and death but with a broad spectrum of disease aggressiveness. Extrapulmonary metastases were a peculiar pattern of myxoid/round cell liposarcoma that require special consideration for treatment and prognosis. Cancer 2007. © 2007 American Cancer Society.
Liposarcoma (LPS) is one of the most common histologic subtypes of adult soft tissue sarcoma. It can be subdivided in 3 variants: well differentiated/dedifferentiated, myxoid/myxoid round cell, and pleomorphic. Pure myxoid LPS (MLPS) accounts for 30% of all LPS.1 Round cell LPS (RCLPS) is defined as a form of MLPS that has an associated round cell component in >5% of the tumor2 and accounts for 15% of all LPS. The same chromosomal translocation t(12;16)(q13;p11) has been demonstrated in both in MLPS and RCLPS,3 resulting in a fusion derived from the t(12;16) malignant LPS (FUS-CHOP) transcript.
Pleomorphic LPS (PLPS) does not belong directly to the MLPS/RCLPS group. PLPS tumors appear to be a distinct category, closer to other pleomorphic sarcomas than to MLPS/RCLPS or well-differentiated LPS.4 Nevertheless, a small round cell variant that contains pleomorphic lipoblasts and small round cells that is virtually indistinguishable from RCLPS has been observed.5 Recently, a rare epithelioid variant of PLPS carrying the typical MLPS/RCLPS FUS-CHOP transcript also has been described.6
MLPS/RCLPS tumors are known as particularly sensitive to radiotherapy and chemotherapy compared with other histologic subtypes of soft tissue sarcoma.7, 8 Recent data have raised new interest in medical treatments for MLPS/RCLPS.9 Therefore, we undertook the current retrospective analysis of our institutional series of patients with MLPS/myxoid RCLPS and PLPS to investigate prognostic factors and long-term outcomes and to gain a better understanding of the possible role of new therapeutic approaches.
Between January 1980 and December 2005, 3025 consecutive patients with soft tissue sarcomas underwent surgery with the intent of eradicating disease at the National Institute for the Study and Cure of Tumors in Milan, Italy. In this series, 329 patients had a diagnosis of localized MLPS/RCLPS or PLPS.
Tumors were located mainly at the extremities and girdles. Retroperitoneal or other abdominal (intra-abdominal and pelvic) locations were grouped together. Paraspinal, chest, and abdominal wall were labeled as trunk locations; whereas the exceptional head and neck locations were left separate. Tumor depth was defined in relation to the investing fascia. All retroperitoneal/other abdominal locations were considered deep.
With regard to the histotypes, over the years, all tumors were reviewed by at least 2 experienced pathologists from our institution. The diagnosis was confirmed molecularly whenever frozen material was available. RCLPS was defined by the presence >5% round cells.
The French Federation of Cancer Centers grading system10 was used to assign tumor grade in untreated primary tumors. Grading for recurrent tumors was performed on the slides from the primary, untreated tumor seen in consultation.
An attempt to perform a complete resection with negative margins was performed in all patients. For tumors of the extremities, the surgical procedure was similar to what was performed normally for all others soft tissue sarcomas. Eleven of 273 patients underwent a major amputation. In retroperitoneal, trunk, head and neck locations, wide margins were achievable only rarely. No patients died from surgical complications. Margins were defined at the time of pathologic assessment by a dedicated pathologist. The surgical specimen always was examined in the presence of the operating surgeon. The margins were inked and were sampled separately. The closest margin was categorized microscopically as positive (tumor within 1 mm from the inked surface) or negative (absence of tumor within 1 mm from the inked surface).
Clinical data were extracted from a prospective database of all adult patients with soft tissue sarcoma who were treated at our institution. The data retrieved included sex, age at diagnosis, tumor site, phase at presentation (primary vs recurrent), type of surgery, tumor size, tumor depth, histotype, pathologic grade, margin status, adjuvant treatments, dates of neoplastic events, site of distant metastasis (DM), and death or last follow-up.
Radiotherapy was delivered as an adjunct in 128 patients (39%; 84 patients with primary tumors and 44 patients with recurrences). The indication for radiation therapy was verified both by the operating surgeon and by the radiation oncologist when, based on clinical grounds, there was a greater risk of recurrence. However, no prospectively selected criteria were used to this end. External beam radiation was used in all such patients, and doses ranged from 45 gray (Gy) to 65 Gy (median, 57 Gy).
Chemotherapy was received by 61 patients (19%), including 39 patients with primary tumors and 22 patients with recurrent tumors, at the discretion of the multidisciplinary Soft Tissue Sarcoma Group at our institution or as part of clinical trials. Antracycline-based regimens were used, in most patients with ifosfamide.
The median follow-up duration for the entire group, as of June 2006, was 119 months (interquartile range, 70–153 months). Ten patients (3%) were lost to follow-up before 10 years. The current study was approved by our Institutional Review Board.
The endpoints of this study were disease-specific survival (DSS), local recurrence (LR) incidence, and DM incidence. According to the clinical literature, curves relating to these endpoints were estimated within each subgroup determined by the status of disease at presentation to our institution (primary or recurrent tumor). DSS curves were calculated by using the Kaplan-Meier method and were compared by using log-rank tests.
Crude cumulative incidence (CCI) curves for LR and DM were calculated in a competing-risks framework, as described by Marubini and Valsecchi,11 and comparisons between curves were performed using of the Gray test.12 The time to occurrence of any event was computed from the date of surgery at our institution to the date when the event first was recorded or was censored at the date of last follow-up assessment in event-free patients.
In the analysis of cause-specific survival, deaths caused by conditions unrelated to sarcoma were censored. For the analysis of disease recurrence patterns, LRs, distant recurrences, and deaths in recurrence-free patients, whichever occurred first, were regarded as competing events. Concomitant LRs and DMs were included in the estimation of the CCI curves as DM.
Multivariable analyses of DSS, LR-free survival, and DM-free survival were based on cause-specific hazards and, thus, were performed using Cox multiple-regression models. Putative prognostic factors that were included in the models were presentation (primary or recurrent), tumor site (extremities or other), tumor size (≤5 cm or >5 cm), tumor grade (Grade I, II, or III), tumor depth (deep or superficial), margin status (negative or positive), chemotherapy (yes or no), radiotherapy (yes or no), and histologic subtype (MLPS, RCLPS, or PLPS). Because of their strong correlation, grade and histologic subtype had to be entered separately into the Cox models. The effects for the remaining factors were estimated from the models that included histology, the effect of which was slightly stronger than that observed for grade.
The distribution of patient and disease characteristics was compared between patients with primary tumors and patients with recurrent tumors by means of the Wilcoxon rank-sum test and the Fisher exact test for continuous and categorical variables, respectively.
We used SAS software (SAS Institute Inc., Cary, NC) and the S-Plus (StatSci, MathSoft, Seattle, WA) Design Library (available at URL: http://lib.stat.cmu.edu) and Cmprsk Library (available at URL: http://biowww.dfci.harvard.edu/≈gray/) to perform the modeling and statistical calculations. For each test, results were considered statistically significant whenever a 2-sided P value <.05 (5%) was achieved.
Clinical characteristics are summarized in Table 1.
|Median age (IQ range), y||46||(34–57)||55||(46–63)||49||(39–59)||<.0001|
|Head and neck||1||1||1||1||2||1|
|Median size (IQ range), cm*||8||(5–13)||10||(6–19)||9||(5–15)||.0009|
Sixty-nine patients died of disease (24 of 214 patients with primary tumors and 45 of 115 patients with recurrent tumors), 27 patients died of intercurrent disease other than sarcoma, and the remaining 233 patients were alive at the last follow-up. The time to disease-specific death in the entire series varied from 5 months to 219 months (median for those who died, 48 months). For patients with MLPS, the time to disease-specific death varied from 5 months to 219 months (median for those who died, 54 months); for patients with RCLPS, the time to disease-specific death varied from 5 months to 127 months (median for those who died, 27 months); and for the patients with PLPS, the time to disease-specific death varied from 6 months to 195 months (median for those who died, 49 months).
The DSS rates were 83% at 5 years and 75% at 10 years. In the patients who had primary tumors, the DSS rates were 90% and 87% at 5 years and 10 years, respectively, whereas, in the patients who had recurrent tumors, the DSS rates were 72% and 56%, respectively, with a significant difference observed between the 2 groups (P < .001).
Among the patients who had primary tumors, the 5-year and 10-year DSS rates were 93% and 92%, respectively, for the MLPS group; 87% and 77%, respectively, for the RCLPS group; and 81% and 81%, respectively, for the PLPS group. No significant differences were observed between the 3 groups (Fig. 1A).
Among the patients who had recurrent tumors, the 5-year and 10-year DSS rates were 80% and 61%, respectively, for the MLPS group; 59% and 47%, respectively, for the RCLPS group; and 64% and 53%, respectively, for the PLPS group. No significant differences were observed between the 3 groups (Fig. 1B).
At multivariable analysis, phase at presentation, tumor size >10 cm, and surgical margins were independent prognostic factors for DSS (Table 2). With regard to histology, there was weak evidence only for RCLPS compared with MLPS (P = .09), although RCLPS and PLPS had a significantly greater proportion of DM in both subgroups of patients (primary tumors and recurrent tumors), as detailed below. Tumor grade was an independent prognostic factor if it was considered instead of histology (P < .05).
|Category (Reference)||HR||95% CI||P|
|Presentation: recurrence (primary)||2.76||1.66–4.61||.0001|
|Depth: deep (superficial)||1.14||0.36–3.57||.8255|
|Tumor size, cm|
|5–10 (≤ 5)||1.27||0.47–3.40||.6388|
|>10 (≤ 5)||3.56||1.40–9.05||.0075|
|Site: extremity (other)||0.64||0.35–1.17||.1451|
|Margin: positive (negative)||2.00||1.13–3.54||.0180|
|RT: yes (no)||0.96||0.58–1.58||.8598|
|CT: yes (no)||1.41||0.77–2.58||.2666|
Eighty-six patients developed LR after they underwent surgery at our institute, including 75 first events (33 patients with primary tumors and 42 patients with LR at presentation). The time to first LR in the whole series varied from 2 months to 151 months (median for those who developed a recurrence, 21 months). For patients with MLPS, the time to first LR varied from 3 months to 151 months (median for those who developed a recurrence, 25 months); for patients with RCLPS, the time to first LR varied from 4 months to 86 months (median for those who developed a recurrence, 15 months); and, for patients with PLPS, the time to first LR varied from 2 months to 69 months (median for those who developed a recurrence, 14 months).
The overall CCI rates for LR were 21.7% at 5 years and 25.2% at 10 years. In the patients who had primary tumors, the LR CCI rates were 13.2% and 17.5% at 5 years and at 10 years, respectively; whereas, in the patients who had recurrent tumors, the respective rates were 37.1% and 39.2%, and a significant difference was obselrved between the 2 groups (P < .001).
Among the patients who had primary tumors, the 5-year and 10-year LR CCI rate was 11.3% and 15.9%, respectively, for the MLPS group; 13.7% and 21.9%, respectively, for the RCLPS group; and 19.4% and 19.4%, respectively, for the PLPS group. No significant difference was observed between the 3 groups (Fig. 2A).
Among the patients who had recurrent tumors, the 5-year and 10-year LR CCI rate was 35.3% and 37.1%, respectively, for the MLPS group; 31.3% and 31.3%, respectively, for the RCLPS group; and 47.8% and 52.2%, respectively, for the PLPS group. No significant difference was observed between the 3 groups (Fig. 2B).
In the multivariable analysis, phase at presentation, tumor site, and radiotherapy were independent prognostic factors for LR-free survival (a borderline P value of .05 was achieved for radiotherapy), as shown in Table 3. With regard to surgical margins, only a weak impact was observed (P = .07).
|Category (Reference)||HR||95% CI||P|
|Presentation: recurrence (Primary)||2.67||1.72–4.16||<.0001|
|Depth: deep (superficial)||1.00||0.47–2.14||.9912|
|Tumor size, cm|
|>10 ≤ 5)||1.43||0.72–2.82||.3035|
|Site: extremity (other)||0.38||0.23–0.64||.0002|
|Margin: positive (negative)||1.67||0.95–2.92||.0758|
|RT: yes (no)||0.39||0.23–0.65||.0003|
|CT: yes (no)||1.64||0.92–2.92||.0913|
Fifty-eight patients had DM, including 46 first events, after the underwent surgery at our institute (21 patients with primary tumors and 25 patients with LR at presentation). The time to DM in the whole series varied from 2 months to 187 months (median for those who developed a recurrence, 20 months). For patients with MLPS, the time to DM varied from 3 months to 73 months (median time for those developed a recurrence, 23 months); for patients with RCLPS, the time to DM varied from 2 months to 140 months (median for those who developed a recurrence, 18 months); and, for patients with PLPS, the time to DM varied from 5 months to 187 months (median for those who developed a recurrence, 34 months).
The overall CCI rate for DM was 14.5% at 5 years and 15.3% at 10 years. In the patients with primary tumors, the DM CCI rates were 10.0% and 10.7% at 5 years and 10 years, respectively; whereas, in the patients who presented with recurrent tumros, the DM CCI rates were 22.4% and 23.4%, respectively, and a significant difference was observed between the 2 groups (P < .05).
Among the patients who presented with primary tumors, the 5- and 10-year DM CCI rates were 4.0% and 5.0%, respectively, for the MLPS group; 21.7% and 21.7%, respectively, for the RCLPS group; and 20.6% and 20.6%, respectively, for the PLPS group. A significant difference was observed between the 3 groups (Fig. 3A).
Among the patients who presented with recurrent tumors, the 5- and 10-year DM CCI rates were 18.0% and 18.0%, respectively, for the MLPS group; 43.2% and 43.2%, respectively, for the RCLPS group; and 13.0% and 17.4%, respectively, for the PLPS group. A significant difference was observed between the 3 (Fig. 3B).
On multivariate analysis, phase at presentation, tumor size >10 cm, and radiotherapy were independent prognostic factors for DM-free survival, as shown in Table 4. Histology also was identified as an independent prognostic factor for DM in addition to tumor grade if it was considered in the multivariable analysis.
|Category (Reference)||HR||95% CI||P|
|Presentation: Recurrence (primary)||2.78||1.62–4.76||.0002|
|Depth: Deep (superficial)||0.94||0.34–2.60||.9025|
|Tumor size, cm|
|Site: Extremity (Other)||1.45||0.65–3.22||.3643|
|Margin: Positive (negative)||0.94||0.49–1.82||.8621|
|RT: Yes (No)||1.69||0.99–2.86||.0529|
|CT: Yes (No)||1.44||0.80–2.62||.2279|
Overall, 58 patients developed DM. Fifteen patients (26%) patients had only lung metastases, 24 patients (41%) had only extrapulmonary metastases, and the remaining 19 patients (33%) patients had both lung and extrapulmonary metastases. Metastatic patterns at single extra-pulmonary sites are detailed in Table 5.
|Regional lymph nodes||1||0/1||1||0/1||—||—||2||0/2|
|Patients with multiple extrapulmonary sites||8||4/4||6||5/1||—||—||14||9/5|
Extrapulmonary metastases affected 49% of patients with MLPS, 43% of patients with RCLPS, and only 8% of patients with PLPS. In 14 patients (33%), multiple sites of either synchronous or metachronous extrapulmonary metastases were documented. Therefore, 43 patients (74% of the patients with metastases) had at least 1 extrapulmonary metastatic site. The extrapulmonary sites that were affected most frequently were abdomen, bone, paraspinal soft tissues, and extremities.
The median survival after pulmonary metastases developed was approximately 2 years for patients with MLPS and 1 year for patients with RCLPS and PLPS. Conversely, the median survival after extrapulmonary metastases developed as a first event was 14 months for patients with MLPS (range, 4 months to >52 months). For patients with RCLPS and PLPS, the limited number of extrapulmonary metastasis as a first event prevented us from calculating the median survival.
In this series of 329 patients with MLPS, RCLPS, and PLPS of all anatomic sites who underwent surgery with intent to eradicate disease at our institution over 25 years, the cumulative incidence of LR and DM at 10 years was 25.2% and 15.3%, respectively. DSS at 10 years after definitive surgery was 77% for the MLPS/RCLPS group. These results are similar to those reported from the main published series (Table 6). Nevertheless, when evaluating the results according to the different histotypes, some differences could be observed. The Memorial Sloan-Kettering Cancer Center (MSKCC) group recently reported a series that included all primary LPS subtypes. In their report, the 12-year DSS rate was 86% for MLPS, 55% for RCLPS, and 53% for PLPS.16 In the current series, the 10-year DSS rate was 90% for primary MLPS, 81% for RCLPS, and 82% for PLPS. This difference in outcomes for both RCLPS and PLPS mainly may reflect differences in the referral pattern to these 2 major national referral centers. The median tumor size in our series was definitively smaller than that reported in the MSKCC series (9 cm vs 15 cm). Furthermore, the cut-off level of a 5% of round cell component to differentiate MLPS from the more aggressive RCPLS variant probably cannot single out a group of patients with uniform risk. Therefore, our patients may have been affected by a disease with a lesser round cell component on average and, thus, with a subsequently more favorable outcome. A better risk stratification in the RCLPS group should be investigated prospectively.
|Series||No. of patients||Age, y||Site||Median size, cm||MLPS, %||RCLPS, %||LR, %||DM, %||Extrapulmonary M, %||DSS, % (y)|
|Smith et al, 19962||29||31||Extremities||14||76||24||31||34||NS||47 (10)*|
|Kilpatrick et al, 199613||86||21||Any||11||57||43%||14||35||NS||67 (10)*|
|Antonescu et al, 200114||79||17||Any||NS||57||43||28||38||67||73 (5)|
|ten Heuvel et al, 200615||49||27||Any||12||83||17||33||27||77||72 (10)|
|Moore Dalal et al, 200616||144||23||Any||NS||100||—||NS||NS||NS||86 (12)|
|Moore Dalal et al, 200616||81||23||Any||NS||—||100||NS||NS||NS||55 (12)|
|Current series||269†||25||Any||9||75||25||24||14||74%||77 (10)|
We also analyzed a number of factors that had a possible influence on the clinical outcome of patients with these tumors. Presentation with either primary or recurrent disease, tumor size, tumor grade, and surgical margin status were identified as the most important prognosticators for cause-specific survival in the current study. Patients who presented to our institute with an LR had worse outcomes compared with patients who presented with primary disease. This negative impact of presentation was significant for all 3 of the endpoints that we investigated. This difference well may reflect the selected subset of patients, but it also suggests that patients with LR need to be followed even more carefully, because they bear a 2.8-fold greater risk of disease mortality.
The impact of tumor size was more evident for tumors that measured >10 cm. The cut-off size of 5 cm did not seem to single out 2 different risk categories. Patient who had tumors that measured >5 cm and <10 cm had a limited increased risk compared with patients who had tumors that measured >10 cm, who bore a 3.5-fold greater risk of dying from their disease.
Tumor grade was identified as a significant factor for survival in the multivariable analysis. The difference was observed mainly between grades I and II, whereas a much weaker difference was observed between grades II and III. This may reflect the pathologic characteristics of the disease. MLPS tumors always were graded as grade I, whereas RCLPS and PLPS tumors were graded II or III. Therefore, the difference in outcome according to tumor grade probably reflects the difference in outcome according to the histologic variant. The presence of a >5% round cell component resulted in a >1.5-fold greater risk of dying from disease. Different from what has been reported by others, this difference did not correspond to the rates for high-grade tumors. In the current series, RCLPS behaved similarly to intermediate-grade tumors, and patients with RCLPS had long-term survival rates between 70% and 80%.
Patients who had positive surgical margins bore a 2-fold greater risk of dying from their disease and a 1.7-fold greater risk of developing an LR compared with patients who did not have positive surgical margins. The same difference was not observed for patients who had metastases. Therefore, patients who had positive surgical margins had a greater risk of dying mainly from locoregional disease independent of tumor location, as reported in other series of patients with softtissue sarcoma.17–21
Two more factors had an affect on recurrence-free survival in our series: tumor location and the administration of radiation therapy. Patients who had tumors of the extremities had more favorable local control rates compared with patients who had tumors located at other sites. Tumors located in the trunk bore a nearly 3-fold greater risk of recurring locally compared with tumors located in the extremities. This difference in local outcome may reflect the impact of tumor location on the ability to perform resection with adequate surgical margins, as reported in many series on patients with retroperitoneal and trunk soft tissue sarcoma,22, 23 who also had a nearly 2-fold greater risk of dying of their disease, mainly from locoregional failure. Patients who received radiation therapy had approximately 50% of the risk of developing an LR compared with patients who did not receive radiation therapy.
With regard to the histologic subtype, a major difference was observed for the development of DM. Patients with primary MLPS had a 10-year incidence of DM of 5%, whereas the incidence of DM was 22% and 21% for patients with RCLPS and patients with PLPS, respectively. Only minor differences in 10-year survival were observed, reflecting the indolent course of the disease even in the metastatic setting and the relative greater sensitivity of this disease to chemotherapy.8, 9 Furthermore, surgical management of extrapulmonary metastases often was carried out, although it sometimes was more challenging in patients with lung nodules, especially with regard to particular anatomic sites (eg, abdomen, mediastinum, and heart/pericaridium) (Table 5).24, 25
In conclusion, we observed that patients with MLPS, RCLPS, and PLPS who had a confirmed have an indolent course, as reported previously using predicting tools and based on large case series analyses,26–28 Presentation with either primary or recurrent disease, tumor size, and tumor grade are of major importance. Complete tumor resection with negative margins should be the objective of surgery with an evident impact on local control. Adjuvant radiation therapy should be delivered to improve local control. High-grade and large RCLPS tumros have a more aggressive behavior; thus, patients with these tumors may be considered for possible new adjuvant medical treatments. The cutoff level for tumors of a 5% round cell component singles out a group of diseases with a broad spectrum of aggressiveness. Further studies will be needed to determine whether more precise tools may be obtained by stratifying the round cell cutoff level further or by examining the different molecular subtypes, allowing a better selection of high-risk patients as candidate for newly available agents.