Desmoplastic small round cell tumors: Multimodality treatment and new risk factors

Abstract Background To evaluate optimal therapy and potential risk factors. Methods Data of DSRCT patients <40 years treated in prospective CWS trials 1997‐2015 were analyzed. Results Median age of 60 patients was 14.5 years. Male:female ratio was 4:1. Tumors were abdominal/retroperitoneal in 56/60 (93%). 6/60 (10%) presented with a localized mass, 16/60 (27%) regionally disseminated nodes, and 38/60 (63%) with extraperitoneal metastases. At diagnosis, 23/60 (38%) patients had effusions, 4/60 (7%) a thrombosis, and 37/54 (69%) elevated CRP. 40/60 (67%) patients underwent tumor resection, 21/60 (35%) macroscopically complete. 37/60 (62%) received chemotherapy according to CEVAIE (ifosfamide, vincristine, actinomycin D, carboplatin, epirubicin, etoposide), 15/60 (25%) VAIA (ifosfamide, vincristine, adriamycin, actinomycin D) and, 5/60 (8%) P6 (cyclophosphamide, doxorubicin, vincristine, ifosfamide, etoposide). Nine received high‐dose chemotherapy, 6 received regional hyperthermia, and 20 received radiotherapy. Among 25 patients achieving complete remission, 18 (72%) received metronomic therapies. Three‐year event‐free (EFS) and overall survival (OS) were 11% (±8 confidence interval [CI] 95%) and 30% (±12 CI 95%), respectively, for all patients and 26.7% (±18.0 CI 95%) and 56.9% (±20.4 CI 95%) for 25 patients achieving remission. Extra‐abdominal site, localized disease, no effusion or ascites only, absence of thrombosis, normal CRP, complete tumor resection, and chemotherapy with VAIA correlated with EFS in univariate analysis. In multivariate analysis, significant factors were no thrombosis and chemotherapy with VAIA. In patients achieving complete remission, metronomic therapy with cyclophosphamide/vinblastine correlated with prolonged time to relapse. Conclusion Pleural effusions, venous thrombosis, and CRP elevation were identified as potential risk factors. The VAIA scheme showed best outcome. Maintenance therapy should be investigated further.


| Staging and tumor distribution
Disease stage was classified according to IRS and TNM. 7,8 For purpose of this analysis, localized disease was defined as one single tumor mass; regionally disseminated disease as multiple lesions in one affected cavity, detected by radiological assessment or inspection during surgical intervention. Extraperitoneal metastases were defined as lesions outside the affected cavity or in solid organs.

| Thrombosis
Venous thrombosis was diagnosed radiographically at diagnosis before implantation of a central venous line. No systematic screening for thromboses was performed.

| Pretreatment blood parameters
Coagulation parameters, blood counts, and C-reactive protein values (CRP) were evaluated. D-dimers were graded as normal (<500 µg/L), moderately increased (500-2000 µg/L), and markedly increased (>2000 µg/L). Scoring for disseminated intravascular coagulation (DIC) was done according to the scoring system of the International Society on Thrombosis and Hemostasis (ISTH). 9 The inflammatory scores neutrophil-to-lymphocyte ratio and platelet-to-lymphocyte ratio were calculated by dividing the absolute neutrophil or platelet count, respectively, by the absolute lymphocyte count. The median was chosen as threshold. CRP was considered elevated when >5 mg/L.

| First-line chemotherapy
The induction therapy used was trial-dependent (Table S1). Early diagnosed patients had received the P6-scheme according to Kushner. 10 Regional hyperthermia in combination with platinum-based chemotherapy 11 and high-dose chemotherapy with autologous stem cell rescue were administered on an individual basis.
Hyperthermic intraperitoneal chemotherapy (HIPEC) procedures with cisplatin and doxorubicin in combination with tumor resections were performed, a decision made by the treating oncologist.

| Maintenance chemotherapy
Remission was defined as no macroscopic residuals at therapy end (achieved with surgery and/or chemotherapy and/or radiotherapy). Maintenance chemotherapy with oral trofosfamide, idarubicin, etoposide (O-TIE), 12 or with cyclophosphamide per os, vinblastine intravenously (Cyc/Vbl) 6,13,14 was administered (Table S1) trial-dependent or individually. Some patients received individual therapy with irinotecan/ temozolomide or irinotecan/trabectidin. In addition to survival from diagnosis, time to relapse was calculated from end of intensive chemotherapy to eliminate the influence of different lengths of intensive therapies.
In regionally disseminated nodular disease with variable response, the lesion showing the least volume reduction was evaluated.

| Local treatment
CWS protocols recommended local treatment of DSRCT analogous to non-rhabdomyosarcoma (NRSTS). Overall treatment strategies did not change substantially over the years. The mainstay of treatment for NRSTS was primary surgery with preferably complete or wide-ranging resection after biopsy. In metastatic disease, primary excision or biopsy of the primary tumor and/or metastases was performed and systemic therapy was administered to reduce tumor volume and improve respectability. This was followed by secondary surgery if feasible. Best surgery was defined as the best surgical result obtained at the end of treatment irrespective of procedure numbers. The surgical result was categorized as biopsy only, presence of macroscopic [R2], or microscopic [R1] residual tumor or as resection with free margins [R0].
Radiotherapy was to be administered with a total dose of 32-54.4 Gy (dose reduced hyperfractionated/accelerated radiotherapy) depending on primary resection status. Postoperative radiotherapy of the primary tumor (and metastases) was recommended. External beam irradiation with 44.8 Gy was recommended if a microscopically complete resection (R0) was not performed. In CWS-2002P, the recommendation was extended to include patients in whom an R0-resection had been performed if tumor size was >5 cm or age >10 years. In SoTiSaR, radiotherapy with 50.4 Gy (conventional fractionated irradiation) was added as an alternative. However, individual physician decisions did not always follow these recommendations. Multivariate analysis was conducted using Cox's proportional hazards regression method to establish independent prognostic significance. A stepwise variable selection procedure (combination of forward and backward selection techniques) was applied to covariates with P-value of at least 0.1 in EFS at univariate analysis. Hazard ratios (HRs) with 95% confidence intervals, calculated according to the Wald method, are reported for significant variables. Statistics were calculated using IBM SPSS ® 24 for all other analyses.
Six patients had a localized tumor (all EWSR1-WT1-positive, 3 extra-abdominal, 3 abdominal). An additional patient with a single primary in the abdomen also had extraperitoneal metastases (liver, lung, mediastinal lymph nodes).
Four patients presented with thrombosis before implantation of a central catheter. Two had a thrombosis in the vena cava, one in the iliac with extension into femoral and popliteal veins while the other had a brachiocephalic and jugular thrombosis.
Response to chemotherapy ranged from partial response to progression ( Table 2). In six patients with regionally disseminated disease, a mixed response with reduction of some and stable or increased volume in other lesions was documented.
Nine patients received high-dose chemotherapy with autologous stem cell transplantation, and six other patients were treated with regional hyperthermia in combination with platinum-based chemotherapy 11 in addition to standard chemotherapy. 21/60 (35%) patients underwent complete resection of primary tumor manifestations (R0 n = 5, R1 n = 16). Two patients with R1 resection suffered disease progression shortly after resection and therefore were not considered as having achieved remission. Five patients were treated with HIPEC, three following R1-and two after R2-resection.

| Outcome
Overall, 51/60 patients died, 49 of disease and two of therapy-related causes (organ failure after chemotherapy, postoperative complications with peritonitis after anus praeter relocation). With a median follow-up of 3.2 years (1.2-10.6), 9/60 patients were alive, six in 1st, one in 2nd remission, one lost to follow-up without ever having achieved remission, and one in first relapse with active disease (Table S4).
Among the four patients initially presenting with thrombosis, two showed progression while still on chemotherapy, while the remaining suffered early recurrences. None survived longer than 1.5 years.

| Outcome according to surgery
In all those 19 patients who only underwent biopsy as best surgical procedure, an event (disease progression or relapse) was documented in a median of 0.7 years (range 0.2-3.3). Median overall survival was 1.6 years (0.6-5.3). None survived. In those other 19 patients with R2-resection as best surgical result at any time, all had an event in a median of 1.0 years (0.2-5.2). Median overall survival was 1.7 years (0.3-6.4). Two were alive at the cutoff date, thereof one lost in progression and the other in first relapse with active disease. Of those 16 patients with R1 resection, 13 had an event in a median of 1.2 years (0.1-2.4). Median overall survival was 2.6 years (1.1-6.8). Four patients were in first remission at the cutoff date with a follow-up of 5.5, 3.2, 2.5, and 2.5 years, respectively. Five were alive. Of those five patients with R0-resection, 3 had an event in a median of 0.6 years (0.3-3.0). Median overall survival was 4.2 years (1.0-10.6). Two were alive at the cutoff date with a follow-up of 10.6 and 6.1 years, respectively.

| Outcome according to chemotherapy
The five patients treated according to the P6-regimen had a median EFS of 12.9 months (7.6-16.5), 15 treated with VAIA 29.4 months (1-127.7), and 37 who received CEVAIE 12.0 months (22.1-306.7). The remaining patient survived 9 months with palliative care. The VAIA scheme correlated with increased chance of R0 or R0/R1 resection (Table S3). Potential correlations between maintenance therapy and outcome were analyzed in the subgroup of 25 patients achieving remission (Table 2, Figure 3). In six patients without maintenance therapy, the median time to relapse calculated from the end of intensive chemotherapy was 8

| Pattern of relapse
Among those 19 patients with relapse, five suffered recurrence with merely intraperitoneal lesions. Nine other patients suffered metastatic recurrence with extraperitoneal lesions at different sites (n = 1 lung and bone lesions, n = 1 lymph node supraclavicular, n = 1 bone lesions, n = 1 mediastinal mass and lung lesions, n = 1 lung and liver lesions, n = 1 lung lesions, n = 1 one spleen lesion, n = 1 liver and brain lesions, and n = 1 pleural sarcomatosis). Five patients suffered relapse with combined intra-and extraperitoneal lesions. Extraperitoneal lesions were documented at liver and skin (n = 1), paratracheal and bone (n = 1), thoracal and lung (n = 1), thoracic lesions subpleural (n = 1), the site of the affected distant lymph nodes was not documented (n = 1).

| Uni-and multivariate analyses
In univariate analysis, extra-abdominal site, localized disease, no effusion or ascites only, absence of thrombosis, and normal CRP were associated with a significantly increased EFS (Table 1). Univariate analysis of metastatic sites did not show any differences in EFS whereas the existence of liver metastases correlated with reduced OS (Table S2).
Among first-line treatment, chemotherapy with VAIA and complete resection of all primary tumor lesions at any time correlated with prolonged EFS ( Table 2).
The distribution of prognostic factors among the chemotherapeutic regimens is shown in Table S3.
A multivariate analysis was conducted to establish independent prognostic significance of the evaluated factors ( Table 3). Absence of thrombosis and chemotherapy according to the VAIA regimen remained significant.

| DISCUSSION
This study of 60 DSRCT patients confirms the poor prognosis. A small subgroup with reasonable outlook for long-term, disease-free survival could be identified, three individuals with localized and extra-abdominal disease, without pleural effusions, or thrombosis and normal CRP who were subsequently treated by complete tumor resection in conjunction F I G U R E 3 The effect of additional maintenance therapy after the end of intensive chemotherapy evaluated in 25 patients who achieved a first complete remission with VAIA chemotherapy. All three patients were diseasefree at last follow-up (6.1, 5.5, 2.5 years).
An inherent weakness of this study is the small number of patients due to the fact that DSRCT has only been recently described and is a very rare tumor. Nevertheless, to the best of our knowledge, it is the largest series of DSRCT patients enrolled in prospective trials to date. However, our analysis is retrospective and the base-line blood parameters are nonspecific. Further evaluation should be prospective and standardized.
Characteristics were similar to previous reports. 20,21 Patients were principally adolescent, male and had large and widespread abdominal and/or retroperitoneal tumors. The majority had extraperitoneal metastases. No typical metastatic pattern was identifiable. No specific risk for disease progression or relapse of any specific metastatic site could be revealed. Nevertheless, though the existence of liver metastases did not correlate with reduced EFS in our series, it correlated with a reduced overall survival probability. Liver metastases have been observed as a potential prognostic factor before. 22,23 No specific pattern of disease recurrence could be determined; therefore, we were not able to identify specific treatment weaknesses.
In contrast to the literature, existence of extraperitoneal metastases did not correlate with survival. 1,24,25 This supposed difference might have been caused by the fact that others had not differentiated between localized and regionally disseminated disease. However, others had estimated tumor burden based on peritoneal cancer index but survival did not differ. 24,25 In contrast to all other solid tumors, size and evidence of distant metastases did not correlate with survival. One cause might be the unique nature of dissemination. It seems that from the point, the cells lose the ability to organize within a single node and start disseminating into multiple nodules, survival decreases significantly. It remains to be elucidated whether this is due to the fact that local therapy is restricted or whether regionally disseminated disease indicates considerably more aggressive biology.
Though the presence of effusions has been described in DSRCT, 26 prognostic impact is unclear. In our series, pleural effusions correlated with inferior survival.
Four patients aged 8, 10, 11, and 17 years presented with venous thrombosis-detected coincidentally in the initial staging investigations. The association between thrombosis and cancer is well known as the Trousseau's syndrome. 27 Occurrence varies by age and cancer type. [28][29][30][31][32] Pathogenesis is multifactorial. 33 A positive association can be observed with 1-year mortality of the cancer type as a measure of biological aggressiveness and associated thrombogenic potential. 29,34 In our series, coagulation activation with thrombus formation seems to indicate more aggressive disease. Unfortunately, detailed analysis of the coagulation cascade was not possible. In only 15/60 (25%) patients, the necessary coagulation values including D-dimers were determined. Elevation was evident in 13 (86%). It is debatable, whether D-dimers alone or the ISTH scoring system is sufficient to capture or even quantify the pivotal phenomenon. 35,36 One contributing factor to coagulation activation is inflammation, which plays an important role in tumorigenesis. 37 Inflammatory scores are predictive in numerous cancers. 19,38 In our series, merely CRP correlated with survival. As there are no acknowledged standard values for the applied ratios, 39 the median was chosen. 40 However, the predictive significance of CRP suggests need to determine clinical utility. It seems worth mentioning that among those 17 patients with normal CRP, five had localized, four regionally distributed, and eight widespread disease with extraperitoneal metastases.
Of interest are the underlying mechanisms. Malignancy is characterized by destructive tumor cell invasion of vascularized organ tissue and formation of leaky new vessels, so-called tumor angiogenesis. The procoagulant phenotype is likely to have profound effect on the hemostatic system as result of plasma leakage into the interstitial space, secretion of inflammatory cytokines, and direct invasion of tumor cells into blood or lymphatic vessels. 27 Coagulation activation, inflammation, and tumor biology form a triangular network with reciprocal interactions. 33,41 The elucidation of these mechanisms might help to understand DSRCT biology.
The most effective chemotherapeutic regime still is debated. Most combinations are based on alkylating agents, similar to those in other small round cell tumors or Ewing sarcomas, which also carry EWSR1 fusions. 25,42 Farhat reported five patients with disease stabilization lasting 4-9 months using a regimen consisting of cisplatin, etoposide, cyclophosphamide, and an anthracycline. 43 Kushner reported 12 patients with median survival of 19 months with the P6-protocol. 10 Bertuzzi reported median survival of 14 months in 17 patients treated in two prospective studies with induction chemotherapy of ifosfamide, epirubicin, and vincristine. 44,45 Wong reported median time to progression of 14.6 months in 13 patients treated according to the VIDE-regimen (vincristine, ifosfamide, doxorubicin, etoposide). 20 Another report states survival of 1.7 years in a single case with VIDE. 46 The use of other agents has been reported, including irinotecan, temozolomide, and vinorelbine, but none of them showed superiority. 47,48 Evidence is emerging on trabectedin. [49][50][51][52][53] Activity was reported for pazopanib 54,55 and eribulin. 56 A recent report gave negative results for imatinib. 57 Limited activity of the antiangiogenics sunitinib, sorafenib, and bevacizumab is reported. [58][59][60] In our series chemotherapy according to the VAIA scheme correlated with longer EFS than did the other investigated regimes. The median EFS of 15 patients treated with VAIA was 29.4 months. To the best of our knowledge, this is the first assessment of the VAIA scheme in DSRCT. Though the distribution of prognostic factors differed among the three analyzed chemotherapeutic regimens, the independent effect of the VAIA scheme was proven in the Cox regression analysis.
However, the fact that response to chemotherapy did not correlate with outcome is striking. 2/6 (33%) survivors had progressive disease. They were then treated with aggressive surgery.
6/60 (10%) patients showed mixed response of different tumor nodes suggesting heterogenous biology, which should be considered when adopting targeted therapeutic approaches.
In accordance with the recent literature, high-dose chemotherapy did not correlate with survival. 61 The value of hyperthermia remains to be specified. However, in our series, patients did not benefit.
While there is considerable investment in the development of innovative therapies, metronomic approaches have been underinvestigated. In our series, metronomic therapy with cyclophosphamide/vinblastine correlated with prolonged time to relapse.
The crucial and decisive role of surgery can be confirmed. 42,62 In our series, no patient who did not have a R0 resection or R1 resection survived long-term or was disease-free at the cutoff date. More precisely, of those 38 patients with only biopsy or R2-resection, only two were alive at the cutoff date, thereof one lost in progression and the other in first relapse with active disease. However, the feasibility of surgery due to extensively disseminated disease remains a problem. Interestingly, in the subgroup of 25 patients achieving remission, extent of surgery was not predictive of outcome. Two patients who only underwent biopsy achieved remission with chemotherapy alone, one with simultaneous irradiation. Unfortunately, the additional value of HIPEC 1,24,63,64 cannot be further elucidated as only 5 patients with different resection results received HIPEC.
Although the CWS trials included radiotherapy recommendations, most patients were irradiated following an individual concept due to extensive disease. Radiotherapy mainly consisted of focused irradiation to particular tumor sites or nonresectable nodes. Whole-abdominal irradiation was performed in two patients. Thus, a reliable conclusion with regard to the specific effect of radiotherapy in DSRCT remains difficult. In our series, irradiation did not correlate with survival. Other authors suggested that a multimodal concept combining chemotherapy, surgery, and irradiation enables prolonges survival. 25,63,65 In the future, the application and evaluation of standardized radiotherapeutic concepts and/or techniques are necessary to specify its potential benefit. 22,23 Nevertheless, in our series no specific pattern of disease recurrence was evident, so we were not able to identify specific treatment weaknesses.
In summary, the VAIA scheme could be specified as best chemotherapy in a multivariable model. The effect of maintenance therapy warrants further investigation. Pleural effusions, venous thrombosis, and CRP elevation were identified as novel potential risk factors for adverse outcomes. The elucidation of the underlying coagulation and inflammatory mechanisms may lead to a better understanding of disease biology, especially when investigating the observed coagulatory and inflammatory phenomena in conjunction with genetic analysis.