Five Decades of Sarcoma Care at Memorial Sloan Kettering Cancer Center

The 1970s

The 1970s began the recognition that conservative operations with or without radiation therapy were equivalent to amputation for extremity lesions. It became clear that amputation was no longer necessary for improved survival in patients with extremity soft tissue sarcoma.[1]

Early studies during this period strongly suggested that radiation was a significant factor to decrease local recurrence in the extremity.[1]

Preoperative evaluation was changed with the development of the first body computerized tomography (CT) scan in 1975 introduced by Hounsfield. This moiety was initially of most value for the recognition of pulmonary metastasis in patients previously screened with chest x-ray or early pulmonary tomograms.[2] However, CT was quickly adapted for evaluation of intra-abdominal and retroperitoneal lesions. This allowed selective surgical approaches most likely to result in complete gross tumor resection. Hounsfield and Commack received the Nobel Prize, for this remarkable intervention in 1979.[3, 4]

Early desktop computer medical record development began, initially utilized to prevent medication errors from occurring, especially with risk of overdose of narcotics or chemotherapy.

Following the pioneering efforts of Joseph Burchenal at Memorial Sloan Kettering Cancer Center (MSKCC) and others of the use of chemotherapy for childhood leukemia and lymphoma,[5] reports of effective systemic treatment of pediatric soft tissue sarcoma were described, with optimism that such results could be transferred to adults.

The 1980s

At MSKCC the 1980s was the beginning of prospective institutional databases. The sarcoma database at MSKCC began in 1982, and quickly provided natural history and demographics capable of predicting local and metastatic spread. Such databases allowed recognition of the different variables that were predictive of different outcomes. An important finding that each and every outcome, local recurrence, metastatic spread, and disease specific survival, had variable prognostic predictors at the time of initial operation.[6]

The increasing use of adjuvant and therapeutic radiation and chemotherapy required serious studies of the impact of these modalities on surgical wound healing. Not only was the dose of such entities important but the timing of such treatment both before or after a surgical procedure was highly predictive of wound complications.[7]

Adriamycin became the major adjuvant and therapeutic agent to be studied. It was soon clear that cumulative dose frequency and duration of administration were major factors in cardiac toxicity. Studies of conversion of bolus to continuous infusion allowed this valuable drug to be used more judiciously.[8]

As a consequence of meticulous prospective studies, grade and histological subtype became major prognostic factors for survival and local recurrence. These studies allowed definition of variable factors based on site to predict outcome. There was early recognition of the importance of interaction of individual prognostic factors such as size and grade.[6]

As prospective studies were converted to randomized controlled trials the definition of wound complications in response to adjuvant chemo and radiation therapy were adjusted based on animal and human wound healing studies.[9]

The introduction of MRI (magnetic resonance imaging) in the definition of extremity sarcoma allowed comparison to CT as to what was the appropriate and most valuable modality for preoperative evaluation.[10] in 1982, the initial prospective randomized controlled trial (RCT) of limb sparing plus radiation versus amputation was reported from the National Cancer Institute.[11]

Following the identification of the RB gene product it’s importance was recognized as a highly predictive variable in outcome.[12]

Subsequent studies however suggested that the importance of this variable was highly correlated with other clinical predictive values such as site, size and grade.[13]

Early studies of adjuvant chemotherapy in adults with soft tissue sarcoma were disappointing compared to studies in children and adolescents.[14]

The 1990s

In the 1990s pathological definition of unique mutational signatures of sarcoma began the era of molecular diagnosis.[15]

Studies to limit the aggressive approach to the desmoid tumor followed recognition of multifocal desmoids in adolescents, especially females[16] many of which could be managed conservatively with a focus on preservation of function rather than complete tumor resection.

Studies documented the low prevalence of lymph node metastasis in soft tissue sarcoma except in epithelial subtypes.[17]

A randomized controlled trial at MSKCC confirmed that radiation therapy improved local control over surgery alone using the brachytherapy technique but with no survival benefit.[6] This observation appeared confined to the high-grade tumor. Subsequent studies of external beam radiation suggested that low-grade tumor did benefit in limiting local control if given by the external beam route. There was significant long term toxicity.[18]

Subsequent studies of intensity modulated radiation therapy, would limit the broad applicability of brachytherapy especially due to the need for prolonged post-surgery hospitalization. Brachytherapy remained a useful adjunct for re-irradiation for recurrence, to avoid amputation.[19]

During this period there was definition of the lack of benefit of extended normal organ resection in the management of retroperitoneal sarcoma.[20, 21] This gave rise to the concept of renal preservation whenever possible as frank parenchymal invasion was uncommon and the limited adhesion could be dealt with by capsular resection.

Open surgical biopsy was recognized as not mandatory. Core needle biopsy was adequate for both diagnosis of malignancy, differentiation of sarcoma from carcinoma and in the majority of cases adequate for grade and histology.[22]

The prognosis and value of molecular diagnosis came as the fusion genes would redefine synovial sarcoma, historically seen as a disease of young adolescents occurring in the extremities and could now be made in essentially any part of the body with the definition of the SYT-SSX gene fusion.[23]

By the late 1990s the power of prospective databases was reinforced and the importance of prolonged, greater than 5 year follow-up clearly demonstrated by the results of the management of 500 patients with retroperitoneal soft tissue sarcoma.[24]

The increasing definition of post radiation breast sarcoma, especially for DCIS as a non-lethal primary entity[25] was a reminder of Stewart and Treves who first reported post radiation, and post surgical lymphedema as precursors to sarcoma development,[26] in the first volume of the journal Cancer. This seminal journal development recognized the evolution of oncology as a vigorous and defined specialty.[26]

The 2000s

By the early 2000s, the prospective clinical database had matured to enable defining long term outcomes in retroperitoneal liposarcoma showing the importance of histologic subtype: well-differentiated, dedifferentiated, myxoid and round cell liposarcoma for improving prognostication and patient stratification for clinical trials.[27] In the extremity location, the natural history of well-differentiated liposarcoma (WDLS) / atypical lipomatous tumors was defined and the sclerosing WDLS subtype was found to associate with a higher rate of local recurrence.[28] We then developed a liposarcoma specific nomogram that included histologic subtype and other clinical variables which improved outcome prediction across all tumor locations.[29] These Mature follow up data on MSK patients treated for their primary extremity sarcoma who then developed a local recurrence allowed for the analysis of clinical prognostic factors such as local recurrence tumor size, grade and the local-recurrence free interval that independently associated with disease-specific survival following complete resection of a locally recurrent extremity sarcoma. Using this recurrence data, we were able to define the clinical observation that only large high-grade recurrence impacted significantly on disease specific survival. A small less than 5 cm recurrence of a low-grade extremity sarcoma occurring after 16 months translated into a 4 year disease specific survival of 81% opposed to the appearance of a large (>5cm)high-grade lesion recurring in less than 16 months which translated into a 4 year survival of 18%.[30]

We also found that local recurrence tumor size and the interval to local recurrence are highly prognostic For outcome in retroperitoneal liposarcoma size. The average local recurrence growth rate (defined as the local recurrence tumor burden / the time interval to local recurrence) was strongly associated with disease-specific survival and local control for patients with completely resected locally recurrent retroperitoneal liposarcoma. Despite aggressive operative management patients with an average local recurrence growth rate greater than 1 cm/mo were associated with poor outcomes and should be considered for enrollment in clinical trials employing novel agents.[31]

The 2000s began the two decades of development of the underlying biology of STS with improved classification of sarcomas based on genetic and cytogenetic alterations, building on the long-standing clinical documentation began two decades earlier. Genetic analysis allowed for dividing sarcomas into two broad groups. The first is sarcomas with simple genetic alterations (e.g. translocations or specific activating mutations). The second is sarcomas with highly complex genomic aberrations, commonly including alterations in the cell cycle genes TP53, MDM2, RB1, CDK4, and CDKN2A.

In the early 2000s the gastrointestinal stromal tumors (GIST) were defined as a specific entity by the discovery of activating mutations in KIT and PDGFRA in GIST and the previously described gastrointestinal autonomic nerve tumors (GANT or GANN) or presumptive retroperitoneal leiomyosarcoma replaced. These could now be defined, and natural history characterized.[32, 33] This paralleled characterization and emphasis of the importance of biological subtype within soft tissue sarcoma and how a type-specific molecular alteration, if it is critical for cell proliferation and survival, can form the basis for effective targeted therapy. This idea is best illustrated by the development of imatinib, an inhibitor of ABL, KIT, and PDGFRA tyrosine kinases for GIST. Dr Singer was a critical member of the team responsible for the first clinical deployment of imatinib for patients with advanced GIST[34, 35] and was the first to demonstrate the importance of KIT mutation type and mitotic activity as important prognostic factors in patients with primary GIST.[36] Upon the discovery of activating mutations in KIT and PDGFRA in GIST, imatinib proved to be an effective, low toxicity therapy. Prior to imatinib, patients with advanced unresectable or metastatic GIST had a rapidly fatal clinical course with no evidence of benefit from standard chemotherapy or radiotherapy. In clinical studies, imatinib provides clinical benefit to 75% to 90% of patients with advanced GIST,[37–39] with a 2-year survival rate of 70–80%.[40] We subsequently showed that selected patients with metastatic GIST who have responsive disease or focal resistance to tyrosine kinase inhibitor therapy may benefit from elective surgical resection.

In 2006, it was shown that hat flavopiridol suppresses KIT mRNA expression through positive transcriptional elongation factor inhibition and decreases KIT promoter activity, resulting in a global decrease in the level of functionally mature KIT at the cell surface. This team also found that targeting KIT expression with flavopiridol represents a novel means to disrupt GIST cell dependence on KIT signaling and collectively renders these cells sensitive to apoptosis.[41]

At MSKCC the Sarcoma Biology Laboratory was outfitted with a state-of-the-art 600MHZ NMR magnet with magic angle spinning capability to afford high resolution metabolite analysis of sarcoma tissue samples from fresh resection specimens to enable metabolite predictors of sarcoma type/subtype, differentiation state and histologic grade.[42–45] The Sarcoma Biology Lab subsequently focused on the integration of large-scale analyses of gene expression, copy number alterations, and mutations in soft tissue sarcoma to characterize the landscape of genetic alterations in specific sarcoma types/subtypes and to identify subtype-specific therapeutic targets, and biomarkers outcome. [46, 47]

This laboratory effort made particular strides in defining 12q-amplified liposarcoma and identified CDK4 and MDM2 as potential therapeutic targets,[47, 48] leading to clinical trials of corresponding inhibitors. This laboratory effort established a large number of sarcoma cell lines across multiple sarcoma types [well-differentiated liposarcoma (WDLS), dedifferentiated liposarcoma (DDLS), undifferentiated pleomorphic sarcoma (UPS), myxofibrosarcoma (MFS), myxoid/round cell liposarcoma] that were genetically characterized for mutations and copy number alterations which were then utilized for functional analysis of biological pathways to identify promising therapeutic targets and test new targeted agents in-vitro and in-vivo.

Given the toxicity of postoperative radiation therapy a randomized controlled trial of pre versus postoperative radiation therapy was reported.[49, 50] This suggested there was an increase in immediate perioperative wound complications in the preoperative radiotherapy (XRT) group, but greater long-term complications with postoperative XRT. This resulted in increasingly selective use of radiation, defining low risk subtypes by histology, size and grade and development of Intensity-modulated radiation therapy (IMRT) for soft tissue sarcoma (STS).[51]

Failure to accrue to a prospective randomized trial for preoperative radiation therapy in RPS trial led by the American College of Surgeons Oncology Group was taken up subsequently by the European study group.

Newer histopathological biological markers such as Ki-67 were recognized as a predictive marker for disease specific survival.

There was continued evaluation of metastatic resection of lung and liver metastases[52–54] showing prolonged survival in highly selected patients.

A new approach allowed major symptomatic relief using Intralesional resection of metastasis to bone with spinal stabilization, providing significant palliation.[55]

The presence of institutional based databases allowed early development of sarcoma nomograms for local recurrence, disease specific survival, post metastasis survival[56, 57]all of which were subsequently validated.[58]

Definition of the relevance of positive microscopic margin following resection of soft tissue sarcoma[59] allowed careful reconsideration of the objective of widely negative margins being sacrificed for close or positive margins, when extended resection would create high morbidity. This approach allowed recognition that many patients with microscopic margin positivity did not recur.

A landmark randomized controlled trial of tumor specific adjuvant therapy for gastrointestinal stromal tumors was conducted and reported.[60] Continued study of trial participants and others allowed recognition and definition of Gleevec resistance based on analysis of cKit mutation. [60]

Multiple trials investigating the role of adjuvant chemotherapy for high risk tumors were reported in meta-analysis,[61] suggested a limited benefit of Adriamycin based therapy, with significant potential toxicity, as to not justify adjuvant chemotherapy as standard of care by most but not all clinicians.

Extensive studies resulted in definition of prognosis dependent on time of local recurrence[30, 62] and emphasized the problem of heterogeneity in stage.[63, 64]

The extensive accumulation of prospective molecular characterization of tumor subtype allowed subtype specific nomograms to be described for liposarcoma[29] and for GIST.[65]

The 2010’s

In this decade there was further expansion of integrative genome-scale analysis of mutations, DNA copy number variations, changes in DNA methylation, and changes in expression of genes and microRNAs in multiple types of soft tissue sarcoma. The results, provided a broad view of the genomic landscape of sarcomas, revealed novel potential drivers and potential therapeutic targets. These discoveries provided the basis for clinical trials of CDK4 and MDM2 inhibitors for patients with liposarcoma, leading to CDK4 inhibitors becoming the standard of care for patients with Rb-positive dedifferentiated liposarcoma.[66]

Biomarkers of CDK4 inhibitor-induced senescence elucidated the mechanism regulating quiescence versus senescence in well-differentiated and dedifferentiated liposarcoma cells.[48, 67, 68] The data generated in these projects have served as the starting point for numerous subsequent investigations.

CEBPα was identified as a tumor suppressor in 12q-amplified dedifferentiated liposarcoma (DDLS)[69] and showed that pharmacologic inhibition of DNA methylation may be a promising therapeutic approach for DDLS harboring CEBPα promoter methylation[70] or miR-193b promotor methylation.[71] miR-193b serves as a tumor suppressor in dedifferentiated liposarcoma by directly targeting FAK, CRKL, and MSRA to regulate focal adhesion signaling and ROS signaling. DNMT1 upregulation in WDLS and DDLS promotes miR-193b promoter methylation, leading to miR-193b under expression in liposarcoma.[71] The transcription factor CEBPα, which cooperates with PPARγ to induce adipogenesis, is often downregulated in dedifferentiated liposarcoma (DDLS) tissues and cell lines, and this downregulation is a marker of poor prognosis. Exogenously expressing CEBPα in DDLS cell lines inhibits proliferation and can restore the ability to induce early adipogenesis markers, eventually inducing G2/M arrest and apoptosis. This decrease in CEBPα expression may result from epigenetic defects, namely CEBPα promoter methylation and HDAC1 mutations. Consistent with this, treating DDLS cells with a demethylating agent and the histone deacetylase inhibitor SAHA dramatically increases CEBPα expression, decreases proliferation, induces apoptosis, and reduces growth of DDLS xenografts by 50%–70%. Taken together, these results suggest that CEBPα acts as a tumor suppressor in DDLS, that its loss may explain the undifferentiated state of DDLS, and that DNA methylation inhibitors may be a promising therapeutic approach for DDLS harboring CEBPα promoter methylation.[70] A screen for compounds that restore expression of C/EBPα yielded 19 compounds that induce C/EBPα expression and inhibit proliferation in two DDLS cell lines, but not in normal adipose-derived stem cells.[72] Among them was SN-38, the active metabolite of the chemotherapeutic irinotecan, which is effective in vitro at low nanomolar concentrations. In xenograft models, irinotecan was more effective than doxorubicin, the current standard of care. These findings have provided the justification for a phase 2 clinical trial of a liposomal formulation of irinotecan in patients with advanced DDLS.

The MSK sarcoma lab, led by the senior author found that integrin-α10 is an important driver of survival and metastasis in myxofibrosarcoma and undifferentiated pleomorphic sarcoma, leading to activation of RAC/PAK and AKT/mTOR pathways.[73] Integrin-α10 expression is highly associated with worse outcomes and distant metastasis in myxofibrosarcoma. Integrin-α10 promotes cell growth, migration, and survival of myxofibrosarcoma cells, and that engagement of its I-domain to its ligand ECM collagen is crucial for the signaling function. This integrin signaling pathway can be targeted downstream using inhibitors of RAC, mTOR, and PI3K, or using antibodies directed against integrin-α10 (under development in Dr Singer’s group). These discoveries have provided the basis for a clinical phase 2 trial of a selective mTOR inhibitor, MLN0128, for patients with myxofibrosarcoma or undifferentiated pleomorphic sarcoma.

Based on evidence that MFS and UPS in about 70% of patients harbor copy number alterations or mutations in the tumor suppressor genes RB1 and TP53, which engender dependence on the oncogenic protein Skp2 and examined Skp2’s function and potential as a therapeutic target in MFS/UPS.[74] Skp2 drives proliferation of patient-derived MFS/UPS cell lines deficient in both RB and p53 by degrading p21 and p27. Inhibition of Skp2 using the neddylation-activating enzyme (NAE) inhibitor pevonedistat decreased growth of Rb/p53-negative patient-derived cell lines and mouse xenografts. This pre-clinical data has motivated us to pursue a phase1/2 trial of pevonedistat in patients with RB- and p53-deficient metastatic MFS and UPS.

We have also shown through molecular analysis that myxofibrosarcoma and undifferentiated pleomorphic sarcoma form a single genetic entity characterized by complex copy number alterations with few recurrent mutations.[48, 68] Furthermore, the percent myxoid component in these tumors is an independent predictor of disease-specific survival and that a 5% myxoid component cutoff is an improved criterion for distinguishing myxofibrosarcoma and undifferentiated pleomorphic sarcoma. [75]

Expression profiling of STS, became standard of care.[76]

There was a major increase in watchful waiting for desmoid tumors with a parallel decrease in the morbidity of historically aggressive surgical and multimodality approaches.[77]

Advancing on our experience with classical nomograms, we began to use artificial intelligence (AI) for modeling decision trees.[78]

As part of our historical prospective databases, we were able to report an analysis of 10,000 patients with STS treated by the surgical service at our institution.[79] We found that for retroperitoneal sarcoma Histologic type/subtype is the most important independent predictor of disease-specific death, local recurrence, and distant recurrence. Histology predicts the pattern and incidence of local and distant recurrence, aids in the design of future clinical trials and enables more accurate counseling of patients.[80]

Continuing prior studies of the impact of prior radiation, we further defined the histopathology and outcome of radiation associated STS.[81] On going analytics allowed the application of Bayesian analytics to predict outcome,[82] and the refinement of nomograms for outcome of extremity STS without XRT,[83] the definition of a nomogram for outcome of desmoid fibromatosis,[77] and the definition of the role of Wnt pathways in desmoid tumors.[84]

As all of our analytics were clinically correlated, we were able to define separate issues such as the role of copy number losses in the poor outcome in dedifferentiated liposarcoma.[85]

Studies on the role of surgical resection for metastatic disease defined the role of pulmonary resection for metastatic STS[86]and the role of surgical resection for metastatic GIST further defined.[87, 88]

In summary, MSK has been a leader in the definition and management of STS. While early studies focused on natural history, the prospective accumulation of data allowed extensive prediction of outcome variable for recurrence and survival. The last two decades have seen an extensive molecular and biological characterization of individual subtypes and the integration of such data to predict outcome and allow innovation of targeted therapies. The integrated approach of using tumor biology to define outcome allows great optimism for increasingly effective diagnosis and treatment, with minimized patient morbidity.