Skip to content

Fanconi anemia phenotypes and allele-level genotype–phenotype relationships

Scope and framing

Fanconi anemia is a genetically heterogeneous DNA interstrand crosslink–repair disorder with pleiotropic phenotypes spanning congenital anomalies, progressive hematopoietic failure, endocrine/metabolic dysfunction, and markedly elevated cancer risks. citeturn46view0turn28view0turn20view3 The clinical variability is substantial: some individuals have neither obvious congenital anomalies nor early bone marrow failure, and phenotypes differ by gene, pathway position, and residual allelic function (e.g., “null” vs “hypomorphic” alleles). citeturn46view0turn28view0turn42view0

The user request has three parts: (a) a comprehensive phenotype list with estimated frequencies and sources, (b) gene- and allele-level genotype–phenotype information, and (c) a practical framework for predicting genome→phenome relationships. citeturn46view0turn28view0turn23view0turn25view0 “Estimated frequency” is reported here in two main formats: (1) percentage estimates compiled in entity["book_series","GeneReviews","ncbi bookshelf clinical reviews"] from 1927–2024 literature (explicitly approximate), and (2) cohort-derived proportions or cumulative-incidence estimates from major registries/cohorts (International registry and NIH/NCI cohort). citeturn46view0turn20view3turn28view0

Phenotype spectrum with best-available frequency estimates

The table below summarizes the most consistently quantified phenotypes across the literature and larger cohorts. Percentages vary by ascertainment and reporting completeness; in particular, clinic-evaluated cohorts detect more anomalies than questionnaire/remote (“field”) cohorts. citeturn46view0turn30view0turn28view0

Frequency of major phenotypic domains

Phenotype domain Specific phenotype Estimated frequency Source and notes
Congenital / growth Any characteristic congenital anomaly pattern ~75% Literature-synthesis estimate. citeturn46view0
Congenital / growth Born small for gestational age ~40% Birth-size estimate; growth issues can evolve later. citeturn46view0
Growth Growth deficiency 65% Literature-synthesis estimate across reports. citeturn46view0
Skin Abnormal skin pigmentation (e.g., café-au-lait, hypo/hyperpigmentation patterns) 40% Literature-synthesis estimate. citeturn46view0
Limb/skeletal Upper-limb skeletal malformations (classically radial ray/thumb spectrum) 40% Literature-synthesis estimate. citeturn46view0
Limb/skeletal Lower-limb skeletal malformations 5% Literature-synthesis estimate. citeturn46view0
Craniofacial/brain Microcephaly 20–25% Literature-synthesis estimate. citeturn46view0
Renal/urinary Kidney/urinary tract anomalies 20–25% Literature-synthesis estimate; includes horseshoe/ectopic/hypoplastic/absent kidney, hydronephrosis/hydroureter. citeturn46view0
Genital/reproductive Internal/external genital anomalies (male) 25% Literature-synthesis estimate. citeturn46view0
Genital/reproductive Internal genital anomalies (female) 2% Literature-synthesis estimate. citeturn46view0
Endocrine/metabolic Any endocrine disorder 50–75% Literature-synthesis estimate; includes hypothyroidism and glucose dysregulation (see below). citeturn46view0
Ophthalmologic Ocular manifestations 15% Literature-synthesis estimate. citeturn46view0
Otologic Hearing loss 10% Literature-synthesis estimate. citeturn46view0
Neurodevelopment Developmental delay / intellectual disability 10% Literature-synthesis estimate; may be under-described. citeturn46view0
Cardiac Congenital heart defect 6% Literature-synthesis estimate. citeturn46view0
GI Gastrointestinal anomalies 5% Literature-synthesis estimate. citeturn46view0
Hematology Progressive bone marrow failure (BMF) and cytopenias typically present in first decade Common; median onset reported ~7.6y in IFAR analysis BMF onset highly variable; IFAR analysis cites average BMF onset 7.6y. citeturn46view0
Myeloid malignancy MDS and/or AML by age 40 ~35% cumulative incidence Competing-risk analyses across IFAR/NCI/German registry converge on ~35% by age 40. citeturn42view0turn42view2

image_group{"layout":"carousel","aspect_ratio":"1:1","query":["Fanconi anemia radial ray thumb malformation","Fanconi anemia cafe au lait macules skin pigmentation","Fanconi anemia microcephaly short stature clinical photo"],"num_per_query":1}

Cohort-derived frequencies for structured clinical patterns (PHENOS and VACTERL-H)

A structured approach to birth-defect patterns in fanconi anemia uses (a) VACTERL-H features (vertebral, anal, cardiac, tracheo-esophageal fistula, esophageal/duodenal atresia, renal, limb, hydrocephalus) and (b) PHENOS features (Pigmentation, small Head, small Eyes, structural Nervous system, Otologic, Short stature). citeturn28view0turn30view0 In a 203-person cohort study conducted in the entity["organization","National Cancer Institute","bethesda, md, us"] inherited bone marrow failure cohort, VACTERL-H (≥3/8 features) was present in 23% of the “field” cohort and 35% of the clinic-evaluated cohort. citeturn30view0turn28view0 In the same study, the most common VACTERL-H abnormalities were upper-limb (60%) and renal (35–37%), while tracheo-esophageal fistula, hydrocephalus, and anal atresia were each <10%. citeturn30view0turn31view0

For PHENOS (≥4/6 features), the study reported 33% (field) vs 56% (clinic) and noted particularly high frequencies of pigmentation abnormalities (68% field; 82% clinic) and “small eyes” (52% field; 82% clinic). citeturn30view0turn31view0 Importantly for prediction, the same study found that PHENOS status was associated with increased risk of bone marrow failure, and VACTERL-H status was associated with hypothyroidism. citeturn28view0turn30view0

Hematology and cancer outcomes from the largest registry-scale analyses

A 20-year analysis of 754 DEB-confirmed individuals in the entity["organization","International Fanconi Anemia Registry","rockefeller university registry"] (IFAR) reported that 601/754 (80%) developed bone marrow failure and 173/754 (23%) developed 199 neoplasms; 60% of neoplasms were hematologic and 40% were non-hematologic. citeturn20view3 Within those 79 non-hematologic neoplasms, squamous cell carcinoma of head/neck/anogenital/skin accounted for 39 cases, liver tumors 18 cases, renal tumors 6 cases (including Wilms tumor 4), and brain tumors 5 cases (including medulloblastoma 4). citeturn20view3

In the NCI cohort study (203 patients), competing-risk analysis estimated cumulative incidence as the first event of severe bone marrow failure 70%, solid tumors 20%, and leukemia 6.5%. citeturn28view0 The same NCI cohort described head-and-neck and gynecologic cancers as the most common solid tumors and noted further increased risk after hematopoietic cell transplantation. citeturn28view0

Toward a comprehensive phenotype catalog

A practical obstacle in “comprehensive phenotype lists” is that many publications focus on a subset of organ systems, and frequency estimates are often missing or inconsistently reported. citeturn46view0turn20view3turn30view0 As a result, exhaustive enumerations are best built using standardized phenotype ontologies (HPO) plus curated disease–phenotype annotation resources that store both (a) phenotype terms and (b) frequency bins and provenance. citeturn33search6turn33search11turn41search0

Frequency bins for “phenotype-level” annotations

Orphadata/Orphanet frequency bins used in phenotype curation are defined as: “always present” (100%), “very frequent” (99–80%), “frequent” (79–30%), “occasional” (29–5%), and “rare/very rare” (4–1%). citeturn33search0turn33search6turn33search11 These definitions are also consistent with the modern HPO “frequency” vocabulary used in disease annotations (obligate 100%; very frequent 80–99%; frequent 30–79%; occasional 5–29%; very rare 1–4%; excluded 0%). citeturn33search11turn41search0

Symptom-term catalogs as a completeness cross-check

The NIH’s Genetic and Rare Diseases Information Center provides an online symptom catalog for fanconi anemia listing 105 symptom terms across systems and explicitly notes that not all possible symptoms may be listed. citeturn36view0 This kind of catalog is useful as a “completeness sanity check” for rare/less standardized features (e.g., specific heart lesions, urologic variants, ocular findings), but the most defensible numeric frequency estimates for clinical decision-making still generally come from GeneReviews-scale synthesis plus registry cohorts. citeturn46view0turn20view3turn30view0

Genetic architecture that drives phenotypes

Gene distribution and mode of inheritance

Fanconi anemia is caused by pathogenic variants across a set of FA/BRCA pathway genes and can be inherited as autosomal recessive (most genes), X-linked (FANCB), or autosomal dominant (a RAD51-associated form). citeturn42view0turn46view0 In the GeneReviews molecular testing summary, FANCA explains ~60–70% of cases, FANCC ~14%, FANCG ~10%, and multiple other genes each contribute a small fraction (often <1–3%). citeturn46view0 This gene heterogeneity is also reflected in curated mutation resources such as the Fanconi Anemia Mutation Database hosted by entity["organization","The Rockefeller University","new york, ny, us"] and displayed on the entity["organization","Leiden Open Variation Database","lovd platform"] infrastructure, which explicitly lists FA genes spanning the upstream core complex through downstream repair factors (including newer/rarer genes). citeturn25view0turn27view0

Pathway position and phenotype severity

A major (and clinically actionable) organizing principle is pathway position: “upstream/core complex,” “ID complex” (FANCI/FANCD2), and “downstream” genes (including BRCA-related subtypes). citeturn28view0turn25view0 In the NCI cohort analysis, patients with upstream-complex variants tended to have less severe congenital patterns (lacking VACTERL-H and/or PHENOS features), while the ID complex was associated with VACTERL-H. citeturn28view0turn29view0

Variant severity class (null vs hypomorphic)

Across multiple genes, “null” alleles (e.g., nonsense/frameshift/large deletions leading to strong loss of function) tend to drive more severe congenital anomalies, earlier hematologic complications, and sometimes earlier myeloid malignancy risk than alleles with residual function (“hypomorphic”). citeturn42view0turn28view0turn21view0 However, the relationship is not absolute: the NCI cohort study explicitly reports that some gene/phenotype relationships seen in literature collections were not reproduced uniformly, and it distinguishes clinic vs field ascertainment as an additional driver of apparent severity. citeturn28view0turn30view0

Allele-level genotype–phenotype signals with highest clinical relevance

The table below focuses on allele-level signals repeatedly highlighted in cohort analyses, curated clinical resources, and/or mechanistic studies. It is not an exhaustive list of all pathogenic alleles (there are thousands), but it covers variants where a specific allele is linked to a specific phenotype class, onset window, or cancer spectrum more strongly than typical “gene-only” effects. citeturn25view0turn46view0turn23view0

Gene (complementation group) Variant (HGVS; legacy name if common) Molecular effect class Key phenotype / risk signal Evidence source(s)
BRCA2 (FA-D1) c.631+1G>A and c.631+2T>G (often referenced as IVS7 splice variants) Splice-site; severe functional impact Particularly deleterious childhood myeloid malignancy timing: myeloid malignancy arising before age 3 is highlighted in GeneReviews; association with AML in FA-D1 noted with strong odds ratio in BRCA2 FA-D1 cancer-correlation analysis. citeturn42view0turn23view0 citeturn42view0turn23view0
BRCA2 (FA-D1) 886delGT and 6174delT Truncating/frameshift In an FA-D1 correlation analysis, these alleles were associated with brain tumors relative to other BRCA2 alleles (reported as very strong association in that dataset). citeturn23view0 citeturn23view0
FANCA (FA-A) Variants within/involving exons 27–30 region (domain/location-defined “allele class”) Region-specific risk (variant location effect) In the NCI cohort, variants within/involving exons 27–30 were associated with increased solid tumor risk (HR ~6.2) and a cumulative incidence of solid tumors of ~60% vs ~25% by age 40 in those with vs without variants in that region. citeturn30view2turn28view0 citeturn30view2turn28view0
FANCA (FA-A) p.His913Pro, p.Arg951Gln, p.Arg951Trp Missense (hypomorphic candidates in GeneReviews) These specific FANCA variants are cited as associated with slower hematologic disease progression (allele-level signal noted in GeneReviews). citeturn42view0 citeturn42view0
FANCC (FA-C) c.456+4A>T (IVS4+4A>T) Splice-region; creates aberrant splicing products; founder allele in some populations Pathogenicity supported by aberrant splicing and functional defects; present in population databases at low frequency overall yet enriched in some subpopulations; carrier frequency in one population estimated ~1 in 89 in ClinVar summary; illustrates how population-allele frequency and founder effects interact with clinical prevalence. citeturn43view0 citeturn43view0
FANCC (FA-C) Intron 4 (IVS4+4A>T) and exon 14 (R548X, L554P) vs exon 1 (322delG, Q13X) groupings Allele classes used in registry survival modeling In the 754-person IFAR analysis, having at least one IVS4+4A>T or exon 14 (R548X/L554P) mutation was associated with poorer overall survival than having at least one exon 1 mutation without exon 14/IVS4 mutations (allele-class survival signal). citeturn20view3 citeturn20view3
FANCC (FA-C) c.165+1G>T (splice donor; noncanonical TT usage phenomenon) Splice-site with measurable residual correct splicing A mechanistic study showed that, despite a splice-donor mutation predicted to prevent correct splicing, residual correct splicing produced low levels of correctly spliced transcript and “minute” levels of normal pathway protein processing, proposed to contribute to milder clinical manifestations in those patients (a direct residual-function→phenotype link). citeturn44view0 citeturn44view0
FANCB (FA-B; X-linked) Deletions/truncations vs missense (variant-type class) Variant-type severity gradient In a 19-child cohort, missense variants were generally associated with delayed onset of hematologic disease and longer survival compared with whole-gene deletion or truncation variants, consistent with residual function being protective. citeturn21view0 citeturn21view0
FANCL (FA-L) c.1092G>A (p.K364=) causing aberrant splicing c.1021_1092del (p.W341_K364del) Synonymous-at-DNA but splice-disrupting; founder allele A founder allele explained 13 cases from a specific ancestry group; this cohort had high rates of skin pigmentation findings (11/13 = 85%), and all 13 had bone marrow failure recorded at diagnosis with mean onset ~5.3 years (gene/allele-specific phenotype profile). citeturn45view0 citeturn45view0

What “allele-level” signals usually mean in practice

Across fanconi anemia genes, allele-level associations most commonly fall into three practical buckets: (1) severe downstream genes (notably BRCA-related subtypes) where specific alleles associate with a specific early cancer spectrum, (2) regional/domain effects (as seen for FANCA exons 27–30 in the NCI cohort) where variant location predicts cancer risk beyond “gene-only,” and (3) residual-function splice phenomena where noncanonical or partially rescued splicing yields measurable clinical moderation despite a seemingly “severe” variant class. citeturn23view0turn30view2turn44view0

Practical framework to predict genome–phenome relationships in fanconi anemia

Stepwise prediction strategy

A defensible prediction strategy is to treat phenotype prediction as a hierarchical inference problem, where each layer narrows uncertainty.

Start by establishing the “FA diagnosis layer” using chromosomal breakage testing (DEB/MMC) and/or molecular confirmation; GeneReviews explicitly notes that the diagnosis can be established via breakage testing and/or identifying the relevant pathogenic variant configuration (biallelic autosomal recessive; hemizygous X-linked FANCB; heterozygous RAD51 for dominant FA). citeturn42view0turn46view0

Next, assign the gene and pathway layer (upstream vs ID vs downstream). In the NCI cohort, pathway position predicted phenotype structure: upstream variants were least likely to show neurodevelopmental abnormalities and VACTERL-H/PHENOS patterns, while ID-complex variants were associated with VACTERL-H. citeturn28view0turn29view0

Then incorporate variant severity layer (null vs hypomorphic, and splice rescue potential). The NCI cohort operationalized null vs hypomorphic classification rules (frameshift/nonsense/major splice disruption vs missense/in-frame/protein-change splicing outcomes) and found meaningful phenotype associations (e.g., PHENOS predicting higher BMF risk). citeturn28view0turn30view0 Mechanistic splice-rescue work provides the rationale: measurable residual correct splicing can preserve small amounts of functional pathway output and moderate clinical severity. citeturn44view0

Finally, apply an allele-specific layer where evidence exists (e.g., BRCA2 IVS7 splice variants and early AML, FANCA exons 27–30 and solid tumor risk). citeturn42view0turn23view0turn30view2 This is the layer most relevant to the user’s goal of predicting genome–phenome relationships at allele resolution, but it is also the layer with the sparsest and most cohort-dependent evidence. citeturn46view0turn28view0turn23view0

Variant databases and how to use them responsibly for phenotype prediction

For allele-level work, the most central curated resources are:

  • entity["organization","ClinVar","ncbi variant archive"] for interpreted variant assertions, supporting evidence summaries, and (often) population allele frequency snippets and mechanistic notes; for example, FANCC c.456+4A>T includes a structured summary of observed allele counts in population datasets, founder-status discussions, and functional splicing evidence. citeturn43view0
  • entity["organization","gnomAD","population variant database"] for background allele frequencies and subpopulation stratification, which helps distinguish plausible recessive carrier alleles from implausibly common variants for a rare recessive disorder (as illustrated in ClinVar’s synthesis for c.456+4A>T across gnomAD subpopulations). citeturn43view0
  • The Fanconi Anemia Mutation Database (Rockefeller/LOVD implementation) for gene-specific variant registries and case aggregation; it explicitly aims to include pathogenicity, population distribution, and phenotypic associations, and provides gene-by-gene access via LOVD. citeturn25view0turn27view0
  • Practice-facing genotype–phenotype summaries in the entity["organization","Fanconi Cancer Foundation","fanconi.org, us nonprofit"] clinical-care materials, which GeneReviews links to for surveillance framing (e.g., endocrinopathy screening). citeturn46view0turn24search8

A key practical point is that database assertions should not be treated as “phenotype guarantees.” ClinVar entries can include single-case “clinical features” and gestational findings (helpful signals), but robust genotype–phenotype prediction requires triangulating: curated assertion + primary literature + cohort context + functional evidence. citeturn43view0turn28view0turn20view3

Limits, confounders, and what is currently predictable

There are structural reasons why genome→phenome prediction in fanconi anemia is challenging even with complete sequencing.

First, frequency estimates are explicitly approximate in GeneReviews because many reports did not include systematic physical descriptions; thus, even high-level frequencies can shift as structured phenotyping and survivorship patterns change. citeturn46view0 Second, registry cohorts contain ascertainment biases and non-standardized clinical evaluations; the IFAR analysis acknowledges limitations such as selective reporting and variable clinical testing intensity across centers. citeturn19view0 Third, competing risks matter: bone marrow failure, transplantation, and survival changes can alter observed cancer incidences over time, so “risk by age” depends on cohort era and management. citeturn42view2turn28view0turn20view3

Within those limits, several genotype–phenotype relationships are currently the most predictively useful:

  • Downstream BRCA-related FA (especially BRCA2/FA-D1): early-onset, high-penetrance childhood cancer patterns, with particularly deleterious splice variants highlighted in curated resources and allele–cancer correlations reported in focused analyses. citeturn42view0turn23view0
  • Pathway-position effects (upstream vs ID vs downstream): reproducible associations with congenital-pattern clusters (VACTERL-H/PHENOS) in a large, systematically analyzed cohort. citeturn28view0turn29view0turn30view0
  • Region-level effects within common genes (FANCA exons 27–30): an example where variant location predicts solid tumor risk beyond gene identity alone, supporting domain-aware modeling for genome–phenome prediction. citeturn30view2turn28view0
  • Residual-function splice rescue: mechanistic evidence that partial correct splicing can measurably moderate clinical manifestations, motivating allele-specific functional annotation (especially for noncanonical splice-region variants and synonymous variants that affect splicing). citeturn44view0turn45view0