Soft Tissue Sarcoma Clinical Trials Landscape

Curated by Dual Brain Lab (csilab.net) · Twelfth tumor type on the map Data cutoff: 2026-04 · Guideline anchor: NCCN Soft Tissue Sarcoma v3.2026 (March 12, 2026)

This post indexes all 42 landmark trials under /trials/sarcoma/ into a full-view timeline, organized by NCCN STS v3.2026’s 6 subtypes bundled into 3 clinical-biology buckets. Soft tissue sarcoma is the most heterogeneous family among solid tumors — >50 WHO histological subtypes, with almost no shared biology or drug sensitivity across them (LMS responds to doxorubicin+trabectedin · myxoid LPS uniquely responds to trabectedin · UPS/ddLPS carries an IO signal · desmoid does not metastasize but invades locally). This heterogeneity forces STS treatment to leave behind the “broad-spectrum” chemotherapy logic and adopt histotype-tailored strategies.

Quick entries by bucket:

→ Adult Mainstream STS (extremity + retro + ALT/WDLPS + cross-subtype systemic)
→ Desmoid + Phyllodes (unique biology)
→ Pediatric / AYA RMS (North America COG vs European EpSSG)

1. Clinical Landscape

Rare and extremely heterogeneous. Soft tissue sarcoma accounts for ~1% of adult solid tumors and ~15% of childhood malignancies. 2022 US estimates: ~13,190 new STS cases and 5,130 deaths. Most common primary sites: extremity 43% · trunk 10% · visceral 19% · retroperitoneum 15% · head and neck 9%. >50 WHO histological subtypes.

NCCN v3.2026 covers 6 subtypes:

STS of extremity / body wall / head and neck
Retroperitoneal / intra-abdominal STS
Desmoid tumors (aggressive fibromatosis)
Rhabdomyosarcoma (RMS)
Atypical lipomatous tumor / well-differentiated liposarcoma (ALT/WDLPS)
Borderline / malignant phyllodes tumor of the breast

Not covered (separate NCCN books): GIST · bone sarcomas (osteosarcoma / Ewing / chondrosarcoma, see Bone sarcoma overview) · uterine sarcomas · DFSP.

Clinical implications of heterogeneity. Rarity × 50 subtypes means single-subtype phase 3 RCTs are extremely hard to conduct (enrollment often takes 5-10 years). International cooperative groups (EORTC · SARC · FSG · ISG · COG · EpSSG · TARPSWG) have been the engine of STS landmark trials for 30 years. The real treatment wisdom to learn from STS is not “which drug” but “histology dictates the path.”

2. Current Treatment Paradigms

STS paradigms must be unpacked by “biology bucket”. Each bucket has fundamentally different treatment logic.

Adult Mainstream STS

Backbone: R0 surgery + radiotherapy + histology-driven systemic therapy. The foundational path has been stable for 30 years; the past decade’s evolution concentrates on 2L+ multikinase TKI + histotype-specific ADC/IO.

Local therapy foundation:

NCIC SR2 (O’Sullivan 2002 Lancet, PMID 12103287) preop vs postop RT head-to-head → preop RT wins standard position via the classical function-preservation vs wound-complication tradeoff
Pisters brachytherapy MSKCC (1996 JCO) intraoperative brachytherapy reduces local recurrence in high-grade disease
EORTC 62961 (Issels 2010 Lancet Oncol, PMID 20434400) neoadjuvant chemo ± regional hyperthermia, DFS gain in large tumors
RTOG 0630 (2015) neoadjuvant RT volume reduction + image guidance to lower late fibrosis

Adjuvant chemotherapy (a 40-year controversy):

SMAC 1997 meta + Pervaiz 2008 meta (PMID 18521899) classic meta-analyses with marginal benefit
Frustaci 2001 JCO (Italian Sarcoma Group) epi+ifos adjuvant in high-risk limb STS — positive OS signal
EORTC 62931 (Woll 2012 Lancet Oncol, PMID 22954508) adjuvant doxorubicin+ifosfamide 5-year OS negative → ended the “all STS adjuvant chemo” paradigm
ISG-STS 1001 (Gronchi 2017 Lancet Oncol, PMID 28499583) first head-to-head: histotype-tailored adjuvant chemo is NOT superior to standard AI (epi+ifos) → negative evidence for histology-driven customization

Advanced 1L:

EORTC 62012 (Judson 2014 Lancet Oncol, PMID 24618336) doxorubicin vs doxo+ifos — combination improves PFS but not OS, with more toxicity → doxo monotherapy remains the 1L standard
GeDDiS (Seddon 2017 Lancet Oncol, PMID 28882536) gem+docetaxel vs doxo 1L STS non-inferior → gem+doc becomes an alternative for doxo-ineligible patients
JGDG phase 1b/2 (Tap 2016 Lancet, PMID 27291997) olaratumab+doxo positive → FDA accelerated approval 2016, then…
ANNOUNCE (Tap 2020 JAMA, PMID 32259228) phase 3 failed to replicate JGDG → FDA revoked approval, ending the PDGFRα-targeted 1L hope — a classical teaching case: phase 2 small samples cannot predict phase 3
LMS-04 (Pautier 2022 Lancet Oncol, PMID 35835135) doxorubicin+trabectedin 1L — LMS-specific positive → LMS dedicated path

Advanced 2L+ multikinase TKI:

PALETTE (EORTC 62072) (van der Graaf 2012 Lancet, PMID 22595799) pazopanib vs placebo in non-GIST STS 2L+ → FDA 2012, first TKI approved for STS 2L+
REGOSARC (Mir 2016 Lancet Oncol, PMID 27751846) regorafenib 2L+ expanding TKI options

Histotype-specific 2L+:

Schöffski 309 (2016 Lancet, PMID 26874885) eribulin vs dacarbazine 2L+ — positive in liposarcoma + LMS → FDA liposarcoma approval 2016
ET743-STS-301 (Demetri 2016 JCO, PMID 26371143) trabectedin vs dacarbazine 2L+ LMS/LPS → FDA 2015
SARC028 (Tawbi 2017 Lancet Oncol, PMID 28988646) pembrolizumab basket — UPS / ddLPS exclusive signal (LMS/synovial negative) → NCCN 2B for UPS/MFS/ddLPS

2020+ IO in the frontline:

SU2C-SARC032 (Mowery 2024 Lancet, PMID 39547252) neoadjuvant pembrolizumab + RT + surgery vs RT + surgery, high-risk UPS/MFS/ddLPS DFS benefit → first positive IO trial in STS neoadjuvant setting

Retroperitoneal STS:

STRASS (EORTC-62092) (Bonvalot 2020 Lancet Oncol, PMID 32941794) preop RT + surgery vs surgery alone — primary endpoint negative but liposarcoma subgroup signal → NCCN limits to “select LPS”
TARPSWG consensus (2015 Ann Surg Oncol, PMID 25316486) international multidisciplinary consensus: R0 extended resection + experience centers
Gronchi 2009 JCO extended multivisceral resection reduces recurrence in well-differentiated LPS

ALT/WDLPS precision era (MDM2 inhibitors):

MANTRA (Chawla 2023+) milademetan vs trabectedin, phase 3 ongoing
Brightline-1 (Schöffski 2024) brigimadlin phase 2/3 in MDM2 amplified ddLPS

Desmoid + Phyllodes

Desmoid’s unique biology. Does not metastasize but invades locally, ~90% carry CTNNB1 or APC mutations driving Wnt/β-catenin pathway. Pre-2015 the mainstream was “diagnose-then-resect + RT” with 15-30% recurrence; now shifted to active surveillance + systemic therapy on progression (limb/function preservation).

Active surveillance:

GRAFITI (van Houdt 2023 Ann Surg, PMID 35166264) first prospective active surveillance phase 3 → ended the “diagnose-then-resect” tradition
Colombo-Bonvalot pooled (2025 CCR) multicenter surveillance cohort supports

Systemic therapy (unresectable / progressive):

FNCLCC/FSG imatinib (Penel 2011 Ann Oncol, PMID 20622000) early KIT/PDGFR TKI signal
Palassini MTX+Vb (2017 Cancer J, PMID 28410293) long-term cohort establishing MTX/Vb as a low-intensity alternative
Alliance A091105 (Gounder 2018 NEJM, PMID 30575484) sorafenib vs placebo phase 3 → first positive RCT → NCCN 2A
DESMOPAZ (Toulmonde 2019 Lancet Oncol, PMID 31331699) pazopanib vs MTX/Vb phase 2 — 6-month non-progression rate 83.7% vs 45.0% → opened the desmoid TKI era
DeFi (Gounder 2023 NEJM, PMID 36884323) nirogacestat (γ-secretase inhibitor) vs placebo phase 3 → FDA 2023 approval, Notch pathway new mechanism — first approved desmoid drug

Phyllodes — limited evidence. Malignant phyllodes is exceedingly rare (<1% of breast tumors); RCTs are nearly impossible:

Barth 2009 Ann Surg Oncol (PMID 19424757) multi-institutional adjuvant RT prospective, first and only prospective evidence (N~25)
KROG 16-08 Choi 2018 Breast Cancer Res Treat (PMID 29808288) Korean multicenter retrospective of 362 patients, RT improves local control in the malignant subset

Pediatric / AYA RMS

NCCN explicit: adult RMS is treated per pediatric protocols (COG / EpSSG). No adult-specific landmark trials exist.

IRS / COG (North America) backbone:

IRS-IV (Crist 2001 JCO, PMID 11408506) VAC (vincristine/actinomycin/cyclophosphamide) backbone established
D9602 (Raney 2011 JCO) low-risk reduced cyclophosphamide feasible
D9803 (Arndt 2009 JCO) VAC vs VAC/VIE — no difference in intermediate risk
ARST0331 (Walterhouse 2014 JCO) low-risk shortened therapy
ARST0431 (Weigel 2016 JCO) intermediate-risk intensified vincristine/irinotecan
ARST0531 (Hawkins 2018 JCO, PMID 30091945) VAC/VI vs VAC non-inferior, lower cyclo dose → alternative option

EpSSG (Europe) divergence:

EpSSG RMS 2005 (Bisogno 2019 Lancet Oncol, PMID 31562043) 6-month vinorelbine + low-dose cyclo maintenance after standard induction — 5-year OS absolute gain 12.8% in high-risk non-metastatic RMS → European standard, not replicated in North America

Metastatic / relapsed:

BERNIE (Chisholm 2017 Eur J Cancer, PMID 28738258) bevacizumab + chemo in metastatic pediatric RMS — survival benefit not significant → anti-VEGF did not enter the standard

3. Current Controversies

Adult Mainstream STS

Adjuvant chemotherapy — 40-year unresolved controversy. EORTC 62931 negative vs SMAC/Pervaiz meta marginal positive vs ISG-STS 1001 histotype-tailored negative. The true conclusion is likely “only a subset of high-risk STS + specific histology (high-grade LMS / synovial / myxoid) benefit”, but no RCT has refined this. 2026 practice: selectively used in FNCLCC grade 3 + high-risk histology, not routine in most centers.

IO histotype-specificity. SARC028 signals only in UPS / ddLPS; LMS / synovial sarcoma show essentially no IO response. PD-L1 IHC in STS does not predict response (unlike other solid tumors). SU2C-SARC032 brought IO to the neoadjuvant setting but only for high-risk UPS/MFS/ddLPS. Future “inflamed vs immune desert” microenvironment classification is the key biomarker direction, but clinical assays are not validated.

The MDM2 inhibitor biomarker paradox. MDM2 amplification is near-universal in ALT/WDLPS (~95%). “Biomarker selection” in this setting is almost equivalent to “all ALT/WDLPS use MDM2i”. If MANTRA / Brightline-1 ultimately read out positive, it will be the first time STS achieves genotype-driven therapy.

China-led STS research is scarce. The core global STS RCTs are almost all driven by EORTC / SARC / FSG / ISG / COG. Chinese original STS research concentrates on real-world data within the CSCO Sarcoma guideline + single-institution retrospectives. This is a clear catch-up direction for Chinese oncology research in the next decade.

Desmoid + Phyllodes

Active surveillance vs early systemic — the entry threshold. GRAFITI established surveillance safety, but the “progression” definition is not unified (RECIST ≥20% vs symptom worsening vs QoL decline). This affects systemic therapy initiation timing and remains the most persistent decision-making uncertainty in desmoid care.

The desmoid three-drug ordering. After DeFi (2023 NEJM), nirogacestat became the only FDA-approved desmoid drug, but no head-to-head evidence exists vs sorafenib (A091105) or pazopanib (DESMOPAZ). NCCN lists all three in parallel. Clinical selection by:

nirogacestat — unique γ-secretase target · characteristic ovarian dysfunction toxicity
sorafenib — longest experience · hand-foot syndrome + hypertension
pazopanib — same family TKI · hypertension + hepatotoxicity

Phyllodes adjuvant RT’s real value. Barth prospective sample is too small (N~25); KROG 16-08 and other retrospectives support RT reducing local recurrence but no OS signal. Should all malignant phyllodes receive adjuvant RT, or only close-margin / high-risk? No RCT. Practice variation is enormous.

RMS

PAX::FOXO1 fusion replacing histological typing. 2020+ data show PAX3/7::FOXO1 fusion status provides stronger prognostic stratification than traditional alveolar (ARMS) vs embryonal (ERMS) histology. ARST2031 / RMS 2021 and other next-gen trials stratify by fusion status, but 2026 NCCN guidelines still use histology — transition-period disconnect.

Maintenance therapy — trans-Atlantic divergence. EpSSG RMS 2005’s 6-month vinorelbine+cyclo maintenance shows clear benefit, but North American COG has not replicated it (no maintenance in the ARST series). This divergence between the two major sarcoma communities directly affects clinical adoption in China — most Chinese centers lean toward the EpSSG model.

Relapsed / refractory RMS — standard gap. After BERNIE’s negative result, vincristine+irinotecan+temozolomide (VIT) or cyclo+topotecan rest on phase 2 evidence. No phase 3 standard exists.

4. Biomarker System

STS histology-as-biomarker

Histological subtype itself (>50 WHO 2020 subtypes) is the most important “biomarker” in STS. Subtype → paradigm switch: LMS → anthra+trabec · myxoid LPS → trabec · UPS/ddLPS → IO signal · ASPS → TKI/IO. LMS / LPS / synovial / undifferentiated / rhabdo / fibrous each walk their own path.
FNCLCC grade (1-3) — adjuvant decision + surveillance frequency; grade 3 + size > 5cm + deep = high risk
MDM2 amplification (FISH / NGS) — ALT vs normal fat differential diagnosis + brigimadlin/milademetan eligibility

Desmoid biomarker

CTNNB1 mutation (T41A / S45F / S45P) vs wild-type (usually APC+) — recurrence risk stratification + spontaneous regression prediction (S45F has lower spontaneous regression rate)
APC mutation — gateway to Gardner / FAP syndrome screening (clinically mandatory)

RMS biomarker

PAX3::FOXO1 / PAX7::FOXO1 fusion — ARMS hallmark, strongly associated with unfavorable outcome. In 2026, COG actually stratifies enrollment by fusion status.
MyoD1 / Myogenin IHC — RMS standard differential (vs synovial / Ewing / NRSTS)
DICER1 germline — pediatric RMS multi-organ syndrome screening

Pan-sarcoma tumor-agnostic

NTRK fusion (larotrectinib / entrectinib) — rare but enriched in infantile fibrosarcoma and select subtypes
MSI-H / dMMR — rare (<2% STS), pembrolizumab tumor-agnostic eligible
TMB-H ≥10 mut/Mb — extremely rare, but NCCN v3.2026 SARC-G includes it in subsequent lines

Why PD-L1 “fails” in STS

In other solid tumors, PD-L1 CPS / TPS broadly predicts IO response, but not in STS. Possible reason: STS microenvironment heterogeneity stems from histology (mesenchymal origin) rather than immune escape. Future “immune signatures” (CXCL9/10 / TILs density / tertiary lymphoid structures) may be more actionable than PD-L1, but no validated clinical assay exists today.

5. Time-Space Overview

Five paradigm lines, 1996-2025 (42 landmark trials sorted by era):

① 1996-2005: Surgery + RT foundation

1996 Pisters brachytherapy MSKCC
1997 SMAC meta-analysis
2001 IRS-IV (RMS VAC backbone)
2001 Italian Sarcoma Group Frustaci (STS adjuvant epi+ifos)
2002 NCIC SR2 preop vs postop RT

② 2005-2012: Adjuvant chemo settles + TKIs enter

2008 Pervaiz meta
2009 D9803 (RMS VAC/VIE negative) · Gronchi retro extended resection
2010 EORTC 62961 hyperthermia
2011 FNCLCC/FSG imatinib desmoid · D9602 · ARST0331 (RMS low-risk de-escalation)
2012 EORTC 62931 (STS adjuvant chemo phase 3 negative) · PALETTE (pazopanib 2L+ FDA)

③ 2012-2020: 2L+ ADC + TKI + IO breakthrough

2014 EORTC 62012 doxo mono 1L standard
2015 TARPSWG consensus · RTOG 0630
2016 JGDG olaratumab phase 2 hope · Schöffski 309 (eribulin LPS FDA) · Demetri ET743-STS-301 (trabectedin FDA) · REGOSARC · ARST0431 (RMS intermediate)
2017 GeDDiS (gem+doc 1L non-inferior) · ISG-STS 1001 (histotype-tailored neoadjuvant negative) · SARC028 (pembro basket UPS/ddLPS signal) · Palassini MTX+Vb desmoid · BERNIE (RMS bev negative)
2018 Alliance A091105 sorafenib desmoid phase 3 +ve · ARST0531 (RMS VAC/VI non-inferior)
2019 DESMOPAZ pazopanib desmoid phase 2 · EpSSG RMS 2005 maintenance

④ 2020-2023: High-signal IO + precision ADC break barriers

2020 STRASS retro RT primary negative (LPS signal) · ANNOUNCE olaratumab phase 3 failure
2022 LMS-04 doxo+trabec 1L LMS +ve
2023 DeFi nirogacestat desmoid FDA · GRAFITI (prospective surveillance)

⑤ 2024-2025: IO moves upstream + MDM2i + maintenance expand

2024 SU2C-SARC032 pembro+RT neoadjuvant UPS/MFS/ddLPS +ve · Brightline-1 brigimadlin ALT/ddLPS
2025 Colombo-Bonvalot surveillance pooled · MANTRA phase 3 ongoing

6. Full Trial Table

Sorted by bucket + sub-subtype + year. Each entry has a PMID hyperlink (✓ = verified, null = NCT[si] not indexed or no PubMed record).

Adult Mainstream STS (25 trials)

Extremity / body wall / head and neck STS (9 trials)

1996 · Pisters brachytherapy MSKCC — STS extremity adjuvant brachytherapy — PMID 8622034
1997 · SMAC meta-analysis — STS adjuvant anthracycline classic meta — PMID 9400508
2001 · Italian Sarcoma Group (Frustaci) — STS adjuvant epi+ifos high-risk limb — PMID 11230464
2002 · NCIC SR2 (O’Sullivan) — STS preop vs postop RT → preop RT wins standard — PMID 12103287
2008 · Pervaiz meta-analysis — STS adjuvant anthra+ifos updated meta — PMID 18521899
2012 · EORTC 62931 — STS adjuvant doxo+ifos phase 3 negative — PMID 22954508
2015 · RTOG 0630 — STS neoadjuvant image-guided RT volume reduction — PMID 25667281
2017 · ISG-STS 1001 — STS neoadjuvant histotype-tailored vs standard AI negative — PMID 28499583
2024 · SU2C-SARC032 — Neoadjuvant pembro+RT+surgery in UPS/MFS/ddLPS positive — PMID 39547252

Cross-subtype systemic (STS_all, 11 trials)

2010 · EORTC 62961 hyperthermia — Neoadjuvant chemo + deep hyperthermia large-tumor benefit — PMID 20434400
2012 · PALETTE (EORTC 62072) — pazopanib 2L+ non-GIST STS first TKI — PMID 22595799
2014 · EORTC 62012 — doxo vs doxo+ifos 1L → doxo mono standard — PMID 24618336
2016 · JGDG (Olaratumab phase 1b/2) — olaratumab+doxo 1L phase 2 positive → FDA accelerated → ANNOUNCE revoked — PMID 27291997
2016 · Schöffski 309 (eribulin) — eribulin vs dacarbazine 2L+ LPS/LMS positive — PMID 26874885
2016 · ET743-STS-301 (trabectedin) — trabectedin vs dacarbazine 2L+ LMS/LPS FDA — PMID 26371143
2016 · REGOSARC — regorafenib 2L+ expanding TKI options — PMID 27751846
2017 · GeDDiS — gem+doc vs doxo 1L STS non-inferior — PMID 28882536
2017 · SARC028 — pembrolizumab basket UPS/ddLPS signal — PMID 28988646
2020 · ANNOUNCE — olaratumab+doxo phase 3 failure — PMID 32259228
2022 · LMS-04 — doxo+trabectedin 1L LMS exclusive positive — PMID 35835135

Retroperitoneal (3 trials)

2009 · Gronchi extended compartmental resection — retro multivisceral resection Italian series — PMID 19273692
2015 · TARPSWG consensus — retro international multidisciplinary consensus — PMID 25316486
2020 · STRASS (EORTC-62092) — retro preop RT vs surgery primary negative (LPS signal) — PMID 32941794

ALT/WDLPS MDM2 era (2 trials)

2023 · MANTRA — milademetan vs trabectedin phase 3 ongoing — abstract only
2024 · Brightline-1 — brigimadlin in MDM2 amplified ddLPS — abstract only

Desmoid + Phyllodes (9 trials)

Desmoid (7 trials)

2011 · FNCLCC/FSG imatinib desmoid (Penel) — early KIT/PDGFR TKI signal — PMID 20622000
2017 · Palassini MTX+Vb — long-term cohort MTX/Vb low-intensity alternative — PMID 28410293
2018 · Alliance A091105 — sorafenib vs placebo phase 3 first +ve RCT — PMID 30575484
2019 · DESMOPAZ — pazopanib vs MTX/Vb phase 2 opens TKI era — PMID 31331699
2023 · GRAFITI — prospective active surveillance phase 3 — PMID 35166264
2023 · DeFi — nirogacestat (γ-secretase) vs placebo → FDA first approved desmoid drug — PMID 36884323
2025 · Colombo-Bonvalot pooled active surveillance — PMID 39620931

Phyllodes (2 trials)

2009 · Barth adjuvant RT phyllodes — multi-institutional prospective N~25, first prospective — PMID 19424757
2018 · KROG 16-08 (Choi) — Korean multicenter retrospective 362 patients — PMID 29808288

Pediatric / AYA RMS (8 trials)

2001 · IRS-IV — VAC backbone established — PMID 11408506
2009 · D9803 — VAC vs VAC/VIE no intermediate difference — PMID 19770373
2011 · D9602 — low-risk reduced cyclo feasible — PMID 21357783
2014 · ARST0331 — low-risk shortened therapy — PMID 25267746
2016 · ARST0431 — intermediate-risk intensified vincr/irinotecan — PMID 26503200
2017 · BERNIE — bevacizumab + chemo metastatic pediatric RMS negative — PMID 28738258
2018 · ARST0531 — VAC/VI vs VAC non-inferior — PMID 30091945
2019 · EpSSG RMS 2005 — European high-risk maintenance vinorelbine+cyclo 6 mo, OS +12.8% — PMID 31562043

Key Research Gaps

STS adjuvant chemo: 40-year controversy, histotype-tailored not superior to standard AI (ISG-STS 1001 negative)
IO histotype-specificity in STS: which subtypes beyond the basket benefit? Lack of predictive biomarker
MDM2 inhibitor phase 3 mature readout: awaiting 2026-2027
China-led STS RCT landmarks: globally scarce
Adult RMS: entirely extrapolated from pediatric protocols, no adult-specific RCT
Desmoid three-drug head-to-head: DeFi / Alliance A091105 / DESMOPAZ not compared
RCT-level evidence for phyllodes adjuvant RT: permanently impossible (rarity + ethics)

7. Sources

7.1 Pivotal Trial PMID List (40 verified / 95% coverage)

PMIDs all double-verified via NCBI esummary + NCT[si]+year+journal+title guard or author+year+journal — see §6 per-entry hyperlinks.

Only 2 null entries: MANTRA (milademetan phase 3 ongoing) and Brightline-1 (brigimadlin phase 2/3 ongoing) — both have only ASCO/ESMO abstracts, primary publications pending.

7.2 Guideline Citation

NCCN Soft Tissue Sarcoma v3.2026 (March 12, 2026) — primary reference
CSCO Soft Tissue Sarcoma Treatment Guideline — China complement (not in this repo, to be added)
ESMO Clinical Practice Guidelines: Soft Tissue and Visceral Sarcomas (2021 + 2025 update)
ESMO/EURACAN/GENTURIS guidelines: Desmoid-type fibromatosis (2020)

7.3 Methodology

This post’s 42 landmark trials were extracted from NCCN STS v3.2026 via a pilot + 3-way parallel agent extract workflow (docs/preflight/sarcoma-agent-shared-methodology.md):

Chunk split: NCCN PDF split into 5 chunks by 6 subtypes, landed in private/preflight/sarcoma/chunks/
Agent A (pilot): Desmoid + Phyllodes (9 trials)
Schema-diff gate: parallel launch only after 0 drift verified
Agent B/C/D (parallel): mainstream STS / retro+ALT/WDLPS / RMS (33 trials)
PMID hunt: agent raw PMID fab rate ~45% (matches pitfalls.md P0 pattern), all dropped; 22 manually esummary-verified + scripts/pmid-hunt.py NCT[si] filled 8 more + author+year+journal esearch filled 10 more → 95% verified coverage (40/42, only 2 ongoing phase 3 left)
Translate: scripts/translate-trials.py via OpenRouter claude-haiku-4.5, 42/42 success × 0% raw fail

All intermediate artifacts are reproducible — scripts in scripts/ · shared methodology doc in docs/preflight/sarcoma-agent-shared-methodology.md.

The Clinical Trials Timeline

(→ /en/trials/sarcoma/ site-wide 42-sarcoma landmark trials, filterable by year + subtype)

Closing

Soft tissue sarcoma is the hardest subject in solid-tumor oncology: >50 subtypes × extreme rarity × independent per-subtype biology. But precisely because of the rarity, every landmark trial in the past 30 years rests on 10-15 years of multicenter accumulation by international cooperative groups (EORTC · SARC · FSG · ISG · COG · EpSSG · TARPSWG).

The treatment wisdom most worth learning is not “STS 1L uses doxorubicin” broad conclusions but “histology dictates the path”:

When you see LMS, think doxo+trabectedin (LMS-04)
When you see myxoid LPS, think trabectedin (Grosso 2007 signal)
When you see UPS / ddLPS, think IO (SARC028 / SU2C-SARC032)
When you see desmoid, don’t rush to cut (GRAFITI → nirogacestat / sorafenib / pazopanib three options)
For RMS, pediatric protocols extrapolate to young adults (COG + EpSSG two schools)

Dual Brain Lab (csilab.net) · Twelfth tumor type on the map · → Next: 12-tumor cross-view