Multimodality Treatment of Desmoplastic Small Round Cell Tumor: Chemotherapy and Complete Cytoreductive Surgery Improve Patient Survival

Desmoplastic small round cell tumor (DSRCT) is an aggressive malignancy that occurs predominantly in young adult males and is characterized by abdominopelvic sarcomatosis exhibiting multilineage cellular nests of epithelial, muscular, mesenchymal, and neural differentiation admixed with desmoplastic stroma (1–6). Owing to the disease's low age-adjusted incidence and peak incidence of 0.3 and 0.74 cases per million, respectively (7), it was not until 1989 that Gerald and Rosai recognized a pathognomonic EWSR1-WT1 t(11;22)(p13:q12) chromosomal translocation that pairs the Ewing sarcoma (ES) gene (EWSR1) with the Wilm's tumor suppressor gene (WT1; ref. 3). The functional loss of the WT1 tumor suppressor protein and the oncogenic effects caused by the aberrant 59 kDa fusion protein results in hundreds to thousands of nodules coating the intraabdominal serosal and subdiaphragmatic surfaces. Most patients with DSRCT present with advanced disease and symptoms of abdominal pain or distention, nausea, constipation, and/or weight loss (8).

Prior to the recognition of the disease as a distinct clinical entity, DSRCT was invariably misclassified as poorly differentiated atypical cancer of the testes, ovary, mesentery, or gastrointestinal tract, and the chemotherapies used for those malignancies elicited poor clinical response. As previously reported (5, 9), a tectonic shift in the treatment of these patients occurred after researchers made two astute observations: (i) DSRCT microscopically resembles other small round “blue cell” sarcoma subtypes (e.g., ES, rhabdomyosarcoma, synovial sarcoma), and (ii) DSRCT and ES have the same N-terminal EWSR1 fusion partner (10, 11). A fuller historical context of these discoveries is summarized in a recent review by Mora and Rosai (12).

The morphological and genomic characterization of DSRCT as distinct sarcoma subtype has provided a relatively recent opportunity to systematically evaluate which treatment modalities yield clinical benefits in DSRCT. Yet, the relative rarity of this malignancy has contributed to the lack of standardization, and most literature to date is in the format of anecdotal case reports. As a tertiary cancer hospital that cares for one of the largest groups of patients with DSRCT in the world, our multidisciplinary team of medical, pediatric, surgical, and radiation oncologists have reviewed our entire clinical experience treating DSRCT since 1990 and, herein, share our seminal findings. As will be discussed, although carefully sequenced trimodality therapy reliant upon chemotherapy, complete cytoreductive surgery (CCS), and WART represents the standard DSRCT treatment approach used at our institution and others, our analysis suggests the current approach should undergo major revision, or at least prompt prospective clinical investigation.

The patient treatments and study analysis were reviewed and approved by the Institutional Review Board of The University of Texas MD Anderson Cancer Center (MDACC) and conducted in accordance with the principals of Declaration of Helsinki. Written informed consent was obtained from all patients for treatments, but it was not required for this retrospective review, and all the patient records were de-identified prior to analysis.

We identified 187 patients with DSRCT treated at MDACC from 1990 to 2016 and collected data in accordance with the guidelines of MD Anderson's Institutional Review Board. Specialist pathologists used clinical information, IHC, and cytogenetic analysis for the EWSR1–WT1 fusion to confirm the DSRCT diagnoses. The date of original biopsy was used to calculate each patient's age at initial diagnosis. In the comprehensive analysis of prognostic factors, patients were used to assess the association between prognostic factors and overall survival (OS) time, that is, from the time of diagnosis until the time of death or last follow-up. More comprehensive clinical information was available which allowed for a thorough statistical analysis of a number of factors, including: (a) the era in which a patient was treated, taking into account the advent of the ifosfamide/etoposide combination used more frequently after 2003, (b) the ability to achieve an optimal surgical cytoreduction, and (c) the tumor response to neoadjuvant chemotherapy. A complete tumor surgical debulking was defined as removal of the primary dominant tumor mass and all visualized metastatic nodules if present. Given the difficulty in accurately measuring the extent of abdominal sarcomatosis, a positive neoadjuvant chemotherapy response was defined as any reduction in tumor size, not necessarily a clinical response as would otherwise be reported using WHO or RECIST criteria that are better suited to assess well-demarcated tumors. Major response to chemotherapy was defined using RECIST criteria (greater than or equal to 30% reduction in one-dimensional measurement in the predominant mass) to any course of chemotherapy whether provide in the neoadjuvant, adjuvant, or relapsed setting. As a tertiary referral center, many patients had received chemotherapy and exhibited chemoresistance prior to referral. In those circumstances, external medical records were reviewed to determine the type, length, and dose of chemotherapy. All but four patients received chemotherapy and despite vast treatment heterogeneity most patients received ES-based chemotherapy regimens that were the standard at their respective times of diagnosis.

Most regimens included combinations of vincristine, dactinomycin, cyclophosphamide, doxorubicin, ifosfamide, and etoposide. Temozolomide/irinotecan, cyclophosphamide/topotecan, and high-dose ifosfamide, which have proven clinical activity in chemotherapy-resistant ES, were common second-line regimens. Less common salvage regimens included cyclophosphamide/vinorelbine, gemcitabine/docetaxel, and dacarbazine. A small minority of patients with DSRCT treated since 2007 received insulin-like growth factor 1–targeted monoclonal antibodies alone or in combination with mTOR inhibitors, which was in keeping with early-phase clinical trials that revealed that these therapies had striking, albeit short-lived, clinical activity in patients with ES. Other biologically targeted therapies, such as polyethylene glycol–conjugated interferon and pazopanib, were used infrequently. Finally, as an induction regimen, busulphan was given to a handful of patients who underwent bone marrow transplantation.

Clinical data incorporated information from newer ifosfamide/etoposide-containing regimens (which were used more frequently after 2003), the ability to achieve a complete surgical cytoreduction, and response to neoadjuvant chemotherapy. An extensive discussion about DSRCT staging is beyond the scope of this publication. No prospectively validated staging system for DSRCT exists, and those for soft tissue sarcomas have the unintended effect of identifying most DSRCT as stage 4 disease. Similar to the American Joint Committee on Cancer's staging system, that proposed by Hayes-Jordan and colleagues is based on tumor burden (peritoneal cancer index as defined by Sugerbaker and colleagues³⁹), regional spread (liver, rather than nodal, metastases), and the presence or absence of extra-abdominopelvic metastases.⁴⁰ Although the latter two metrics of extra-abdominopelvic metastases and liver spread were readily assessed, our experience suggests that surgeons, radiologists, and pathologists do not consistently document the peritoneal cancer index score, whose calculation requires that each of 13 anatomical regions be scored from one to three based upon maximal lesion size, and in our opinion, this is too time-consuming to gain widespread traction in clinical practice. We suggest a pragmatic staging alternative that replaces the peritoneal cancer index with a simpler method based on the American Joint Committee on Cancer's practice of recording the maximal tumor diameter of the largest mass. In addition, given its clear utility, fluorodeoxyglucose (FDG) PET/CT should be part of DSRCT staging.

Continuous variables were summarized by their means, standard deviations, and ranges, whereas categorical variables were summarized by their frequencies or percentages. A univariate analysis included demographic information (age and gender), treatment type [radiotherapy, surgery, and stem cell transplantation (SCT)] and treatment era (two analyses of patients diagnosed before 2003 or after 2003). The association between diagnosis time (≤2003 vs. >2003, when more effective chemotherapy regimens became available for pediatric sarcomas) and other relevant clinical variables were assessed using the Fisher exact test. Kaplan–Meier curves were calculated to estimate unadjusted OS. A Cox proportional hazards model was used to evaluate the ability of the covariates to predict OS. Within the multivariate Cox regression model, nonsignificant variables were eliminated in a step-down fashion with a P value cut-off of 0.05. All computations were carried out in SAS 9.4 and S-Plus version 8 software.

The demographic and treatment information of 187 evaluable patients with DSRCT is summarized in Table 1. Consistent with prior reports, an overwhelming majority of patients were male (82.9%). The patients' mean age ± SD was 22.6 ± 10.6 years (range, 0.5–52.9 years), a relatively broad spectrum compared with earlier reports from children's hospitals, which naturally have a younger referral base. Local control modalities included cytoreductive surgery (106 patients; 57%) and whole abdominal radiotherapy (WART, 92 patients; 49.2%). A subset of patients had either hyperthermic intraperitoneal perfusion of chemotherapy (HIPEC) with cisplatin (200 mg/m²) following CCS (82 patients; 43.9%) or SCT (12 patients; 6.4%). Treatment evolved as we gained more experience using HIPEC, radiation, and SCT in patients with DSRCT, and each of these modalities were more commonly used after 2003.

When assessing the effect of treatment modality upon OS, chemotherapy (P < 0.01; HR, 0.12), radiotherapy (P < 0.01; HR, 0.44), and CCS (P = 0.006; HR, 0.59) were statistically significant in the univariate analysis (UVA; Supplemental Table S1). In the subsequent multivariate analysis (MVA), chemotherapy (P < 0.01; HR, 0.17) and radiation (P < 0.01; HR, 0.46) retained statistical significance (Supplemental Table S2). Notably, WART—the predominant form of XRT provided—was not statistically significant when CCS was used as the reference timepoint (Supplemental Table S2). Our results indicate that patients diagnosed after 2003 (P < 0.01; HR, 0.47) had improved survival (Fig. 1A). Most patients (88.2%) were treated after 2003, when results from INT-0091 demonstrated that ifosfamide/etoposide, added to a Vincristine, Adriamycin, Cyclophosphamide (VAC) backbone, improved clinical outcomes for those with localized ES.(13) Hearteningly, we observed a striking time-dependent improvement on 3- and 5-year OS rates that had risen, respectively, to 50.9% and 25.1% after 2003 compared with earlier OS rates of just 31.8% (3-year) and 4.6% (5-year).

As DSRCT almost universally presents with advanced-stage disease and is remarkably chemosensitive, all but four patients in this study received systemic chemotherapy. Because nearly all patients received chemotherapy, it was not possible to prove that chemotherapy was superior to observation. Instead, as an indirect measure of chemotherapy effect, we assessed whether chemotherapy response prolonged patient survival, recognizing that most patients with DSRCT eventually die from their malignancy. Of the 156 patients that had CT and/or FDG-PET/CT imaging available for review, 90.4% achieved clinical benefit (SD/PR/CR) and just over half the patients (51.9%) had a presurgical response (CR/PR) according to RECIST at some point during treatment. OS was significantly better for those who achieved clinical benefit versus no benefit from their initial chemotherapy regimen (Fig. 1B), as could be further stratified by clinical response (Fig. 1C). Consistent with the findings of Schwarz and colleagues, our univariate analysis revealed that patients whose disease responded to chemotherapy in the neoadjuvant setting (P = 0.03; HR, 0.63; Supplementary Table S1) had better OS than patients whose disease did not respond to chemotherapy (14).

Although not previously assessed in DSRCT, our univariate data suggest that dose-dense VAC/IE (VAC/Ifosfamide and Etoposide)-based therapy could be superior to 3-week dosing (68.4% vs. 48.2% response rate; P = 0.05). However, our study was not powered to detect a difference in dosing schedules and the Kaplan–Meier curves failed to reach significance (Fig. 1D). Almost half the patients received variants of the Children Oncology Group's alternating VAC/IE protocol (AEWS-0031), VAI, Europe's VIDE (EURO-EWING99) protocol, or less frequently the P6 protocol when the patients were referred from other cancer centers (Fig. 2A), and no regimen proved superior to another (Supplementary Fig. S1; ref. 9). Nearly nine in 10 patients received neoadjuvant chemotherapy, the standard of care at MDACC, and a small minority first underwent chemotherapy as part of an adjuvant or salvage approach when the initial surgery was performed prior to their arrival at MDACC (Fig. 2B). Most received standard interval therapy every 3 weeks (72.7%) and roughly one-quarter of the patients received dose-dense therapy every 2 weeks (Fig. 2C). The second-line regimens used for salvage therapy is shown in Fig. 2D. Our analysis was underpowered to detect any difference in the response rates to second-line chemotherapies (data not shown). However, salvage therapy was extensively used because 61% of the patients (114 of 187) had metastatic disease at the time of diagnosis (Fig. 2E). A univariate analysis of the number and location of metastatic sites is included in Supplemental Table S3.

Fifty-seven percent of the patients underwent CCS when clinically appropriate, and this group had substantially better OS (P = 0.006; HR, 0.59) than those who were gauged to be inappropriate for CCS (Fig. 3A; Supplementary Table S1). Our data suggest the preoperative imaging is reasonably accurate assessing a patient's suitability for CCS, as 75.5% of the patients achieved the expected surgical outcome by eliminating all macroscopically evident disease [i.e., a completeness of cytoreduction score (CCR) score of 0; 61.3% of patients with CCS] or leaving behind measurable tumor implants less than 2.5 mm in diameter (14.2% of patients). Care should be taken in extrapolating this result, because CCS was performed by only a handful of surgeons at our center.

Importantly, our institutional practice has been to offer patients CCS only if they achieved clinical benefit from neoadjuvant chemotherapy because most chemo-resistant tumors quickly recur. As this approach, naturally biased the CCS-selected patients toward an improved survival outcome, we performed a secondary analysis limited to those who achieved clinical benefit from their initial neoadjuvant chemotherapy regimen. Shown in Fig. 3B, the benefit of CCS persisted in those who had responded to neoadjuvant chemotherapy treatment (P < 0.01). An identical subset analysis was conducted to determine whether CCS improved survival in those whose tumors progressed to their first neoadjuvant chemotherapy regimen, however only eight patients met that criteria (data not shown). Of those, four patients obtained clinical benefit from a second neoadjuvant regimen and one achieved clinical benefit from a third neoadjuvant regimen, rendering them operable candidates. Without more evidence, we cannot routinely advocate for CCS in patients that progress on neoadjuvant chemotherapy.

Seventy-two percent of the patients also underwent HIPEC with cisplatin after CCS in an attempt to eliminate micrometastatic cells inevitably left behind at the time of surgery. A subtle, statistically insignificant trend toward improved survival was observed in the first 2 years following CCS; however, this effect was transitory and did not alter 3- or 5-year OS (Fig. 3C, P = 0.16). The results were unchanged when CCS and HIPEC were secondarily evaluated in patients with abdomen-only disease devoid of hepatic surface implants (Supplementary Fig. S2A–S2C) or within an even more stringent subset of patients that achieved complete cytoreductive resection (CCR0) status from CCS (Supplementary Fig. S3). As this study was underpowered to detect a subtle benefit from HIPEC, if one exists, more definitive results are anticipated later this year from a HIPEC-specific trial that continues to enroll at our institution.

Radiation was commonly used for palliation at the time of tumor recurrence. However, our clinical practice evolved over time, and we increasingly used WART as an adjuvant therapy before consolidative chemotherapy after several publications seemed to indicate that trimodality therapy (i.e., chemotherapy, radiation, and surgery) led to better clinical outcomes than bimodality treatment (i.e., chemotherapy and surgery; ref. 15). Our results supported this trimodality approach when the time of diagnosis was used as the reference timepoint to gauge OS (Supplementary Fig. S4). However, because WART cannot contribute to improved OS before it is provided, we conducted a time-variant analysis of the effect of WART versus no WART, this time using the date each patient underwent their respective CCS as the reference point. Surprisingly, in this more homogeneous subset of patients who had undergone WART shortly after recovering from CCS (n = 69), no benefit was observed (Fig. 3E). Subsequent analysis in patients with abdomen-only disease yielded identical results (Supplementary Fig. S2D). This unexpected result conflicts with our current treatment paradigm and prompted an updated treatment recommendation to consider WART only in highly selected patients that are prospectively monitored in clinical trials specifically designed to quantify the post-WART toxicity and in-field recurrence rate.

In an era where immunotherapy plays a prominent role in treatment for many different cancer types, the use immunotherapies, check-point inhibitors, or SCT and other cellular therapies can also be studied in DSRCT. At the time of this writing, we are unaware of any patients with DSRCT who have responded to immunotherapy, and to the extent one can draw from the sarcoma community's experience with ES and other translocation-positive sarcomas that have otherwise relatively quiescent genomes, current evidence suggests limited clinical effect (16–18). SCT is not standard of care at MD Anderson, and only 12 of 187 patients with DSRCT underwent this treatment modality throughout the almost 2-decade-long experience presented in this work. Nine of the 12 patients with DSRCT who underwent SCT received this before 2006 when this treatment modality was more commonly used and all but one of the patients had an autologous rather than allogeneic SCT, so we have insufficient data to evaluate the effect of SCT in the present era. Nevertheless, rapid advancements in the immunotherapy space provide a tantalizing opportunity to study them in patients combating DSRCT.

Our retrospective study of patients with DSRCT is the largest conducted to date at a single institution and unique for its inclusion of children and adults treated throughout a two-decade long period. The treatment approach taken by our multidisciplinary team has steadily improved (Fig. 4), and significant strides in surgical technique, radiation delivery, supportive care, and novel chemotherapies led to improved time-dependent patient survival. We report the 5-year OS rate of patients with DSRCT is considerably higher than reported in recent case series and now approaches 25.1% when patients receive trimodality care from skilled experts (19–21). Nonetheless, despite clear clinical advancements, most survivors harbor residual disease that requires near-perpetual systemic chemotherapy to control.

Our study was not adequately powered to discern regimen-specific differences in chemotherapy response. The majority of patients received the P6 protocol or a similar ES-oriented regimen (e.g., VID or VAC/IE) as first-line neoadjuvant treatment, and we favor this approach because it typically leads to rapid tumor control and selects chemosensitive patients likely to benefit from CCS (22). High-dose ifosfamide and topoisomerase-containing regimens such as temozolomide/irinotecan or cyclophosphamide/topotecan have proven activity as second- or third-line options, and trabectedin (recently approved for soft-tissue sarcoma) has demonstrated clinical responses in several case reports (23–25). A substantial number of DSRCTs overexpress vascular endothelial growth factor receptor 2, and a handful of patients have had clinical responses to sunitinib (26), sorafenib, or pazopanib (27, 28). Other biological therapies, including anti-ganglioside GD2 antibodies (29, 30), imatinib (31), mTOR inhibitors (32, 33), and a combination of insulin-like growth factor 1 and mTOR inhibitors (34), have shown limited success. As reported by Fine and colleagues, DSRCT highly expresses the androgen receptor (AR), which remains to be investigated as a cancer target (35).

Expectedly, chemotherapy responders exhibited a better OS than nonresponders. To tease apart the confounding effect between chemotherapy response and surgery, we performed a secondary Kaplan–Meier analysis limited to 141 patients who derived preoperative clinical benefit from their first chemotherapy regimen. As hypothesized, CCS performed by surgeons that routinely care for patients with DSRCT was associated with improved the OS of patients whose disease responded to neoadjuvant chemotherapy, solidifying this clinical practice recommendation. Insufficient data existed to determine whether CCS improves the OS of chemotherapy nonresponders and further pursuit of this question would likely require international collaboration by the handful of cancer centers that have surgeons experienced in CCS for DSRCT.

At MDACC, surgeons routinely performed CCS when possible since incomplete resections lead to inferior outcomes and this study validates this approach. HIPEC has been used extensively at MDACC, with the goal of eradicating micrometastatic disease, however our study was underpowered to detect a statistically significant survival benefit. We eagerly await the results from a prospective clinical trial studying this treatment modality (36).

Given the exquisite sensitivity of ES to radiation and the molecular and morphological similarity of ES and DSRCT, radiotherapy for patients with DSRCT has been evaluated and is most often applied as an adjunct to chemotherapy and surgery. First described by Kushner and colleagues in 1996, the treatment approach at Memorial Sloan Kettering Cancer Center relied on conventional two-dimensional WART (30 Gy plus boosts to sites of residual disease; ref. 9). In that initial series (37), 86% of tumors responded at the expense of serious gastrointestinal and hematologic complications.

Recent data using intensity-modulated radiotherapy is associated with considerably less toxicity than conventional WART (38). Nevertheless, despite safer methods to deliver radiation, we were concerned that the toxicity of WART would outweigh its putative benefits. At first pass, our univariate analysis indicated that radiotherapy conferred a survival benefit (P = 0.01; HR, 0.44) when analyzed from time of diagnosis. Because radiotherapy was given almost exclusively to patients whose disease responded to chemotherapy and who later underwent successful CCS (two variables independently linked to improved outcomes), we sought to remove these confounding effects using a planned subset analysis that compared trimodality therapy (chemotherapy/CCS/WART) with bimodality therapy (chemotherapy/CCS). Unlike prior studies that assessed the effects of WART from the date of diagnosis, our analysis used CCS as the study start date, which enabled an unbiased measure of WART's impact in the most common clinical scenario (chemosensitive patients who underwent CCS).

Remarkably, in this more uniform patient population, WART did not improve OS. Reconciling the potential benefit of focal radiation but not WART, one explanation is that radiation was usually provided in an attempt to control solitary extraperitoneal disease in patients that harbored less extensive tumor spread as an adjunct to CCS. Thus, in many instances, radiation and CCS were carefully coordinated in an attempt to control all known detectable sites of disease. Compared with historical controls, WART appears to delay extrahepatic intraabdominal tumor recurrence and may still provide value by controlling symptoms and preserving quality of life (39). However, those end points must be balanced by the potential side effects, cost, and time required for radiation administration, and should be preferably assessed in any future trial implementing WART for DSRCT.

Among our key findings, multimodality care by experts well-versed in DSRCT treatment can achieve 5-year OS rates of 25.1%, which exceeds the 15% survival rate cited historically (7) and is considerably better than the pre-2003 OS survival rate (4.6%). Nevertheless, few patients are cured, and most have rapid tumor recurrence unless they perpetually receive systemic chemotherapy. Importantly, our data indicate that CCS and chemotherapy are essential modalities of proven clinical benefit, whereas HIPEC and WART warrant further study in randomized prospective clinical trials.

Persistent mesenteric implants and common extraabdominal metastases continue to pose formidable clinical challenges for patients diagnosed with DSRCT and durable disease-free intervals are brief when chemotherapy is withheld for any significant period. In this sense, the treatment of DSRCT more closely resembles that of metastatic ES, in which traditional cytotoxic chemotherapies and nascent biologically targeted therapies are used to halt or reverse tumor growth in an effort to alleviate cancer-related symptoms and extend patient survival. Although chemotherapy and CCS were associated with improved OS, our research calls into question the role of HIPEC and WART. Prospective, randomized, multicenter cooperative trials will be required to evaluate those local-control modalities and to advance novel biologically targeted therapies.

No potential conflicts of interest were disclosed.

Conception and design: V. Subbiah, S.-E. Lamhamedi-Cherradi, B. Cuglievan, S. Patel, N.C. Daw, A. Hayes-Jordan, J.A. Ludwig

Development of methodology: V. Subbiah, S.-E. Lamhamedi-Cherradi, V. Ramamoorthy, P.M. Anderson, M.F. McAleer, S. Patel, A. Hayes-Jordan, J.A. Ludwig

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V. Subbiah, S.-E. Lamhamedi-Cherradi, B. Cuglievan, B.A. Menegaz, P. Camacho, W. Huh, P.M. Anderson, D.C. Lev, M.F. McAleer, R.S. Benjamin, S. Patel, C.E. Herzog, N.C. Daw, A.J. Lazar, A. Hayes-Jordan, J.A. Ludwig

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): V. Subbiah, S.-E. Lamhamedi-Cherradi, B. Cuglievan, B.A. Menegaz, W. Huh, W. Qiao, S. Patel, N.C. Daw, B.W. Feig, A.J. Lazar, A. Hayes-Jordan, J.A. Ludwig

Writing, review, and/or revision of the manuscript: V. Subbiah, S.-E. Lamhamedi-Cherradi, B. Cuglievan, B.A. Menegaz, W. Huh, R.E. Pollock, M.F. McAleer, R.S. Benjamin, S. Patel, C.E. Herzog, N.C. Daw, B.W. Feig, A.J. Lazar, A. Hayes-Jordan,

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V. Subbiah, S.-E. Lamhamedi-Cherradi, B. Cuglievan, B.A. Menegaz, P. Camacho, D.C. Lev, A.J. Lazar, J.A. Ludwig

Study supervision: V. Subbiah, J.A. Ludwig

The University of Texas MD Anderson Cancer Center is supported by the NIH through Cancer Center Support Grant CA016672. J.A. Ludwig is supported by R01-CA180279-01A1.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.