CD33

Chr 19

CD33 molecule

Also known as: CD33rSiglec, SIGLEC-3, SIGLEC3, p67

NCBI Gene: 945 ENSG00000105383 UniProt: P20138

Enables protein phosphatase binding activity; protein tyrosine phosphatase activator activity; and sialic acid binding activity. Involved in several processes, including immune response-regulating signaling pathway; negative regulation of cytokine production; and negative regulation of monocyte activation. Located in Golgi apparatus; external side of plasma membrane; and peroxisome. [provided by Alliance of Genome Resources, Jul 2025]

Active trials

Pathogenic / LP

ClinVar variants

Pubs (1 yr)

-0.1

Missense Z

1.32

LOEUF

Clinical Summary— CD33

⚡

Population Constraint (gnomAD)

Low constraint (pLI 0.00) — loss-of-function variants are relatively tolerated in the population.

📋

ClinVar Variants

14 Pathogenic / Likely Pathogenic· 39 VUS of 83 total submissions

💊

Clinical Trials

7 active or recruiting trials — potential therapeutic options may be available

Population Genetics & Constraint

gnomAD v4 — loss-of-function & missense intolerance

gnomAD

Tolerant — LoF & missense variants common in population

LoF Constraint?LOEUF (Loss-of-function Observed/Expected Upper bound Fraction) is the upper bound of the 90% CI for LoF OE — the preferred gnomAD v4 metric. Lower = more intolerant to LoF. LOEUF < 0.35 = highly constrained.

1.32LOEUF

pLI 0.000

Z-score 0.70

OE 0.80 (0.50–1.32)

Tolerant

Highly tolerant — LoF variants common in population

Missense Constraint?Missense Z-score: standard deviations fewer missense variants observed vs. expected. Z > 3.09 (p < 0.001) = gene does not tolerate missense variation. OE missense < 0.6 is also considered constrained.

-0.14Z-score

OE missense 1.03 (0.92–1.15)

211 obs / 205.3 exp

Tolerant

Tolerant to missense variation

Observed / Expected Ratios?Shaded band = 90% confidence interval. Vertical tick = point estimate. Grey threshold line = gnomAD constraint cutoff for that variant class.

LoF OE?Ratio of observed to expected LoF variants. Upper CI bound (LOEUF) ≤ 0.35 = strong LoF constraint signal.0.80 (0.50–1.32)

0≤0.351.4

Missense OE?Ratio of observed to expected missense variants. OE ≤ 0.6 = fewer missense variants than expected by chance.1.03 (0.92–1.15)

0≤0.61.4

Synonymous OE?Control metric — synonymous variants are largely neutral and expected near OE = 1.0. Significant deviation may indicate annotation issues.0.93

0≤1.21.6

LoF obs/exp: 11 / 13.8Missense obs/exp: 211 / 205.3Syn Z: 0.54

0.82top 10%

GOF

0.73top 25%

LOF

0.2385th %ile

This gene has evidence for multiple mechanisms of pathogenicity (dominant-negative and gain-of-function). Both the Badonyi & Marsh prediction and the broader genomic evidence point to dominant-negative as the predominant mechanism. Different variants in this gene may act through different mechanisms — interpret in context of the specific variant.

DNprediction above median

GOFprediction above median

Note: In-silico variant effect predictors (SIFT, PolyPhen, REVEL, CADD) may underestimate pathogenicity of missense variants in genes with GOF or DN mechanisms. Consider functional evidence and clinical context.

Predictions from Badonyi M, Marsh JA. PLoS ONE. 2024;19(8):e0307312.

ClinVar Variant Classifications

83 submitted variants in ClinVar

ClinVar

Classification Summary

Pathogenic13

Likely Pathogenic1

VUS39

Likely Benign11

Benign19

Pathogenic

Likely Pathogenic

VUS

Likely Benign

Benign

Curated Variants Distribution

Classified variants from ClinVar · 5 ACMG categories

Classification	LoF	Missense + Inframe	Non-coding	Synonymous	Total
Pathogenic	0	0	13	0	13
Likely Pathogenic	0	0	1	0	1
VUS	0	36	3	0	39
Likely Benign	1	8	0	2	11
Benign	2	10	5	2	19
Total	3	54	22	4	83

LoF = frameshift, stop gained/lost, canonical splice · Counts from ClinVar esearch · Updated hourly

View in ClinVar →

Protein Context — Lollipop Plot

CD33 · protein map & ClinVar variants

Showing all ClinVar variants across the protein. Search a specific variant to highlight its position.

External Resources

Links to major genomics databases and tools

Franklin by Genoox

AI-powered variant curation and clinical analysis

DECIPHER

Genomic variants and phenotypes in rare disease patients

Clinical Trials

Active and recruiting trials from ClinicalTrials.gov

ClinicalTrials.gov

Relapsed/Refractory AML

Phase I/II Clinical Trial of Universal Donor CD33 CAR Natural Killer Cells for AML

NOT YET RECRUITING

NCT07026942Phase PHASE1, PHASE2Nationwide Children's HospitalStarted 2026-04-01

Universal donor derived CD33 CAR-NK

Leukemia, Myeloid, Acute

CD33KO-HSPC Infusion Followed by CART-33 Infusion(s) for Refractory/Relapsed AML

RECRUITING

NCT05945849Phase PHASE1University of PennsylvaniaStarted 2024-02-23

CD33KO-HSPC; CART33

Acute Myeloid Leukemia

A Study to Compare Standard Chemotherapy to Therapy With CPX-351 and/or Gilteritinib for Patients With Newly Diagnosed AML With or Without FLT3 Mutations

RECRUITING

NCT04293562Phase PHASE3Children's Oncology GroupStarted 2020-07-21

Allogeneic Hematopoietic Stem Cell TransplantationAsparaginase Erwinia chrysanthemiBiospecimen Collection

Relapsed/Refractory Acute Myeloid Leukemia (AML)

Genetic Ablation of CD33 in HSC to Broaden the Therapeutic Index of CD33-directed Immunotherapy in Patients with AML

NOT YET RECRUITING

NCT05662904Phase PHASE1German Cancer Research CenterStarted 2028-01

Donor-derived CD34+ HSC with CRISPR/Cas9-mediated CD33 deletionGemtuzumab Ozogamicin

Recurrent Glioblastoma, IDH-WildtypeRecurrent GliosarcomaRecurrent WHO Grade 4 Glioma

Testing the Combination of Anti-cancer Drugs Actimab-A and Cemiplimab (REGN2810) to Improve Outcomes for Patients With Recurrent Glioblastoma

NOT YET RECRUITING

NCT07422363Phase PHASE1National Cancer Institute (NCI)Started 2026-06-28