Minimal Increase in Survival Throughout the Years in Patients with Soft Tissue Sarcoma with Synchronous Metastases: Results of a Population‐Based Study

A population‐based analysis was performed to determine the effect of changes in treatment policies on overall survival in patients with soft tissue sarcomas who presented with metastatic disease at time of diagnosis and to establish whether the survival rate had improved over the years.

Abstract

Background.

Treatment options for patients with metastatic soft tissue sarcoma (STS) have increased in the last decade. We aimed to examine whether this is associated with improved overall survival (OS) in patients with STS with synchronous metastases.

Patients and Methods.

Patients diagnosed with STS and synchronous metastases from 1989 to 2014 were queried from The Netherlands Cancer Registry. Trends in OS were assessed by the Kaplan‐Meier method and log‐rank test in time intervals of 5 years, for the whole study population and in subgroups for liposarcomas, leiomyosarcoma, and other STS subtypes. A multivariable Cox regression analysis was performed to identify characteristics prognostic for OS.

Results.

Median OS of the 1,393 identified patients did not improve significantly over the years from 5.8 months in 1989–1994 to 8.1 months in 2010–2014, but there was an evident trend. Median OS was prolonged in the subgroups of liposarcomas (3.6 to 9.3 months), leiomyosarcomas (11.3 to 14.6 months), and other STS subtypes (5.7 to 6.3 months), although there were no significant improvements in OS over the years. Primary tumor site in one of the extremities and surgery in an academic center had a favorable effect on OS, whereas significant negative predictors were no treatment, elderly age, STS subtype other than liposarcoma or leiomyosarcoma, high or unknown grade, and nodal involvement.

Conclusion.

Although overall survival of patients with STS with synchronous metastases in this nationwide and “real‐life” population has improved over the years, the improvement was not statistically significant, despite new treatment options.

Implications for Practice.

Treatment of patients with metastatic soft tissue sarcoma (STS) has changed in the past years, with new drugs such as trabectedin (2007) and pazopanib (2012) becoming available. By using data from the nationwide Netherlands Cancer Registry, the impact of these changes in treatment policies on survival is analyzed in a “real‐life” population of patients with STS with synchronous metastases, rather than in a strictly selected trial population. Unfortunately, overall survival improved only minimally and not significantly for these patients diagnosed from 1989 to 2014. Hopefully, the advent of novel treatment options, such as eribulin and olaratumab, will further improve the outcome of this patient group.

Introduction

Soft tissue sarcomas (STSs) are a group of rare and heterogeneous tumors of mesenchymal origin, comprising over 50 different histological subtypes and accounting for approximately 1% of all adult malignancies [1], [2]. Roughly 650 to 700 new patients are diagnosed annually with STS in The Netherlands, with a slightly increasing incidence over the years [3]. Leiomyosarcomas and liposarcomas form the two most prevalent subtypes, each representing approximately 20% of patients with STS [1]. When STS is diagnosed at an early stage and complete resection of the tumor can be carried out, cure may be achieved in up to 90% of patients [4], [5], [6]. Unfortunately, cure is generally not attainable in case of metastatic disease, a situation for which only palliative treatment remains. Most of these cases occur as a relapse after primary treatment, which may take several years of follow‐up. Approximately 10% to 15% of patients present with synchronous metastases, that is, metastases that are diagnosed before or simultaneously with the primary tumor [7].

In the majority of patients with advanced STS, the disease cannot be cured, and only treatment with palliative intent remains. However, in selected cases with oligometastatic disease, mostly isolated and solitary lung metastases, long‐term survival and in rare cases even cure can be achieved by surgical resection of the metastasis, also called metastasectomy [8], [9], [10], [11]. Besides surgical resection, other local treatment options can be applied. Examples include radiofrequency ablation, isolated limb perfusion and (stereotactic) radiotherapy [12]. In contrast to metastasectomy of solitary (lung) metastases, these treatment options are used rarely with therapeutic intent but rather with a palliative intent. Although they might prolong remission or prevent or slow down progression [13], these treatment modalities are usually used to reduce symptoms and thereby improve quality of life.

In recent years, the number of systemic palliative treatment options for patients with metastatic STS has increased. Whereas doxorubicin‐based chemotherapy is the mainstay of first‐line treatment, mostly as a single agent and sometimes in combination with ifosfamide [14], [15], two new agents have been approved in the last decade for patients in whom doxorubicin‐based chemotherapy fails or is unsuitable. In The Netherlands, the alkylating agent trabectedin became available in 2007 for adults who have advanced STS and who fail on treatment with anthracyclines and/or ifosfamide or who are unsuited to receive these agents. Although registered for all STS subtypes, it is mostly applied in patients with liposarcomas or leiomyosarcomas, as efficacy was proved most pronounced in these entities [16], [17]. Secondly, the tyrosine kinase inhibitor pazopanib was introduced in 2012. In the Netherlands, it is registered for advanced STS after prior chemotherapy for metastatic disease or advanced STS with progressive disease within 12 months after neoadjuvant/adjuvant therapy, irrespective of subtype except for patients with liposarcoma [18], [19]. Besides these two new agents, other chemotherapeutic drugs are increasingly used in daily clinical practice, although these have not been formally registered for STS. Examples include gemcitabine in leiomyosarcoma and taxanes in angiosarcoma, which are also mentioned in international guidelines [13], [20]. These drugs can be used in multiple lines of therapy or in combination with other locoregional treatment options [13].

For this study, a population‐based analysis was performed to determine the impact of the changes in treatment policies on overall survival in patients with STS presenting with metastatic disease at time of diagnosis (also known as synchronous metastases) and to establish whether the survival has improved over the years. Because our data source, the nationwide database of The Netherlands Cancer Registry (NCR), does not include information on patients who relapsed after initial treatment for nonmetastatic disease, the focus in this study is on the 10% to 15% of patients with soft tissue sarcoma with synchronous metastases.

Materials and Methods

Data Collection

From the NCR, data on all patients with STS diagnosed with synchronous metastases between 1989 and 2014 were identified and extracted. Synchronous metastases were defined as metastases detected prior to or during screening in the diagnostic workup before start of (neoadjuvant) treatment of the primary tumor. Children (age at diagnosis <18 years), gastrointestinal stromal tumors (GIST) and small blue round cell tumors (SBRCTs: Ewing's sarcomas, mesenchymal chondrosarcomas, peripheral neuroectodermal tumors, and rhabdomyosarcomas) were excluded because of the different tumor biologies, different treatment regimens, and outcomes.

Information in the NCR on patient characteristics, disease characteristics, and primary treatment was retrieved from patients' medical records by trained registration employees of The Netherlands Comprehensive Cancer Organisation (IKNL). Follow‐up information on vital status was obtained through yearly linkage with the Municipal Personal Records Database (MPRD). For our study, the last linkage was performed in February 2016.

Primary tumors were staged according to the TNM classification and STS subtypes were classified following the World Health Organization classification [2], [21]. No central pathology review was performed. Primary tumor site as well as metastatic site(s) and STS subtype were coded under the International Classification of Diseases of Oncology topography and morphology codes, respectively. STS subtypes that are acknowledged for exhibiting aggressive behavior but usually not graded (malignant peripheral nerve sheet tumors, angiosarcoma, extraskeletal chondrosarcoma, epithelioid sarcoma, clear cell sarcoma not otherwise specified, alveolar soft part sarcoma) were pooled with grade III tumors and classified as high‐grade tumors. Grade I and II tumors were pooled and classified as low‐grade tumors. For the study period, the NCR database lacked information on the type of systemic therapy (cytostatic drugs, kinase inhibitors, etc.) or intent of surgery (resection of primary tumor, palliative debulking, etc.). In addition, no data were available on patients' performance score or comorbidities.

The study was performed in accordance with local ethics committee guidelines and national legislation.

Statistical Analysis

Median follow‐up time with corresponding interquartile range (IQR) was calculated by the reversed Kaplan–Meier method [22]. Overall survival (OS) was defined as time in months between diagnosis (first pathological confirmation) and death or last follow‐up. Patients alive at date of last linkage to the MPRD were censored. To assess trends in overall survival over the study period, the data were analyzed in intervals of 5 years, using the Kaplan–Meier method. Overall survival was assessed for the total study population, as well as in subgroups for patients with liposarcomas, leiomyosarcomas, and other STS subtypes. Survival times are noted as median survival times in months, together with the corresponding IQRs. Differences between subgroups were tested by the log‐rank test. An additional survival analysis focusing on patients who received systemic therapy was performed to explore the effect of the new agents. Because of small subgroups, liposarcomas and leiomyosarcomas were combined (“L‐sarcomas”) for this analysis.

A multivariable Cox proportional hazards regression analysis for OS was performed to identify relevant patient, tumor or treatment related characteristics. Also, we explored possible influences on survival of the type of hospital (academic vs. nonacademic) where patients were diagnosed and treated. Factors that tested significantly at an α‐level of 0.05 in univariable analyses were included in the multivariable Cox regression analysis. The definitive model was obtained with a backward stepwise elimination method. Results are described as hazard ratios (HRs) with corresponding 95% confidence intervals (95% CI). P values ≤.05 were considered statistically significant. SPSS was used for the statistical analyses (SPSS Statistics for Windows; IBM, Armonk, NY).

Results

Between 1989 and 2014, 1,689 patients were diagnosed with STS and synchronous metastases, including children and those diagnosed with GIST and SBRCTs, representing roughly 12% of all patients diagnosed with STS. Over the years, an increase in incidence was noticed from 52 (10.2%) to 97 (14.4%) patients a year (Fig. 1). After exclusion of all children and patients with GISTs and SBRCTs, 1,393 patients with STS and synchronous metastases were included in the analyses. There was a slight male predominance, and the trunk was the primary tumor site in approximately half of the patients (Table 1). The most common localizations of metastases were the lungs (42.9% of patients), liver (13.3%), bones (11.6%), and lymph nodes (7.4%). In 30.8% of the patients, the sites of metastases were unknown or unspecified.

Figure 1.

Incidence of STS and metastatic STS at time of initial diagnosis in The Netherlands. The blue fractions of the bars represent patients with nonmetastatic/localized STS. The red fractions and the numbers above the bars represent the proportion of patients with metastatic STS at initial diagnosis.

Abbreviation: STS, soft tissue sarcoma.

Table 1. Characteristics of all 1,393 patients with soft tissue sarcoma with metastatic disease at time of initial diagnosis.

Abbreviations: CTx, chemotherapy; FU, follow‐up; IQR, interquartile range; NA, not applicable; NOS, not otherwise specified; RTx, radiotherapy.

Trends in Survival

Median OS over the whole period and for all subtypes was 6.3 months (IQR 2.4–15.5). Throughout the years, median OS did not improve significantly; it increased from 5.8 months (IQR 2.3–14.8) in 1989–1994 to 8.1 months (IQR 2.7–17.1) in 2010–2014 (p = .095), although there was an evident trend (log‐rank trend test, p = .015; Fig. 2A).

Figure 2.

Overall survival of all patients with STS with synchronous metastases per 5‐year time intervals (A), specified for liposarcomas (B), leiomyosarcomas (C), and other STS subtypes (D). p values were calculated by the log‐rank test. For (A), an additional log‐rank trend test was performed, showing a significant trend over the years (p = .015).

Abbreviations: IQR, interquartile range; STS, soft tissue sarcoma.

When analyzing the different STS subtypes, median OS for patients with liposarcoma did not change significantly from 3.6 months (IQR 1.7–18.5) in 1989–1994 to 9.3 months (IQR 3.8–28.8) in 2010–2014 (p = .180; Fig. 2B). Neither did the median OS for patients with leiomyosarcomas improve significantly (11.3 months, IQR 3.5–19.5, to 14.6 months, IQR 5.5–21.0, p = .449; Fig. 2C). Also, for patients with one of the other STS subtypes, median OS remained stable from 5.7 months (IQR 2.1–12.7) in 1989–1994 to 6.3 months (IQR 2.2–13.5) in 2010–2014 (p = .559; Fig. 2D).

In our study population, almost one third of patients did not receive any treatment. This subgroup had a poor median OS of 2.1 months (IQR 0.9–5.8) compared with 9.5 months (IQR 4.2–20.0) for patients who received any type of treatment (p < .001; Table 2). Among the latter group, those who underwent multidisciplinary treatment had a better OS, and patients treated with both radiotherapy and surgery had the most favorable median OS (19.9 months, IQR 8.0–45.2).

Table 2. Median overall survival in months of patients with metastatic STS at initial diagnosis per (combination of) treatment modality.

Combinations of treatment modalities can be in any order, i.e., RTx‐CTx is pooled with CTx‐RTx, surgery‐RTx is pooled with RTx‐surgery, etc. p values were calculated by using the log‐rank test (p < .0001).

Abbreviations: CTx, chemotherapy; IQR, interquartile range; OS, overall survival; RTx, radiotherapy.

A subanalysis of the subgroup of patients who received chemotherapy (n = 503) was performed to explore the effect of the new systemic therapies. Median OS over the total period in the chemotherapy subgroup was 10.8 months (IQR 5.5–20.4) and improved minimally from 10.5 months (IQR 4.9–17.1) in 1989–1994 to 13.0 months (IQR 5.8–24.3) in 2010–2014 (p = .446; Fig. 3A). In the different STS subgroups, median OS also did not improve for the L‐sarcomas (13.0 months, IQR 4.9–34.5, in 1989–1994 to 18.1 months, IQR 8.4–29.7, in 2010–2014, p = .485, Fig. 3B) or for the other STS subtypes (10.1 months, IQR 4.9–14.1, in 1989–1994 to 10.6 months, IQR 4.5–20.8, in 2010–2014, p = .789; Fig. 3C).

Figure 3.

Overall survival of all patients with STS with synchronous metastases receiving chemotherapy per 5‐year time interval (A), specified for L‐sarcomas (liposarcomas and leiomyosarcomas) (B) and other subtypes of metastatic STS (C). p values were calculated by the log‐rank test.

Abbreviations: IQR, interquartile range; STS, soft tissue sarcoma.

Prognostic Factors for Overall Survival in Metastatic STS at Initial Diagnosis

In the univariable Cox regression analyses almost all factors tested significantly, except for the variables gender, socioeconomic status, pulmonary metastases, and the annual volume of the hospital where patients received their (first‐line) chemotherapy (Table 3). In multivariable analysis seven factors remained independently prognostic (Table 4). Whereas an elderly age, STS subtype other than liposarcoma or leiomyosarcoma, high or unknown grade, and nodal involvement had a negative effect on survival, a primary tumor located in the upper or lower extremity, any type of treatment (chemotherapy, radiotherapy, and/or surgery), and undergoing surgery in an academic center (compared with a nonacademic center) had a favorable effect on survival.

Table 3. Results of univariable Cox proportional hazards regression analysis for overall survival of patient‐, tumor‐, and treatment‐related characteristics.

^aVariables with p value < .05 are included in multivariable analysis.

^bTreatment modalities used, in any order (i.e., CTx and RTx can be first CTx followed by RTx, but also RTx first followed by CTx).

Abbreviations: CI, confidence interval; CTx, chemotherapy; HR, hazard ratio; NA, not applicable; Nx, lymph nodes not assessed, unknown involvement; RTx, radiotherapy.

Table 4. Results of multivariable Cox proportional hazards regression analysis for overall survival of patient‐, tumor‐, and treatment‐related characteristics.

^aVariables significant at an α‐level of .05.

^bTreatment modalities used, in any order (i.e., CTx and RTx can be first CTx followed by RTx, but also first RTx followed by CTx).

Abbreviations: CI, confidence interval; CTx, chemotherapy; HR, hazard ratio; NA, not applicable; Nx, lymph nodes not assessed; RTx, radiotherapy.

Discussion

Several new therapeutic agents and/or regimens for the treatment of patients with advanced STS have been introduced in the last decade, including trabectedin in 2007 and pazopanib in 2012. Despite these new options, the overall survival of patients with STS with synchronous metastases has improved only minimally and not statistically significantly over the years.

As probably only a small proportion of patients received one of the new agents, the survival benefit of trabectedin reported in several trials apparently has not translated to patients with advanced disease at initial diagnosis on a population level [16], [17], [23], [24], [25]. A possible contributing factor could be the different composition of STS subtypes included in our study population compared with the study populations of clinical trials. For instance, the efficacy of trabectedin was most pronounced in the L‐sarcomas and especially myxoid liposarcomas, and whereas in most trabectedin trials 50% to 100% of the included patients had liposarcoma or leiomyosarcoma, only approximately 35% of the patients in our study population were diagnosed with an L‐sarcoma, of whom only 3.0% had myxoid liposarcoma.

In line with the results from the PALETTE trial, in which a significant difference in progression‐free survival but not in overall survival was observed between the pazopanib and placebo arm [19], we did not observe a significant difference in overall survival. It must be noted that a possible beneficial effect might be masked because of short follow‐up, with 2012 being the year of introduction of pazopanib [26].

Throughout the years, we observed a median overall survival across all STS subtypes of 6.3 months, which is poorer than reported in clinical trials for patients with metastatic STS, in which median OS times of 12 to 24 months have been described [9], [27], [28], [29], [30], [31], [32], [33], [34], [35]. However, these studies have included only patients who received any kind of treatment, whereas this study also included a substantial number of patients who did not receive any treatment. As expected, patients who did not receive therapy, probably because of a poor performance status or multiple comorbidities, have a poorer prognosis than patients receiving (any kind of) treatment. Unfortunately, no data on comorbidities or performance status were available to confirm this hypothesis or to correct for possible interactions in our analyses. When focusing only on the group receiving any kind of treatment, we observed a median OS of 9.5 months, which is still poorer than described in literature. It is likely that the patients who received treatment in our cohort did not meet the strict eligibility criteria that patients included into clinical studies have to fulfil. Remarkably, patients who made it to a combination of local treatment only (i.e., surgery and radiotherapy) had the most favorable median OS of 19.9 months. Probably, this group of patients represents the subgroup of patients fit enough to undergo both these treatment modalities, but also a subgroup with minimal or oligometastatic disease, although the exact intents of surgery and radiotherapy are unknown. Furthermore, because we only focused on patients with STS with metastatic disease at initial diagnosis, and clinical trials generally do not limit their inclusion to patients with synchronous metastases, these cases may represent a different, perhaps more aggressive, subgroup of STS compared with patients who initially present with nonmetastatic localized STS and experience a relapse at a later point in time. Dossett et al. showed that after pulmonary metastasectomy patients with synchronous metastases had poorer median OS than patients with metachronous metastases, with synchronous metastases also being a negative significant prognostic factor in multivariate analyses [36]. These findings may give support to the abovementioned hypothesis of patients with synchronous metastases representing a more aggressive subgroup of patients with STS, resulting in a poorer survival.

Although the current study only focuses on patients with STS with synchronous metastases, one of its strengths is that it shows survival in a “real‐life” population rather than in a strictly selected trial population. Many patients are excluded from clinical trials by strict eligibility criteria but nonetheless receive these drugs when available in routine care. This, in combination with the large number of patients included in the cohort, makes it likely that this study gives a reliable reflection and accurate estimation of the “real” survival in daily clinical practice of patients with STS with synchronous metastases.

A limitation of this study is that for the patients who received chemotherapy, it is not specified which types of drugs were administered, how many lines of treatment patients received, and whether these patients received (one of) the new agents. Second, health care in general has improved over the years, and sarcoma care in The Netherlands has been largely centralized in five designated sarcoma centers [1]. Additionally, the “Will Rogers phenomenon” might have an effect [37], [38]. As mentioned, the incidence of metastatic disease at initial diagnosis increased slightly over the years, which is likely to be explained by better imaging techniques and thereby the ability to detect smaller metastases. Subsequently, it might be possible that patients with minimal metastatic disease in former years have been categorized as having localized/nonmetastatic STS, and they theoretically perform worse than “true” localized STS, but in later periods, because of advancements in imaging, they are categorized as having metastatic STS, and they theoretically do better compared with other patients with (more extensive) metastatic STS. In this way, survival improves in both groups. Therefore, the trend in improvement of survival cannot completely be attributed to the new drugs alone.

Finally, the NCR started to register all patients with cancer in 1989. In this period, GIST was not yet recognized as a distinct entity, and most of these nonepithelial gastrointestinal tract tumors were classified as leiomyosarcomas. It was not until the late 1990s, after discovery of the cell of origin [39], the presence of c‐KIT proto‐oncogene mutations [40], and effectiveness of imatinib in these tumors [41], [42], that GIST was distinguished and treated as a separate entity. Therefore, it cannot be ruled out that in the earlier years of registration some of the registered leiomyosarcomas in fact were GISTs.

Recently, in 2016 another two new agents were registered for advanced or metastatic STS, thereby expanding the limited amount of (palliative) treatment options even more. One of these agents is eribulin, which in comparison with dacarbazine significantly improves OS in patients with advanced liposarcoma who received at least two systemic treatment regimens (including an anthracycline) [43]. The second new drug is olaratumab, a PDGFRα‐inhibitor. It has been conditionally approved, pending the results of the phase III trial, and temporary access for the treatment of adults with advanced STS not amenable to curative treatment (with surgery and/or radiotherapy) has been established. It is used in combination with doxorubicin as first‐line treatment, improving median OS by almost a year compared with doxorubicin alone in a randomized phase II study [44]. Although only a few trials have been conducted with these new drugs, they seem promising and hopefully they will increase the survival of patients with metastatic STS further.

Conclusion

Despite new treatment options and improved health care, overall survival of patients with STS and synchronous metastases treated in ‘real’ life has improved only minimally and not statically significantly over the years. Nonetheless, the relatively small increase of a few months in survival might entail a valuable difference for individual patients. Hopefully, the advent of novel treatment options, such as eribulin and olaratumab, will further improve the outcome of this patient group.

Author Contributions

Conception/design: Melissa Vos, Cornelis Verhoef, Stefan Sleijfer

Provision of study materials or patients: Vincent K. Y. Ho

Collection and/or assembly of data: Vincent K. Y. Ho

Data analysis and interpretation: Melissa Vos, Vincent K. Y. Ho, Astrid W. Oosten, Stefan Sleijfer

Manuscript writing: Melissa Vos, S. Sleijfer, Vincent K. Y. Ho, Astrid W. Oosten, Cornelis Verhoef

Final approval of manuscript: Melissa Vos, Vincent K. Y. Ho, Astrid W. Oosten, Cornelis Verhoef, Stefan Sleijfer.

Disclosures

The authors indicated no financial relationships.