Bone Tumor Clinical Trial Timeline: In-Depth Report

Coverage: 35 landmark trials cited by NCCN Bone Cancer V2.2026 (34 published, all PMIDs traceable + 1 rEECur phase 3 primary manuscript pending) + 5 subtypes (osteosarcoma / Ewing sarcoma / chondrosarcoma / chordoma / GCTB)

Curated by Dual Brain Lab (csilab.net)

1. One-Sentence Definition

This report traces the evolution and current decision landscape of systemic therapy for bone sarcomas (osteosarcoma + Ewing sarcoma + chondrosarcoma + chordoma + GCTB — five subtypes combined) over the past 40 years (1986-2026), focusing on landmark clinical trials cited in NCCN Bone Cancer V2.2026, to provide frontline clinicians with a traceable panoramic map for 2026 decisions on “who, what, and why.”

Iron rule: every data point of every trial is traceable to PubMed (PMID) or ClinicalTrials.gov (NCT id) — every [PMID xxxxxxxx] in the text can be opened directly in PubMed for verification.

2. Vertical: Timeline of Four Treatment Paradigms

Bone tumor systemic therapy has evolved over 40 years along a logic completely different from NSCLC or BTC — the mainstream major subtypes (osteosarcoma + Ewing sarcoma) have held a 40-year chemo backbone (MAP / VDC-IE) undefeated; the rare subtypes (chondrosarcoma + GCTB + chordoma) instead achieved precision-targeted breakthroughs first after 2010; and IO has failed across virtually all subtypes. These three forces together shape the “dual-track reality” of the 2026 bone tumor decision landscape: mainstream stagnation + rare-subtype precision breakthroughs + IO failure.

The scale of each paradigm shift is far smaller than the five leaps seen in NSCLC — the reason lies in the unique biology of bone tumors: low TMB (tumor mutational burden) + cold tumor immune-desert + osteosarcoma / Ewing lacking a unified driver gene + rare subtypes conversely having clear druggable drivers (IDH1 / RANKL / PDGFR).

2.1 Osteosarcoma MAP 40-Year Backbone Era (1986-2026): From Foundation to All Intensification Failures + Marginal TKI Gains in Relapse

Story: MIOS 1986 moved adjuvant chemotherapy from “optional” to “mandatory,” pushing 2-year RFS (relapse-free survival) from 17% to 66%. Over the following 40 years all intensification strategies — adding ifosfamide, adding interferon, adding MTP-PE, switching to neoadjuvant sequencing, switching to targeted agents — failed to further improve OS. Relapsed osteosarcoma entered the multi-TKI (tyrosine kinase inhibitor) era, where sorafenib / regorafenib / cabozantinib / pazopanib — all VEGFR-axis drugs — delivered “4-6 month PFS marginal wins” without anyone changing OS.

• MIOS [PMID 3520317] (Link 1986 NEJM, N=113): non-metastatic high-grade extremity osteosarcoma, postoperative adjuvant multi-drug chemotherapy (high-dose methotrexate + doxorubicin + BCD + cisplatin) vs surgery alone with observation. 2-year RFS 66% vs 17%, OS similarly significantly improved. Established the foundation that “osteosarcoma adjuvant chemo is mandatory”; all subsequent trials used this as baseline, and the 60-70% EFS (event-free survival) ceiling has been fixed here ever since.
• POG-8651 [PMID 12697883] (Goorin 2003 JCO, N=106): neoadjuvant chemotherapy + surgery + adjuvant chemotherapy vs upfront surgery + adjuvant chemotherapy. 5-year EFS 61% vs 69%, OS no difference. A rarely cited “inconvenient truth” — the “neoadjuvant is standard” narrative has no RCT support on survival data; it became standard because neoadjuvant enables limb-sparing surgery + makes histologic response stratification a prognostic tool.
• COSS-86 [PMID 9789613] (Fuchs 1998 Ann Oncol, N=171): high-risk osteosarcoma MAP backbone plus ifosfamide. vs historical control COSS-82, 10-year EFS/OS no significant improvement. Europe’s earliest intensification-failure signal, foreshadowing EURAMOS-1.
• INT-0133 [PMID 18235123] (Meyers 2008 JCO, N=662): MAP ± ifosfamide ± MTP-PE (mifamurtide, muramyl tripeptide phosphatidylethanolamine), 2×2 factorial design. MTP-PE arm 6-year OS 78% vs 70% (p=0.03); ifosfamide addition failed (EFS p=0.39). The only “possible” frontline OS-positive signal in 40 years, but factorial interaction left interpretation contested: FDA refused approval of mifamurtide, EMA approved. Ifosfamide failed for the first time here.
• EURAMOS-1 Good Responders [PMID 26033801] (Bielack 2015 JCO, N=716): patients with good histologic response after neoadjuvant MAP (<10% viable tumor cells), adjuvant MAP + pegylated interferon alfa-2b maintenance vs MAP alone. 3-year EFS 77% vs 74%, HR 0.83 (p=0.21) negative. IFN toxicity caused about 40% of patients not to complete planned treatment.
• EURAMOS-1 Poor Responders [PMID 27569442] (Marina 2016 Lancet Oncol, N=618): patients with poor histologic response after neoadjuvant MAP (≥10% viable tumor cells), postoperative MAPIE (MAP + ifosfamide + etoposide) vs continued MAP. EFS HR 0.98 (95% CI 0.78-1.23, p=0.86) completely negative. Risk of secondary AML (acute myeloid leukemia) rose notably. The largest frontline RCT in osteosarcoma history (two-arm total N=1334), both arms negative — the 2000s strategy of “intensify for poor responders” was definitively shut down here.
• SORAFENIB-ISG [PMID 21527590] (Grignani 2012 Ann Oncol, N=35): sorafenib monotherapy in relapsed / unresectable osteosarcoma. 4-month PFS 46%, mPFS 4 months, mOS 7 months, ORR 8%. The first positive TKI signal in osteosarcoma, benefit short but reproducible; paved the way for subsequent VEGFR TKIs.
• SORAFENIB-EVEROLIMUS [PMID 25498219] (Grignani 2015 Lancet Oncol, N=38): sorafenib + everolimus (mTOR inhibitor) combination in relapsed osteosarcoma. 6-month PFS 45%, failed to meet prespecified 50% threshold; notable toxicity. Adding mTOR didn’t help — the osteosarcoma TKI story is mainly about the VEGFR angiogenesis axis.
• SARC024 [PMID 31013172] (Davis 2019 JCO, N=42 randomised): regorafenib vs placebo (crossover allowed) in adult relapsed / metastatic osteosarcoma. mPFS 3.6 vs 1.7 months (HR 0.42, p=0.017), 8-week progression-free rate ~65% vs 0%, no RECIST response. The first positive PFS signal in modern relapsed osteosarcoma.
• REGOBONE [PMID 30477937] (Duffaud 2019 Lancet Oncol, N=38): independent European validation of SARC024, also regorafenib vs placebo. 8-week progression-free rate 17/26 vs 0/12, mPFS 16.4 vs 4.1 weeks. Cross-continental replication is rare in sarcoma — SARC024 + REGOBONE together pinned regorafenib as a 2L option in relapsed osteosarcoma.
• CABONE [PMID 32078813] (Italiano 2020 Lancet Oncol, dual cohorts N=45 each): cabozantinib (60 mg/d) monotherapy in relapsed osteosarcoma + relapsed Ewing sarcoma. Osteosarcoma 6-month PFR 33%, ORR 12%; Ewing sarcoma 6-month PFR 26%, ORR 26% (one of the highest ORRs ever recorded in relapsed Ewing). The only relapsed-osteosarcoma TKI producing real RECIST tumor shrinkage.
• PAZO-OSTEO [PMID 35075361] (Frankel 2022 J Oncol, N=12, slow accrual and early termination): pazopanib 800 mg/d in relapsed osteosarcoma with lung metastases. PFS numbers in the same band as sorafenib / regorafenib / cabozantinib, confirming VEGFR TKI class effect.
• AOST1321 [PMID 41159913] (Janeway 2026 Clin Cancer Res, N=40): denosumab (RANKL inhibitor) monotherapy in relapsed / refractory pediatric + AYA osteosarcoma, two cohorts. Failed to meet prespecified efficacy threshold, both cohorts negative. The “hijacking osteoclast biology” hypothesis was refuted in osteosarcoma — sharp contrast with the near-universal response of GCTB to denosumab: the same drug is landscape-changing in one bone tumor and inactive in another, a textbook example of bone tumor heterogeneity.

Takeaway: in 2026, frontline osteosarcoma SoC (standard of care) = the MIOS 1986 MAP chemotherapy framework continued. All intensification failed (EURAMOS-1 both arms / COSS-86 / MTP-PE contested). Relapsed osteosarcoma = VEGFR TKI marginal PFS (regorafenib the archetype, double-validated by SARC024 + REGOBONE; cabozantinib produces real shrinkage). Denosumab has no antitumor activity in osteosarcoma itself — retained only for skeletal-related-event prevention.

2.2 Ewing Sarcoma: Interval Compression Marginal Wins + High-Risk Transplant + First-Ever Relapse RCT rEECur (2003-2024)

Story: after INT-0091 established the Ewing sarcoma VDC-IE backbone (vincristine / doxorubicin / cyclophosphamide / ifosfamide / etoposide) in 2003, subsequent progress came not from new drugs but from trial-design optimization: AEWS0031 compressed dosing interval from every 3 weeks to every 2 weeks (“timing matters more than new drugs”) to grab 8 percentage points of EFS; EURO-EWING-99-R2 added busulfan-melphalan (BuMel) + autologous HSCT in high-risk localized disease to grab 14 percentage points. rEECur 2024 used Bayesian adaptive design to give relapsed Ewing sarcoma its first RCT-level answer.

• INT-0091 [PMID 12594313] (Grier 2003 NEJM, N=518): newly diagnosed localized or metastatic Ewing family tumors (<30 years), alternating VDC/IE vs VDC alone. Localized 5-year EFS 69% vs 54% (p=0.005), 5-year OS 72% vs 61%; no benefit in metastatic subgroup. Established VDC-IE as the North American Ewing sarcoma standard and for the first time displayed the biological watershed between localized and metastatic disease.
• AEWS0031 [PMID 23091096] (Womer 2012 JCO, N=587): newly diagnosed localized Ewing, VDC/IE every 2 weeks vs every 3 weeks (interval compression with G-CSF support). 5-year EFS 73% vs 65% (p=0.048), no additional toxicity. “Same chemo, more frequent” = 8 percentage points of EFS. Dose-dense VDC/IE q2w became the North American SoC. This is the only clear frontline improvement in Ewing sarcoma — not a new drug, but a trial-design win.
• EURO-EWING-99-R1 [PMID 24982464] (Le Deley 2014 JCO, N=856): standard-risk localized Ewing, VAC (vincristine+actinomycin+cyclophosphamide) vs VAI (vincristine+actinomycin+ifosfamide) consolidation after VIDE induction. 3-year EFS 75.4% vs 78.2%, HR 1.12 (CI 0.89-1.41) non-inferiority achieved. Cyclophosphamide consolidation is non-inferior to ifosfamide with less renal toxicity — the de-escalation answer for standard-risk disease.
• EURO-EWING-99-R2 [PMID 30188789] (Whelan 2018 JCO, N=240): high-risk localized Ewing (poor response after induction or large axial tumor), BuMel + autologous HSCT vs conventional VAI consolidation. 3-year EFS 67.1% vs 52.9%, HR 0.64 (CI 0.43-0.95, p=0.026); OS improved similarly. One of the few positive results in modern osteosarcoma / Ewing intensification trials — the trick was “use high intensity only in the high-risk subgroup.”
• VIT-WAGNER [PMID 16317751] (Wagner 2007 Pediatr Blood Cancer, N=14 pediatric): irinotecan + temozolomide (no vincristine in original version) in progressive pediatric Ewing family tumors. ORR 63%, median response duration 8.3 months. Subsequently MSK / CHOP / Warsaw / Virginia groups added vincristine to evolve into the VIT three-drug regimen — the historical seed of the entire VIT narrative.
• VIT-RACIBORSKA [PMID 23776128] (Raciborska 2013 Pediatr Blood Cancer, N=22, Poland): VIT (vincristine + irinotecan + temozolomide) three-drug regimen in relapsed / refractory Ewing. ORR ~68%, mOS ~20 months. VIT consistently produced ~60-70% ORR across multiple countries and cohorts yet never entered an RCT — a textbook case of an “evidence-weak regimen becoming practical standard by pragmatic convergence.”
• rEECur [NCT03416517] (McCabe 2024 Lancet Oncol manuscript pending): the first Bayesian adaptive multi-arm multi-stage RCT in relapsed Ewing (N=451 cumulative). Four arms vs historical topotecan+cyclophosphamide (TC) reference: gemcitabine+docetaxel (GD) eliminated at 2020 interim; irinotecan+temozolomide (IT) eliminated at 2022 interim; high-dose ifosfamide won in the phase 3 stage and became the new reference standard for relapsed Ewing. PMID not yet indexed in PubMed at the time of this report (2026-04-21); cited by NCT ID + conference readout.

Takeaway: in 2026 Ewing sarcoma frontline = North American AEWS0031 dose-dense VDC/IE q2w + high-risk localized disease goes to EURO-EWING-99-R2 BuMel HSCT; relapsed 2L = rEECur winner high-dose ifosfamide (replacing the pragmatic-standard VIT of many years), VIT retained in NCCN guidelines but demoted; cabozantinib (CABONE Ewing arm ORR 26%) has genuine tumor-shrinking activity as an off-label monotherapy.

2.3 Rare-Subtype Precision Therapy: IDH1 / RANKL / PDGFR — First Three Driver Lines Reaching Targeted Therapy (2010-2025)

Story: osteosarcoma + Ewing sarcoma have no clear druggable driver genes; but the rare subtypes each have an independent driver — chondrosarcoma IDH1/2 mutation ~50% central type, GCTB RANKL signaling axis, chordoma brachyury + PDGFR pathway. “Where there’s a driver, there’s a way” also holds true in bone tumors — the stage just moves to the rare subtypes.

Naming pitfall warning: ivosidenib (Tibsovo / AG-120, Agios/Servier) = selective IDH1 inhibitor for chondrosarcoma + AML + BTC; vorasidenib (Voranigo / AG-881, Agios/Servier) = dual IDH1/2 inhibitor for glioma. These are most easily confused in 2024-2025 literature — chondrosarcoma uses ivosidenib, not vorasidenib.

Chondrosarcoma IDH1 + BRCAness (synthetic lethality) Path

• IVOSIDENIB-CHONDRO-P1 [PMID 32208957] (Tap 2020 JCO, N=21 single-arm phase 1): advanced IDH1 R132-mutant conventional chondrosarcoma, ivosidenib 500 mg QD. RECIST ORR 0/21, DCR (disease control rate) 52%, mPFS 5.6 months, 39% of patients with PFS > 6 months; plasma 2-HG (2-hydroxyglutarate, the IDH1-mutant metabolite) suppressed > 93%. No RECIST PR (partial response) but PFS + biomarker dual evidence shows “on-target effect.” The first “precision therapy works” signal in bone tumors.
• IVOSIDENIB-CHONDRO-LT [PMID 40100120] (Tap 2025 Clin Cancer Res, N=21 long-term follow-up): same cohort with long-term follow-up. Updated mPFS ~7.4 months; subgroup on treatment > 2 years; safety benign (manageable QTc prolongation, fatigue). The mature readout anchored IDH1 inhibition as the first real precision-therapy win in bone sarcoma, directly supporting NCCN listing of ivosidenib for IDH1+ chondrosarcoma.
• OLAPARIB-IDH-SARC [PMID 34994649] (Eder 2021 JCO Precis Oncol, N=15): olaparib (PARP inhibitor) monotherapy in IDH1/2-mutant advanced mesenchymal sarcomas (mostly chondrosarcoma). CBR (clinical benefit rate) 11/15 (73%), no RECIST PR, mPFS ~8 months. Mechanistic basis: IDH mutation → 2-HG accumulation → inhibits α-KG-dependent DNA demethylation / HR repair → BRCAness (BRCA-deficient-like) phenotype → PARP-inhibitor sensitivity. The “second shot” of the IDH narrative — synthetic lethality rather than direct enzyme inhibition.
• SARC-PAZO-CHONDRO [PMID 31509242] (Chow 2020 Cancer, N=47 single-arm phase 2): pazopanib 800 mg QD in unresectable / metastatic chondrosarcoma. 16-week DCR 43%, ORR ~5%, mPFS 7.9 months, conventional-type mOS ~27 months. Dedifferentiated type has worse prognosis. Disease-stabilization signal for anti-angiogenic TKI in chondrosarcoma — a non-IDH option.
• REGOBONE-CHONDRO [PMID 33895682] (Duffaud 2021 Eur J Cancer, N=46 randomised placebo-controlled): regorafenib vs placebo in metastatic / locally advanced chondrosarcoma. 12-week non-progression rate 44% vs 24%, mPFS 19.8 vs 8.0 weeks, no RECIST PR. Running a placebo-controlled RCT in a rare subtype is difficult — another anti-angiogenic TKI disease-stabilization piece of evidence.

GCTB (Giant Cell Tumor of Bone) RANKL Path

• DENOSUMAB-GCTB-P2 [PMID 20149736] (Thomas 2010 Lancet Oncol, N=37): denosumab 120 mg SC (days 1, 8, 15, 29 + Q4W) in relapsed / unresectable GCTB. Histologic response rate 86% (30/35, defined as >90% giant-cell elimination or no radiographic progression), symptoms markedly relieved. RANKL validated as central driver in GCTB — a concept-breakthrough phase 2.
• DENOSUMAB-GCTB-INTERIM [PMID 23867211] (Chawla 2013 Lancet Oncol, N=282): three-parallel-cohort (unresectable / high-morbidity surgery / phase 2 rollover) expansion. Unresectable cohort 6-month progression-free 96%, high-morbidity-surgery cohort 48% avoided planned surgery or underwent downsized surgery, ONJ (osteonecrosis of the jaw) ~1%. FDA-approval basis dataset; established the “surgical de-escalation” role.
• DENOSUMAB-GCTB-LT [PMID 31704134] (Chawla 2019 Lancet Oncol, N=532 long-term follow-up): the largest GCTB denosumab dataset, followed up to ~5 years. Unresectable cohort annual progression rate still < 5%; cumulative ONJ ~5%, atypical femur fracture rare. Mature readout locking denosumab as modern GCTB SoC + establishing the “use-and-pause” intermittent dosing paradigm. GCTB evolved from “unresectable = fatal” to “medical therapy → limb-sparing surgery” — the clinical pathway was completely rewritten.

Chordoma brachyury / PDGFR Path

• IMATINIB-CHORDOMA-P2 [PMID 22331945] (Stacchiotti 2012 JCO, N=50): imatinib 800 mg QD in PDGFRB-positive advanced chordoma. RECIST ORR 2%, DCR 64%, mPFS 9 months; Choi-criteria response rate higher than RECIST (density change). PDGFRB-pathway inhibition = the first systemic therapy in the smallest bone-tumor histology; Choi provided an alternative yardstick for chordoma assessment.
• IMATINIB-EVEROL-CHORDOMA [PMID 30216418] (Stacchiotti 2018 Cancer, N=43): imatinib + everolimus combination after imatinib failure. ORR 2/43 (5%), Choi response 20/43, mPFS 11.5 months. Combination yielded limited increment over monotherapy.
• GI6301-CHORDOMA [PMID 33594772] (DeMaria 2021 Oncologist, N=34 randomised): brachyury vaccine (GI-6301, inactivated yeast expressing brachyury) + standard radiotherapy vs placebo + radiotherapy in locally advanced unresectable chordoma. 6-month ORR 35% vs 18% (not significant, small sample); brachyury-specific immune response detected in most vaccinated patients. The only randomised brachyury-directed study in chordoma — not formally statistically significant but keeps the brachyury-targeting program alive.

Takeaway: in 2026 rare-subtype bone tumors have entered the “where there’s a driver, there’s a way” precision era: (a) chondrosarcoma IDH1+ → ivosidenib (NCCN-listed, based on two CHONDRO-P1 + LT readouts) + PARP as the BRCAness second shot (OLAPARIB-IDH-SARC); (b) GCTB → denosumab rewriting surgical pathway (Thomas 2010 → Chawla 2013 / 2019 three phase 2 mature datasets); (c) chordoma → imatinib as backbone (Stacchiotti 2012), with combinations + brachyury vaccine providing modest increments.

2.4 Cross-Subtype Immunotherapy (2014-2024): SARC028 All-Failed + Alliance A091401 Soft-Tissue Door Left Ajar + Chordoma Exception

Story: IO (immune checkpoint inhibitor) posts 40%+ ORR in MSI-H colorectal cancer, 33% in melanoma, and rewrote the 1L backbone in NSCLC. In bone tumors it has failed almost completely — the reason is biology: low TMB + immune-desert + low PD-L1 expression. But SARC028 saw 20% response in dedifferentiated chondrosarcoma + Alliance A091401 combination-IO data + Migliorini chordoma exceptional case series left three narrow windows open for the “IO in bone” story.

• SARC028 [PMID 28988646] (Tawbi 2017 Lancet Oncol, bone subgroup n=40): pembrolizumab 200 mg Q3W monotherapy phase 2 in advanced osteosarcoma / Ewing / chondrosarcoma / dedifferentiated chondrosarcoma. Osteosarcoma ORR 5% (1/22), Ewing 5% (1/13), chondrosarcoma 0% (0/5), dedifferentiated chondrosarcoma 20% (1/5). The landmark trial for “IO across-the-board rout in bone tumors” — not a trial-design problem, but a biologically determined one. The 20% in dedifferentiated chondrosarcoma is the only signal worth chasing.
• ALLIANCE-A091401 [PMID 29370992] (D’Angelo 2018 Lancet Oncol, N=96 across bone + STS): nivolumab monotherapy vs nivolumab + ipilimumab combination, randomised non-comparative (no head-to-head p value). ORR 5% vs 16%; responses concentrated in soft tissue sarcomas (UPS, leiomyosarcoma, myxofibrosarcoma, angiosarcoma), with osteosarcoma contributing little. Combination IO tripled monotherapy response in soft tissue — leaving a narrow IO-combination door open.
• ALLIANCE-A091401-EXPANSION [PMID 39343511] (Seligson 2024 J Immunother Cancer, expansion N=89): histology-enriched expansion cohorts (UPS, dedifferentiated liposarcoma, dedifferentiated chondrosarcoma). Nivo monotherapy ORR ~10%, nivo+ipi ~21% in selected histologies; osteosarcoma cohort failed to meet prespecified activity threshold. Dedifferentiated chondrosarcoma is the bone subtype most likely to respond to IO — still not at practice-changing magnitude.
• CHORDOMA-IO-MIGLIORINI [PMID 28919999] (Migliorini 2017 Oncoimmunology, N=3 compassionate case series): anti-PD-1 compassionate use in recurrent chordoma after prior surgery / RT / imatinib failure. All 3/3 patients had clinical + radiographic response (SD to PR lasting several months); correlative analysis showed PD-L1 expression and immune infiltration in chordoma tissue. The first report of chordoma response to IO — despite small N, hypothesis-generating value is high: chordoma is the only potential exception to the IO-all-failed rule in bone, worth large-sample validation.

Takeaway: in 2026 IO use in bone tumors is restricted to the following three categories only: (a) dedifferentiated chondrosarcoma (SARC028 + Alliance expansion signal); (b) chordoma compassionate use (Migliorini + subsequent prospective pending); (c) tumor-agnostic MSI-H / dMMR / TMB-H subgroups (rare). IO monotherapy not recommended for osteosarcoma + Ewing sarcoma + classic chondrosarcoma — off-label use offers extremely low clinical benefit plus unexpected toxicity plus insurance-denial risk.

3. Horizontal: The 2026 Decision Landscape (Six Dimensions)

Projecting the vertical evolution onto the specific 2026 clinical decision tree, below are six key branchpoints and the evidence for each.

3.1 Newly Diagnosed Osteosarcoma / Ewing: NGS Panel Mandatory + 5-Subtype Classification

Subtype classification determines everything downstream in bone tumors. Every newly diagnosed bone tumor requires: (a) central pathology review to confirm histology (osteosarcoma / Ewing / chondrosarcoma / chordoma / GCTB); (b) molecular testing with a comprehensive NGS panel covering at minimum IDH1/2 R132 hotspot (chondrosarcoma), H3F3A G34W/V (GCTB adjunct), EWSR1-FLI1 / EWSR1-ERG fusion (Ewing confirmation), BRCA1/2 + SDH + PDGFR + MSI status (cross-subtype actionable targets). Missing IDH1 = missing the ivosidenib path; missing MSI-H = missing tumor-agnostic IO.

3.2 Osteosarcoma Local + Adjuvant: MAP Remains SoC / No Gain from Intensification

2026 mainstream: resectable non-metastatic high-grade osteosarcoma = neoadjuvant MAP (high-dose methotrexate + doxorubicin + cisplatin) → limb-sparing surgery → adjuvant MAP, maintaining the framework established by MIOS 1986. POG-8651 [PMID 12697883] showed no survival difference, yet neoadjuvant remains SoC because of surgical feasibility for limb-sparing + histologic-response stratification (not a survival advantage).

Key branchpoints:

Not recommended: adding ifosfamide or mTOR or any “intensification” strategy on top of the MAP backbone; neoadjuvant SBRT / proton-boosted regimens have no phase 3 positive evidence in frontline osteosarcoma.

3.3 Ewing Sarcoma: Interval Compression vs High-Dose Alkylator + rEECur New Standard

2026 mainstream:

Key controversy: North American dose-dense and European VIDE + BuMel have no head-to-head RCT; in practice, choose by patient location / center experience.

3.4 Rare-Subtype Target Matching: IDH1 Chondrosarcoma / RANKL GCTB / PDGFR Chordoma — Three Paths

2026 mainstream:

3.5 IO Use Restrictions: IO Monotherapy Not Recommended in Osteosarcoma + Ewing + Classic Chondrosarcoma

2026 mainstream:

• Osteosarcoma + Ewing sarcoma + classic chondrosarcoma → Not recommended any IO monotherapy (SARC028 [PMID 28988646] three subtypes ORR 0-5%)
• Dedifferentiated chondrosarcoma → SARC028 + Alliance A091401 expansion [PMID 39343511] signal supports consideration of nivolumab + ipilimumab; still not practice-changing
• Chordoma → Migliorini [PMID 28919999] 3-case positive signal; compassionate use or prospective trial; not casual off-label use
• Tumor-agnostic MSI-H / dMMR / TMB-H (rare in bone tumors, < 1-2%) → pembrolizumab / nivolumab based on cross-tumor basket data
• Alliance A091401 [PMID 29370992] supports nivo + ipi combination > monotherapy in soft tissue, but did not deliver significant improvement in the osteosarcoma subgroup

Key principle: the low-TMB + immune-desert biology of bone tumors means IO is not a “haven’t-found-the-right-drug problem” but a “biology doesn’t support it” problem. Off-label use requires careful assessment of insurance coverage + toxicity + futility risk.

3.6 Relapse Setting: All Existing Regimens Are Marginal + Clinical Trial Priority

2026 mainstream:

Universal principle: all existing regimens for relapsed bone tumors deliver marginal PFS benefit — none cure. Every newly diagnosed relapsed osteosarcoma / Ewing / chondrosarcoma patient should be evaluated first for clinical trial enrollment — ADC (antibody-drug conjugate, e.g., TROP-2), adoptive cell therapy (TIL / CAR-T), novel targets (anti-TIGIT / anti-LAG3) are the main directions.

4. Research Gaps: Ten Unresolved Clinical Questions

This report identifies the following gaps, all definable specific questions (not “more research needed” boilerplate):

1. Biological explanation for 40-year MAP / VDC-IE unsurpassability in osteosarcoma + Ewing: after EURAMOS-1 both arms + COSS-86 + INT-0091, every intensification strategy has failed — is MAP / VDC-IE already at the “chemo ceiling,” or do osteosarcoma / Ewing have unidentified resistance pathways? The molecular basis of intrinsic chemoresistance has not been systematically characterized.
2. Biomarkers for IO failure (TMB / PD-L1 / MHC): in SARC028 [PMID 28988646] bone subgroup ORR 5% — is this uniformly low or does it hide responders? TMB / PD-L1 / HLA loss / neoantigen prediction models lack prospective validation; the mechanism of the 20% responder rate in dedifferentiated chondrosarcoma is unclear.
3. IDH1-selective inhibition vs pan-inhibitor in chondrosarcoma: ivosidenib (IDH1-only, AG-120) in CHONDRO-P1 + LT — ORR 0% + DCR 52%; vorasidenib (IDH1/2 dual inhibitor, AG-881, glioma indication) has no prospective chondrosarcoma data — do selective vs dual-inhibition strategies have different activity in bone tumors?
4. GCTB denosumab post-discontinuation relapse risk: denosumab-GCTB-LT [PMID 31704134] confirmed long-term control, but the rebound risk after discontinuation and the biology of relapse (RANKL-signaling restoration vs de novo osteoclast generation) is not systematically characterized; the optimal use-and-pause intermittent rhythm has no RCT.
5. Chordoma PDGFR monotherapy vs combination + integration with proton radiotherapy: imatinib [PMID 22331945] → imatinib + everolimus [PMID 30216418] combination delivers limited increment; optimal sequencing / concurrent use with proton / heavy-ion radiotherapy has no prospective data.
6. Role of neoadjuvant SBRT / proton in rare subtypes: proton / carbon-ion therapy is standard for local control in chondrosarcoma / chordoma, but its integration with systemic therapy (ivosidenib / denosumab / imatinib) — concurrent / sandwich / sequential — lacks RCTs.
7. Pediatric vs adult osteosarcoma biology differences: pediatric / AYA osteosarcoma MAP cure rate ~60%, while adult (> 40 years) cure rate is significantly lower — is this “cannot tolerate full-dose MAP” or “adult osteosarcoma is a different subtype”? Genomic classification has yet to answer.
8. Genomic / transcriptomic subtype stratification: osteosarcoma is highly heterogeneous (MYC / CDK4 amplification, TP53 complex rearrangements, gene-signature subtypes), yet classification has not yet driven treatment selection; STAG2 / TP53 / CDKN2A co-mutation subtypes in Ewing have not entered stratification.
9. High-dose ifosfamide cardiotoxicity / renal-toxicity sequelae in relapsed Ewing: rEECur established high-dose ifosfamide, but long-term safety of high cumulative ifosfamide dosing in pediatric / AYA — long-term renal / cardiac function / secondary tumor risk — requires registry-based long-term follow-up.
10. Early phase 1 signals of ADC / CAR-T / TIL in bone tumors: TROP-2 ADC, B7-H3 CAR-T, GD2 CAR-T, TIL — all have anecdotal early activity in osteosarcoma / Ewing phase 1; but due to rare-case + pediatric-trial complexity, phase 2 confirmatory progress is slow.

5. 2024-2026 Latest Developments

5.1 FDA / NMPA New Approvals / Guideline Expansion

• ivosidenib bone tumor extension: FDA first approved ivosidenib in 2021 for IDH1+ AML and BTC; NCCN Bone V2.2026 formally lists ivosidenib as a recommended option for IDH1+ conventional chondrosarcoma — based on CHONDRO-P1 [PMID 32208957] + CHONDRO-LT [PMID 40100120] two readouts. The chondrosarcoma indication is not a standalone indication on the FDA label, but NCCN-guideline-level recommendation is in place.
• denosumab GCTB update: FDA approved denosumab for GCTB in 2013 (based on Chawla INTERIM [PMID 23867211]); after 2019 LT [PMID 31704134] data, prescribing information was updated for long-term safety (ONJ ~5%, atypical femur fracture monitoring).
• No new osteosarcoma approvals: no osteosarcoma / Ewing systemic therapy received new FDA approval in 2024-2026. Denosumab’s AOST1321 [PMID 41159913] 2026 negative readout in osteosarcoma explicitly refuted extending denosumab to osteosarcoma as antitumor therapy.
• China NMPA: domestic TKI (anlotinib) is outside NCCN but used in Chinese practice off-label for relapsed soft tissue sarcoma + osteosarcoma; no large-scale phase 3 osteosarcoma data.

5.2 Key Conference Readouts (2024-2025, Weighted Lower)

The following entries serve as candidate pool only before formal peer review; they do not enter the main library. Those with a PMID have been promoted to the main library.

• rEECur phase 3 winner (McCabe 2024 ESMO / 2024 Lancet Oncol manuscript pending): high-dose ifosfamide won as the new reference standard for relapsed Ewing sarcoma; full manuscript still not indexed on PubMed as of 2026-04. Cited by NCT03416517 + conference readout.
• AOST1321 2026 readout: [PMID 41159913] fresh publication (Janeway 2026 Clin Cancer Res); denosumab as antitumor therapy in osteosarcoma negative in both cohorts — explicitly refutes the “hijacking osteoclast biology” hypothesis in osteosarcoma.
• IVOSIDENIB-CHONDRO-LT 2025 readout: [PMID 40100120] long-term follow-up confirming durability of CHONDRO-P1 PFS + biomarker signal; included as NCCN Bone V2.2026 reference.
• Alliance A091401 Expansion 2024: [PMID 39343511] histology-enriched expansion cohorts reconfirming dedifferentiated chondrosarcoma as the bone subtype most likely to respond to IO, with osteosarcoma still inactive.

5.3 Ongoing Phase III / Early Signals (Selected)

• Adoptive cell therapy (TIL / CAR-T) in osteosarcoma: B7-H3 CAR-T, GD2 CAR-T have anecdotal response signals in pediatric / AYA osteosarcoma / Ewing phase 1; Stanford / MSK / COG multicenter phase 2 accrual ongoing.
• TROP-2 ADC (sacituzumab govitecan / datopotamab deruxtecan) early signals in osteosarcoma / Ewing basket trials; no phase 3 yet.
• SARC032 (pembrolizumab + radiotherapy) primarily designed for soft tissue sarcoma — bone-related subgroup data to watch.
• Anti-TIGIT / anti-LAG3 combination IO has no positive signals in bone tumors to date.

6. Synthesis and Judgment

6.1 Vertical × Horizontal: The 2026 Bone Tumor Landscape Is Shaped by Three “Dual-Track Resonances”

Overlaying vertical paradigm evolution on the horizontal current decision landscape reveals that the 2026 bone tumor landscape exhibits “dual-track resonances” completely different from NSCLC / BTC:

1. Mainstream major subtypes (osteosarcoma + Ewing) 40-year stagnation vs rare subtypes (chondrosarcoma + GCTB + chordoma) precision breakthroughs in parallel: osteosarcoma MIOS 1986 MAP backbone unchanged for 40 years, all intensification failed (EURAMOS-1 both arms / COSS-86 / MTP-PE contested / all relapse TKIs marginal); meanwhile in the same period, chondrosarcoma IDH1 → ivosidenib, GCTB → denosumab, chordoma → imatinib all delivered phase 2-level positive signals. “Biology determines outcomes” — mainstream major subtypes lack clear druggable drivers, while rare subtypes each have an independent driver. This is the “anti-scale phenomenon” unique to bone tumors (the commonest histologies have the most resources yet are hardest to break through).
2. IO cross-subtype all-failure + three narrow windows left open: SARC028 osteosarcoma / Ewing / classic chondrosarcoma ORR 0-5% confirms IO is not a cure-all; but dedifferentiated chondrosarcoma (SARC028 + Alliance expansion) + chordoma (Migliorini + PD-L1 expression evidence) + MSI-H (rare but tumor-agnostic effective) three narrow windows remain. “Biology gate-keeps” rather than “haven’t tried the right drug” — low TMB + immune-desert keep the IO monotherapy ceiling very low in osteosarcoma / Ewing.
3. Trial-design innovation vs new-drug approvals ratio inversion: in bone tumors, the most valuable frontline improvements in the past 25 years came from trial design, not new drugs — AEWS0031 compressed VDC/IE q3w to q2w grabbing 8 percentage points of EFS; EURO-EWING-99-R2 used high-risk-stratified BuMel HSCT to grab 14 percentage points; rEECur used Bayesian adaptive design to give relapsed Ewing its first RCT answer. This “design beats new drug” phenomenon is almost unseen in NSCLC (driver era) / BTC (biomarker era).

These three resonances together explain a clinical phenomenon: for a newly diagnosed bone tumor patient in 2026, the core branchpoint in the decision tree is not “which newest drug to pick” but “first confirm the subtype + driver diagnosis → follow mainstream chemo backbone or rare-subtype precision → IO mostly hands-off”.

6.2 Clinical Decision Takeaways (for Junior-Mid Oncologists)

1. “Confirm subtype first, then decide” is an iron rule: five bone tumor subtypes (osteosarcoma / Ewing / chondrosarcoma / chordoma / GCTB) have completely different treatments. Any non-specialist center receiving a case should immediately arrange central pathology review + an NGS panel covering at minimum IDH1/2 / EWSR1 fusion / H3F3A / SDH / PDGFR / MSI. Missing IDH1 = missing the chondrosarcoma ivosidenib path; missing H3F3A = missing GCTB confirmation.
2. Do not add drugs to the osteosarcoma MAP backbone casually: the MIOS [PMID 3520317] 1986 framework has been undefeated for 40 years. EURAMOS-1 both arms (poor responder + good responder) negative [PMID 27569442 + 26033801] explicitly tell us: adding ifosfamide / interferon on top of MAP is useless — only more toxicity.
3. Dose-dense is a free win in Ewing: AEWS0031 [PMID 23091096] compressed q3w to q2w grabbing 8 percentage points of EFS — no new drug, no extra toxicity. This is the single most important frontline decision in Ewing sarcoma.
4. High-risk localized Ewing goes BuMel: EURO-EWING-99-R2 [PMID 30188789] high-risk subgroup BuMel + autologous HSCT delivers 3-year EFS 67% vs 53%, a rare positive modern osteosarcoma / Ewing intensification RCT — prerequisite is correctly identifying “high risk” (poor induction response or large axial tumor).
5. IDH1 testing mandatory in chondrosarcoma: IDH1+ chondrosarcoma goes to ivosidenib (IVOSIDENIB-CHONDRO-P1 + LT [PMID 32208957 + 40100120]), with olaparib BRCAness second shot [PMID 34994649] after failure. ivosidenib is not vorasidenib — chondrosarcoma uses AG-120, glioma uses AG-881 — do not confuse them when prescribing.
6. GCTB = denosumab is SoC: Thomas 2010 [PMID 20149736] + Chawla 2013 [PMID 23867211] + 2019 LT [PMID 31704134] three progressive phase 2 datasets lock it in — GCTB changed from “unresectable = fatal” to “medical therapy → limb-sparing surgery.” Note long-term ONJ ~5% + atypical femur fracture monitoring + use-and-pause intermittent dosing.
7. Chordoma = multidisciplinary + proton + imatinib: chordoma prefers proton / heavy-ion radiotherapy + surgery; when systemic therapy is exhausted, imatinib [PMID 22331945] serves as backbone, with everolimus combination [PMID 30216418] or brachyury vaccine clinical trial after failure.
8. IO is not a cure-all: osteosarcoma + Ewing + classic chondrosarcoma (non-dedifferentiated) — not recommended to empirically use IO monotherapy (SARC028 [PMID 28988646]). Off-label use combines triple negatives: extremely low clinical benefit + unexpected toxicity + insurance-denial risk.
9. Rational expectations for relapse 2L: relapsed osteosarcoma regorafenib / cabozantinib / pazopanib — all VEGFR TKIs — offer 4-6 month PFS marginal wins, none curative; relapsed Ewing high-dose ifosfamide was just established as the new standard by rEECur; relapsed chondrosarcoma goes by IDH classification or anti-angiogenic TKI. Clinical trial enrollment (ADC / CAR-T / TIL / novel targets) is the benefit-window opportunity.
10. Lesson from AOST1321 denosumab osteosarcoma negative: the same drug (denosumab) is landscape-changing in GCTB and inactive in osteosarcoma [PMID 41159913]. Five-subtype bone tumor heterogeneity is far greater than one imagines — do not casually extrapolate one subtype’s success to another.

7. Information Sources

Metadata for the 35 trials in this report was independently verified through PubMed and ClinicalTrials.gov. Every [PMID xxxxxxxx] in the text can be verified directly in PubMed.

• Published trials: 34, covering 1986-2026 (PMIDs verifiable)
• Ongoing / primary manuscript pending: 1 (rEECur phase 3 McCabe 2024 ESMO readout, NCT03416517, primary publication not yet PubMed-indexed)
• NCCN guideline citations: 27/35 (77%) directly hit NCCN Bone Cancer V2.2026 reference section; the remaining 8 are expected misses within NCCN literature-scope limits (Alliance A091401 IO series, Migliorini chordoma compassionate 3 cases, AOST1321 just-published in 2026, MIOS 1986 historical, etc.)
• Key 2024-2026 new data: 3 (AOST1321 [PMID 41159913], IVOSIDENIB-CHONDRO-LT [PMID 40100120], Alliance-A091401-Expansion [PMID 39343511])
• Research gaps: 10

7.1 In-Text Citation List (by PMID Ascending)

The following table is the bracket-cited PMID list in this report text — each can be clicked through to PubMed URL for verification.

(The rEECur phase 3 primary manuscript is cited by NCT03416517 + McCabe 2024 ESMO readout; PMID will be added to the table once indexed.)

7.2 Verification Conventions

• Every PMID can be accessed directly via https://pubmed.ncbi.nlm.nih.gov/{PMID}/ for verification
• Every NCT id can be accessed via https://clinicaltrials.gov/study/{NCT_id}/
• Conference abstracts (ASCO / ESMO / CTOS) are searched through official conference systems; all conference citations in this report are “weighted lower” — not peer-reviewed, final data defers to journal publication
• After rEECur manuscript publication, the corresponding PMID will be updated
• If you find a discrepancy between a PMID’s trial name / year / conclusion in this report and PubMed, corrections are welcome

Clinical Trial Timeline Is Here

Chinese: /trials/bone/ English: /en/trials/bone/

Each trial has an independent detail page, including:

• Complete intervention / comparator regimen
• Primary endpoint values + 95% CI
• Key findings + clinical significance
• Clickable links to PMID / NCT originals

35 trials · 5 subtypes · 1986 to 2026 · synced with NCCN Bone Cancer V2.2026.

Closing

Bone tumors over the past 40 years are the oncology field with the strongest “dual-track reality” flavor — the mainstream major subtypes (osteosarcoma + Ewing) held the MAP / VDC-IE backbone undefeated for 40 years, with all intensification failed; in the same period, the rare subtypes (chondrosarcoma + GCTB + chordoma) each leveraged an independent driver gene to achieve precision-therapy breakthroughs; IO at the cross-subtype level failed almost completely, leaving three narrow windows (dedifferentiated chondrosarcoma, chordoma, rare MSI-H subgroups) open.

This landscape of “mainstream stagnation + rare breakthroughs + IO failure” forms a sharp contrast with NSCLC’s “5-paradigm leaps + 10 drivers all entering 1L + IO rewriting the backbone.” The driver behind it is not resource investment (osteosarcoma / Ewing have far more global decades of RCT resources than chordoma), but biology — mainstream major subtypes lack clear druggable drivers, while rare subtypes each have independent drivers; low TMB + immune-desert in bone tumors keep the IO ceiling low. This is a domain where “biology determines the treatment ceiling,” not a “haven’t-found-the-drug yet” problem.

For a newly diagnosed bone tumor patient in 2026, the core branchpoints in the decision tree are not “which newest drug to pick” but “confirm subtype first → follow mainstream chemo backbone or rare-subtype precision → IO mostly hands-off → relapse prioritizes clinical trial.”

The value of this report lies not in “exhaustively listing all trials” (PubMed can do that), but in compressing 40 years of evolution + current decisions + unresolved gaps into a single reading bandwidth. Next time you face a newly diagnosed bone tumor patient, every branchpoint in the decision tree has this map to consult — checkable, traceable, debatable.

Clinician × AI = Research Superpower + Clinical Decision Amplifier

—— Dual Brain Lab · 2026-04-21