Dan S. · Astrocytoma data

Patient-led research data · Nov 2026

IDH-mutant astrocytoma, grade 4: a single-patient multiomic dataset

Dan, 48, is sharing his brain tumor's sequencing, single-cell data, imaging and immune monitoring. His disease is stable, and the goal is to keep it that way and look for a cure.

3independent tumor profiles
75,930single cells sequenced
17brain MRI studies
18vaccine neoantigens
The tumor by single-cell RNA. Each dot is about 50 cells, 60,575 in all; percentages are shares of all cells. The white dots are the T and NK cells: 0.3%.

Purpose

Why these data are shared

The situation

Dan has a grade 4 IDH-mutant astrocytoma. He has had no progression since chemoradiation in 2024, and he is living a full life.

The aim

Keep the disease stable, have well-reasoned options ready if anything changes, and find what in this tumor points toward a cure.

What analysis can change

Findings go to his treating physicians. He is open to single-patient INDs and compassionate use, and will travel for treatment.

The case

Clinical course: no progression on five concurrent therapies

Diagnosis
Astrocytoma, IDH1 R132H, WHO grade 4
MGMT
Methylated
Site
Right frontotemporal, insular
Function
KPS 90, normal exam
Latest MRI
4 Nov 2026, no progression

Bars are continuous treatment; ticks are single doses or sessions; rings are scans. Ivosidenib was held 25 Jul – 12 Aug 2025. Hover any mark for its date.

Dataset

Data inventory

All tumor data come from one pre-treatment specimen, the 28 Dec 2023 resection (UCSF S23-42259). Blood and imaging run through Nov 2026.

DatasetSpecimenContentsFormat
Exome + RNA, CeGaTBrain, block B3Raw reads; 283 passing somatic calls; gene-level TPM; neoepitope reportFASTQ/BAM · CSV · PDF
Exome + RNA, BostonGeneBrain, block B4Raw reads; 32 passing somatic calls; expression with percentiles; copy number; PD-L1 IHCFASTQ/BAM · XLSX · PDF
Single-cell RNA, 10x FlexBrain, FFPECount matrices; Loupe file; per-cell labelsMTX/H5 · CLOUPE · CSV
UCSF 500 and MGMTBrain, block B4Paired tumor-normal panel report; methylation assayPDF
Germline and HLABlood, salivaCeGaT, Ambry 90-gene, BostonGenePDF
ctDNABlood, serialNorthstar Select and Response; Guardant360; Personalis NeXT PersonalPDF
Immune monitoringPBMC, serialPeptide-pool T-cell response reportsPDF
Brain imaging17 MRI studies including FET PET-MRI, with reportsDICOM · PDF
Clinical recordPathology, operative reports, clinical summary, lab historyPDF · CSV

What the 2023 tissue can and can't tell you

The tumor was sampled once, in Dec 2023, before any treatment. Since then it has had radiation (60 Gy), 10 cycles of temozolomide, ivosidenib, tumor treating fields, a peptide vaccine and pembrolizumab. So treat the tissue data as a baseline, not a picture of the tumor today.

Likely still true
  • Germline variants and HLA type
  • Founding mutations: IDH1 R132H, TP53 R273C, ATRX
  • The IDH1 and TP53 vaccine targets
May have changed
  • Cell-state mix, possibly toward MES-like
  • Surface targets such as DLL3, GD2 and SSTR2
  • Immune microenvironment, after vaccine and checkpoint therapy
  • Mutation load: temozolomide can cause hypermutation; not yet tested
Current
  • Blood T-cell responses to the vaccine, through Oct 2026
  • Brain MRI, through Nov 2026

If new tissue is ever taken, the 2023 data become the comparison point for how the tumor changed under treatment. Analyses that plan for that comparison now are especially useful.

Open questions

Unresolved findings where analysis could change decisions

01

Two RNA profiles of the same tumor differ by up to 47-fold

Drug-target genes differ far more than a general glial marker (GFAP) does. Which profile is closer to protein is unknown.

BostonGene, block B4, coding RNACeGaT, block B3, total RNA

TPM, log scale. Sources: Valius target profile (May 2026); CeGaT raw TPM file.

02

Copy-number calls differ by tissue block

The two labs that sequenced block B4 agree. The lab that sequenced B3 calls two extra gains. Regional heterogeneity or method is unresolved.

B3UCSF diagnostic stains; CeGaT exome, RNA, neoantigen selection B4UCSF 500, MGMT assay, BostonGene exome and RNA ?10x single-cell RNA: block not recorded
CallUCSF 500
B4
BostonGene
B4
CeGaT
B3
CCND2 amplification, 12p
EGFR gain, 7p
CCNE1 gain, 19q

Filled = called, ring = not called.

03

Immune-cold tumor, measurable vaccine response in blood

The 2023 tissue has almost no T cells and no PD-L1 protein. Blood T cells now respond to at least 8 of 18 peptides. Whether those T cells reach the tumor is unknown.

0.3%T and NK cells in the tumor (single-cell)
30.6CD4 stimulation index to IDH1 R132H, Jul 2025
0%PD-L1, TPS and CPS (22C3)
8 of 18vaccine peptides with a T-cell response, Jan 2026
04

DLL3 and GD2: RNA-level targets without protein confirmation

Both are tumor-selective in single-cell RNA. Neither has been confirmed at the protein level, and tissue for stains is limited.

05

Pericavity T2 change: treatment effect or infiltrative growth

Non-enhancing T2 signal around the cavity has risen slowly since about Dec 2024. Radiology calls it indeterminate. FET PET showed no uptake.

Genomics

Driver alterations are concordant across three labs

Variant allele fraction by lab

UCSF 500BostonGeneCeGaT

IDH1 p.R132H · TP53 p.R273C with 17p copy-neutral LOH · ATRX c.662+2T>C, hemizygous. BostonGene reports no fraction for ATRX.

Genome-wide

  • TMB 0.7–1.14 mut/Mb
  • Microsatellite stable
  • No HRD
  • MGMT methylated, 10 of 17 CpG
  • CCND2 amp, ~3–4 copies
  • Tumor content 66%
  • Ki-67 ~15%

Not found

  • 1p/19q codeletion
  • CDKN2A/B deletion
  • TERT promoter
  • H3 K27M
  • +7/−10
  • EGFR mutation
  • BRAF/NTRK/RET fusions

Losses (UCSF 500)

  • 3
  • 8p
  • distal 9p
  • distal 10q
  • 11
  • 12q
  • distal 21q

Vaccine neoantigens

18 peptides (10 class I, 8 class II) from 11 mutations. CeGaT, Dec 2024.

  • IDH1 R132H
  • TP53 R273C
  • PPP2R2B V293I
  • RFC1 A980T
  • AGTPBP1 S330T
  • MT-ND4 G146S
  • ALS2 E76D
  • BCL11A R335Q
  • CFAP418 K102E
  • WNK2 P802A
  • NKD1 R288H

HLA and germline

  • A*02:01
  • A*24:02
  • B*18:01
  • B*52:01
  • C*12:02
  • C*12:03
  • DRB1*07:01
  • DRB1*15:02
  • DQB1*02:02
  • DQB1*06:01
  • DPB1*04:01

Germline: LDLR p.Gly219del (familial hypercholesterolemia). 90-gene hereditary cancer panel negative. UGT1A1 *1/*28.

Pathology

Grade 4 on histology, with no molecular high-grade marker

UCSF surgical pathology S23-42259, resection of 28 Dec 2023. Integrated diagnosis: astrocytoma, IDH-mutant, CNS WHO grade 4.

What makes it grade 4

CriterionPresentFinding
Mitotic activityUp to 7 per 2 mm²
Microvascular proliferationPresent
NecrosisNot identified
CDKN2A/B homozygous deletionAbsent (UCSF 500)

Abundant gemistocytic forms are present (pathology report); UCSF 500 links them to the 12p / CCND2 gain.

Immunohistochemistry

IDH1 R132HPositive
p53~70% strong
Ki-67~15%
ATRXLost
PD-L1 (22C3)TPS 0%, CPS 0

IDH1, p53, Ki-67 and ATRX stained by UCSF pathology on block B3. PD-L1 by BostonGene, 31 Mar 2026.

Microenvironment, bulk RNA estimates

NK cells7.7%
Fibroblasts7.4%
Neutrophils7.0%
Macrophages7.0%
Tregs1.0%
CD8 T cells<0.5%

BostonGene, Apr 2026. Cytotoxic genes (GZMB, PRF1, IFNG, CXCL9/10) are at or near zero; HLA class I and TAP1 sit at the 0–1st percentile of astrocytomas.

Tissue remaining

  • FFPE blocks B3, B4: limited
  • Flash-frozen, held at UCSF
  • No living cells
  • No organoids or cell lines
  • DLL3, SSTR2 stains not done

Everything is from before treatment. The amount of frozen tissue is being confirmed.

Single cell

Composition: 62% tumor, microglia-dominated, 0.3% T and NK cells

Tumor cell states

37,706 tumor cells of 60,575 after doublet removal.

NPC-like32.5%
OPC-like26.6%
AC-like25.3%
MES-like9.6%
Dividing6.0%

Surface targets by cell type

GeneTumor-selectiveNote
DLL3Spread across all tumor states
ST8SIA1, B4GALNT1GD2 synthesis; B4GALNT1 mainly NPC/OPC-like
SSTR2Very low, mostly AC/MES-like
IL13RA2Under 1% of tumor cells
ERBB3Also in normal brain cells

Other tumor-selective candidates: CRB1, MYO16, LRP1B, PCDH15, GRM5. Valius analysis, Jun 2026, investigational.

Immune monitoring

Peptide vaccine T-cell responses

3 Apr 2025 · after dose 21 pool

Baseline. One long-peptide pool responds (BCL11A, PPP2R2B, ALS2, MT-ND4).

23 Jul 2025 · dose 7SI 30.6

Strong CD4 response to IDH1 R132H, and to a PPP2R2B / ALS2 / MT-ND4 pool.

19 Jan 2026 · dose 10SI 17.6

IDH1 response persists. CD4 to 7 and CD8 to 2 of 13 pools. Low cell numbers.

20 Oct 2026 · dose 13Pending

First draw after six months of pembrolizumab.

No response to TP53 R273C at any draw. Peptides are tested in pools. Source: CeGaT immune monitoring reports.

Imaging

Seventeen studies, no progression reported

DateStudyRadiology read
4 Nov 2026MRI, 1.5 TNo progression vs Aug. No enhancement, raised perfusion or reduced diffusion
10 Aug 2026FET PET-MRINo FET uptake anywhere in the brain
6 Aug 2026AddendumIndeterminate Pericavity T2 signal up since about Dec 2024
Jun 2024 – May 202613 MRIsNo progression on each
23 Apr 2024MRIResponse to radiation: less infiltrative tumor
26 Jan 2024MRIResidual nodular enhancement at the medial cavity margin

Second primary: papillary thyroid carcinoma

Found on a surveillance brain MRI in 2024. Genetically distinct from the astrocytoma; no clear recurrence.

  • BRAF V600E, 27%
  • pT1b(m)N1b, stage I
  • 10 of 21 nodes positive
  • Thyroidectomy 21 Jan 2025
  • No radioactive iodine
  • Tumor-informed ctDNA not detected

Access

Data access

Raw and processed files are shared on request with academic and non-profit researchers. Tell us who you are and what you plan to do, and we'll reply by email.

Request data access