Intimal sarcoma, dedifferentiated liposarcoma and their relatives share one amplified stretch of chromosome 12 — and a very low mutation count. That count is usually read as "immunotherapy will not work." The published evidence says something more useful, and more dangerous, than that: some patients respond durably, the test that predicts it is not the one usually run, and one specific mistake can make the tumour grow faster.
Start here
Immunotherapy with checkpoint inhibitors — drugs that block PD-1, PD-L1 or CTLA-4 — works best, in most cancers, when the tumour carries many mutations. Each mutation can create a new protein fragment the immune system has never seen. Tumour mutational burden (TMB) counts them. Melanoma and smoking-related lung cancer have hundreds; that is a large part of why checkpoint inhibitors transformed those diseases.
Sarcomas are the opposite case. Adult soft-tissue sarcomas are among the least mutated of all solid tumours,1 and the MDM2-amplified family sits at the low end even of that: typical values are one to two mutations per megabase, with microsatellite-stable, mismatch-repair-proficient tumours. These cancers are driven not by point mutations but by copy-number amplification — extra copies of MDM2 and usually CDK4 on chromosome 12q13–15. On a TMB report, an MDM2-amplified sarcoma looks like a tumour that immunotherapy should ignore.
And yet the responders exist, and they were found early. In SARC028 — the first prospective trial of pembrolizumab in sarcoma — the objective response rate across soft-tissue sarcomas was 18%, concentrated in two histologies: undifferentiated pleomorphic sarcoma (23% in the expanded cohort) and dedifferentiated liposarcoma (10%), the most common MDM2-amplified sarcoma.2,3 Nobody selected those patients by mutation count. They responded anyway.
The strongest predictor found so far is not TMB and not PD-L1 alone. It is the presence of tertiary lymphoid structures (TLS) — organised clusters of B cells and T cells inside the tumour, visible on an ordinary slide. In a landmark 2020 analysis, sarcomas with this B-cell-rich immune class had the best survival and the highest response to pembrolizumab in SARC028.4 A trial then selected patients on this feature prospectively: in the TLS-positive cohort of PEMBROSARC, the response rate to pembrolizumab plus low-dose cyclophosphamide was 30%, against 2.4% in the unselected cohorts of the same trial.5
The lesson is simple and, for this family of tumours, largely unapplied: the question is not how many mutations the tumour has, but whether the immune system has already found it. That is a question about the tissue, and it is answerable from the block that is already in the pathology archive.
The tissue
Here the honest answer is that the field does not yet know, and the two best datasets on intimal sarcoma point in different directions.
An integrated genomic and immune analysis of five pulmonary artery intimal sarcomas found all five to be MDM2-amplified and all five PD-L1-positive, with an immune infiltrate, upregulated interferon-γ and IL-6/JAK/STAT3 signalling, and M2-type macrophages. Two of the five carried an amplification of the PD-L1 gene itself (CD274).6 That last detail matters more than it sounds: across cancers, CD274 amplification is one of the strongest single predictors of checkpoint response — roughly two-thirds of a small treated series responded.7 The authors' reading was "inflamed but suppressed": the immune system is present and has been switched off, rather than absent.
The largest dataset says something different. A Seoul group profiled 42 intimal sarcoma samples and split them into two molecular subtypes. The copy-number-high subtype — 33 of 42, the typical MDM2/CDK4-amplified tumour — was predominantly immune-desert, with few infiltrating lymphocytes. A small second subtype (5 of 42), defined by MLH1 loss and resembling microsatellite instability, was immune-inflamed; two of those five received pembrolizumab and their tumours shrank.8
In the amplicon's common relative, prospective data exist. A randomised neoadjuvant trial of nivolumab with or without ipilimumab in retroperitoneal dedifferentiated liposarcoma (n=17) found a median pathological response of only 8.8% — against 89% in undifferentiated pleomorphic sarcoma treated with concurrent radiotherapy in the same trial. Within the liposarcoma group, response tracked the immune contexture: fewer regulatory T cells before treatment predicted a major pathological response, and B-cell infiltration after treatment was associated with survival.9 A separate Italian analysis of chemotherapy-treated sarcomas found that CD20-positive B cells were favourable and PD-1-positive infiltrates unfavourable, and that anthracycline chemotherapy itself remodels the infiltrate.10
Five inflamed tumours and thirty-three mostly desert ones are not a contradiction; they are a distribution nobody has measured properly. The likeliest truth is that the MDM2-amplified family is immunologically heterogeneous — most tumours cold, a meaningful minority inflamed, and the inflamed minority is where the responders come from. Which means the single most useful thing a patient can know is which group their own tumour is in. That is measurable today, on archived tissue, and in most patients it has simply never been measured.
The evidence
Below is every published case or cohort we could identify in which a patient with an MDM2-amplified sarcoma responded to immunotherapy, with the setting, the regimen and the biomarker where reported. It is a short table. Read the note beneath it before drawing conclusions.
| Report | Tumour | Regimen | Outcome | Biomarker context |
|---|---|---|---|---|
| Ribeiro 2024 (Toronto)11 | 3 advanced MDM2-amplified intimal sarcomas | Pembrolizumab-based | Partial response in all three | Low TMB, mismatch-repair proficient; PD-L1 expression and TLS proposed as the explanation |
| Wilky 2025 (Colorado, NCT04028063)12 | Phase 2, mixed sarcomas (n=28 evaluable); patient 002 was an intimal sarcoma | Doxorubicin + anti-PD-1 + anti-CTLA-4 | Trial: ORR 33%, disease control 80%, PFS at 6 months 46% (primary endpoint not met). Patient 002: complete response of target lesions, two years on treatment, no active disease at last follow-up | Responders were mostly TMB-low and microsatellite-stable; sequence mattered (see below) |
| Park 2025 (Seoul)8 | 2 of 5 MSI-high-like intimal sarcomas (MLH1-altered subtype) | Pembrolizumab | Tumour shrinkage in both | Inflamed subtype — distinct from the typical MDM2/CDK4-amplified tumour |
| Nigi 2026 (Japan)13 | Pulmonary artery sarcoma, MSI-high, Lynch syndrome | Checkpoint inhibitors after surgery | Disease stability beyond 65 months | High TMB — a biology that does not characterise the typical tumour |
| SARC028 (multicentre)2,3 | Dedifferentiated liposarcoma, 39 patients in the expanded cohort | Pembrolizumab alone | Objective response 10% | Not selected by biomarker |
| Roland 2024 (MD Anderson)9 | Resectable retroperitoneal dedifferentiated liposarcoma, n=17 | Neoadjuvant nivolumab ± ipilimumab | Median pathological response 8.8%; a minority achieved major response | Low regulatory-T-cell density predicted major response |
| Waldschmidt 2025 (Würzburg, NitraSarc)14 | Retroperitoneal dedifferentiated liposarcoma, seventh relapse | Trabectedin priming, then nivolumab | Immediate response; R0 after minor surgery; no therapy needed for 52 months | Sequential priming before checkpoint blockade |
| Abe 2025 (Niigata)15 | Recurrent dedifferentiated liposarcoma, TMB 13/Mb | Pembrolizumab | Partial response, then pathological complete response at surgery; disease-free at 15 months | The rare TMB-high exception; PD-L1 negative, macrophage-rich |
Case reports are published because they are remarkable. For every intimal sarcoma patient whose response to pembrolizumab reached a journal, an unknown number did not respond and were never written up. A list of responders without a denominator is a signal, not a rate. The only entries above with a real denominator are the prospective cohorts — SARC028, Roland, Wilky — and in the MDM2-amplified subgroups they report response in roughly one patient in ten with checkpoint blockade alone.
What the table does establish beyond doubt: low mutation count does not exclude a durable response in this family of tumours. Responders have been documented at TMB of one to two per megabase, microsatellite-stable, in both intimal sarcoma and dedifferentiated liposarcoma. The question is who — and that is what the passport below is for.
One finding from the Colorado trial deserves its own paragraph, because it is easy to miss and may matter more than any single drug. The trial ran in two stages with different timing. When checkpoint antibodies were given three weeks before the first doxorubicin dose, the reported objective response rate was 56%; when they were started concurrently with doxorubicin, it was 7%.12 The investigators' hypothesis is that doxorubicin — and the steroid anti-nausea drugs given with it — suppress the immune priming that checkpoint blockade needs, exactly in the window when it is being attempted. This was a post-hoc observation in a small trial and needs confirmation. But it converges with the Würzburg case (a priming drug, then checkpoint blockade) and with a mechanistic argument made in the next section: in an MDM2-amplified tumour, the order in which things are done may decide whether immunotherapy helps or harms.
Safety
This is the single most important paragraph on this page, and the one least likely to reach a patient in time.
In 2017, a group at UC San Diego analysed 155 patients treated with checkpoint inhibitors and asked which genomic alterations were associated with hyperprogression — the tumour growing markedly faster after immunotherapy started than it had before. The clearest signal in the whole dataset was MDM2 or MDM4 amplification. All six patients with it failed treatment within two months, and four of the six met the formal definition of hyperprogression, with growth rates that more than doubled compared with their pre-treatment scans.16
Six patients is a small number, and the finding has been debated since. But it has never been refuted in MDM2-amplified tumours, the biological rationale is plausible — interferon signalling released by immune activation can drive MDM2 transcription, degrading p53 faster in a cell that already has too much MDM2 — and it converges with what the sarcoma trials show: checkpoint blockade alone, in this family, rarely helps and may accelerate. Every documented durable response in the table above occurred either in a combination, in a sequence, or in the atypical MSI-high subtype.
This page cannot tell you whether to accept. It can tell you what to ask, so that the decision is made with the full picture:
None of this is a reason to refuse immunotherapy. It is a reason to insist that it be given in a sarcoma centre, in a considered combination or sequence, with a biomarker rationale, and ideally in a trial — which is exactly what the responders in the table had.
Practical
Every item below can be performed on the archived tumour block from a biopsy or operation, or on a single blood sample. None requires a new procedure. Together they answer the question that TMB cannot: has the immune system found this tumour, can it see it, and which immunotherapy routes are open or closed? Take this list to your pathologist or oncologist. Most of it is standard immunohistochemistry that any sarcoma centre can order.
The immune contexture of intimal sarcoma has been characterised in fewer than fifty patients worldwide, in two studies that disagree. For MDM2-amplified sarcomas outside the liposarcoma family, it has not been characterised as a series at all. Every passport that is run and shared adds to the first such series. If you complete these tests, we would like to record the results — anonymised, with your explicit consent, and under Swiss and European data protection law. See the bottom of this page.
Experimental
No immunotherapy is approved for any MDM2-amplified sarcoma. What follows is the state of play as of September 2026, in three tiers of maturity. Trial status changes constantly; check the registry link before acting on any entry.
The concept behind the Colorado trial — doxorubicin as an immunogenic partner — is now in a randomised phase 3: doxorubicin with or without pembrolizumab in dedifferentiated liposarcoma and undifferentiated pleomorphic sarcoma (NCT06422806, n=365, 260 sites in North America). It excludes patients who have already received an anthracycline. Its readout, expected around 2032, will be the first level-1 answer on immunotherapy in this family.
The Colorado trial itself continues (NCT04028063, University of Colorado): its second part uses botensilimab, an Fc-enhanced anti-CTLA-4 antibody, allows any number of prior therapies, and runs to 2027.12
Reading: the most mature route. The open question is not whether the combination has activity — it does — but which patients, and in which order.
MDM2 inhibitor + anti-PD-1. Blocking MDM2 restores p53, which increases antigen presentation and may remove the very substrate that drives hyperprogression — the logic of alrizomadlin (APG-115) with toripalimab in TP53-wild-type, MDM2-amplified liposarcoma (NCT04785196, phase 1b/2, China). Results in sarcoma are pending; the MDM2-inhibitor class carries dose-limiting thrombocytopenia.
FGFR inhibitor + anti-PD-1. FRS2, an FGFR-pathway gene, sits on the same amplicon as MDM2 and is frequently co-amplified. PERELI (NCT06389799, Lund/Oslo/Gothenburg/Stockholm) tests pemigatinib with retifanlimab in dedifferentiated liposarcoma on the premise that FGFR inhibition remodels the microenvironment.
CDK4/6 inhibitor as immune partner. CDK4/6 inhibition — the axis with the first positive phase 3 in this family — is itself immunomodulatory in preclinical work: it increases antigen presentation and suppresses regulatory-T-cell proliferation.18 No sarcoma trial combines the two yet.
Reading: mechanistically the most elegant tier, clinically the least proven. Every combination here is a hypothesis with a trial attached.
Tumour-infiltrating lymphocytes (TIL). A phase 2 of lifileucel with explicit dedifferentiated liposarcoma and undifferentiated pleomorphic sarcoma cohorts opened in September 2026 (NCT07741877, n=80). TIL therapy is agnostic to HLA type and to which antigen the T cells recognise — the one cellular route that does not depend on the antigen tests above. It requires measurable, previously treated disease.
Engineered T-cell receptors. Afamitresgene autoleucel (MAGE-A4) is approved for synovial sarcoma only; PRAME-directed products are in phase 1. All are gated on HLA type and antigen expression — the last two items of the passport.
B7-H3 (CD276). Broadly expressed in sarcoma and on tumour blood vessels; antibody-drug conjugates and T-cell engagers are in development, and the first positive phase 3 of a B7-H3 conjugate in a sarcoma (osteosarcoma) was announced by its sponsor in 2026, with data not yet published. HLA-independent. Whether intimal sarcoma expresses B7-H3 has never been reported.
Reading: for a patient whose tumour has lost antigen presentation, these HLA-independent routes — TIL, B7-H3 — may be the only immunological options left. That is another reason to test B2M early.
Trials that enrol on the MDM2/CDK4 amplification rather than on histology — including the CDK4 and MDM2 inhibitor studies — are listed separately on our page Trials by amplification.
The gaps
Five of five in Bologna; a minority of thirty-three in Seoul. The true proportion — and whether it differs between a tumour in a pulmonary artery and one in the retroperitoneum — has never been measured in a series large enough to say.
Ribeiro's three partial responses, the Colorado complete response, the two Seoul responders: are these the same biology? No two reports measured the same markers.
The 2017 signal came from mixed tumour types. No sarcoma series has tested it prospectively, and no sarcoma series has refuted it. Patients are being treated in that gap.
56% versus 7% by timing alone is a post-hoc finding in fewer than thirty patients. If it replicates, it changes how every combination should be given.
The same amplicon in a different organ — with a different immune environment, different blood supply, and possibly a different antigen landscape. Intimal sarcoma may be the natural experiment that explains dedifferentiated liposarcoma, or the reverse. Nobody has looked.
The mechanism is clean — restore p53, restore antigen presentation, remove the hyperprogression substrate, then add checkpoint blockade. It has not been tested in that order in any sarcoma trial.
What you can do
Dedifferentiated liposarcoma has trials, cohorts and research programmes at large centres. The rest of the amplicon family does not. Intimal sarcoma of the pulmonary artery, aorta or heart; MDM2-amplified undifferentiated pleomorphic sarcoma; the rare MDM2-amplified bone sarcomas; and the tumours whose report says only "high-grade sarcoma, MDM2 amplified" — these are treated by extrapolation from liposarcoma, and their immune biology has never been recorded as a series.
If your MDM2-amplified sarcoma arose inside a large blood vessel or in the heart, and your report calls it undifferentiated pleomorphic sarcoma, "high-grade sarcoma" or even liposarcoma, it may in fact be an intimal sarcoma. The distinction changes surgical planning and follow-up. Our page on reading the pathology report explains what to look for and what to ask.
We are building the first immune-contexture series for MDM2-amplified sarcomas outside the liposarcoma family — anonymised, consented, and shared with the researchers who can use it. Patients with dedifferentiated liposarcoma are welcome as a comparison group. Two things move this forward:
If your tumour has been tested for PD-L1, lymphoid structures, mismatch repair or any of the markers above, or if you have been treated with a checkpoint inhibitor, cell therapy or an immune combination — whatever the result — we want to record it. Non-responses are as valuable as responses; they are the denominator this page is missing.
Share your immune dataWe are a non-profit initiative with funding and a mandate to accelerate work on the MDM2/CDK4 amplicon. If you have archived MDM2-amplified cases without immune characterisation, a multiplex panel that could be run on them, or a hypothesis about why the vessel-wall tumour behaves differently — tell us. We would rather fund the right experiment than commission the obvious one.
Propose an experimentSources
Literature identified through PubMed and ClinicalTrials.gov. Every reference links to the original publication via DOI or to the trial registry entry. Where evidence is limited to case reports, this is stated in the text.