Contents
20 — Chromosome abnormalities Part 3 — Genome instability

20 — Chromosome abnormalities

Before this: Ch 09 · Ch 03 · Ch 14 · Time: ~40 min

Chapters 16–19 dealt with damage measured in base pairs. This one deals with damage measured in megabases — the scale at which the packaging is the .

What you'll be able to do

The core idea

There are only two things you can do to a genome at this scale: change how many copies of a region exist, or change where the regions sit. Each has one consequence.

Copy number is a dosage problem. Transcript output is roughly proportional to gene copy number. Multiply a region by 1.5 and you multiply the output of a few hundred genes by 1.5. Most genes tolerate that. Subunits of fixed-stoichiometry complexes, acting near a threshold, and developmental timers do not — and a few hundred genes shifted at once will contain some of each. The damage is combinatorial, not additive.

Position is a pairing problem. A rearranged that retains all its sequence has normal dosage, so the carrier is usually fine. But must physically align homologous sequence before it can segregate it, and a chromosome whose layout differs from its partner's forces that alignment into a contorted configuration. inside the contorted region produces broken products.

A balanced rearrangement is a statement about the carrier's genome, not about their gametes. It is a refactor: same content, different file layout, everything compiles. The problem arrives at the merge — and meiosis is the merge, run against a partner who has the original layout.


1. The notation, in sixty seconds

serialise in ISCN format: count, , then a list of deviations. It is a diff against 46,XX. p is the short arm, q the long arm, band numbers count outward from the .

StringReading
47,XY,+2147 chromosomes, male, one extra chromosome 21
46,XX,del(5)(p15.2)terminal deletion of 5p from band p15.2
46,XY,inv(9)(p12q13)pericentric — the interval spans the centromere
46,XX,t(9;22)(q34;q11)reciprocal , breakpoints at 9q34 and 22q11
45,XX,der(14;21)(q10;q10)Robertsonian: one derivative replaces a 14 and a 21
47,XY,+21[12]/46,XY[18]mosaic — 12 of 30 counted cells trisomic

A Robertsonian carrier has 45 chromosomes and is genomically balanced. Counting chromosomes is not counting sequence.

2. Ploidy versus aneuploidy

Euploid changes multiply the whole set; aneuploid changes alter individual .

Triploidy (69 chromosomes) occurs in perhaps 1–2% of recognised conceptions and never survives. Its two origins give visibly different outcomes, which is the cleanest demonstration of imprinting you will meet:

OriginHowOutcome
Diandry — two paternal setsusually dispermyOvergrown cystic placenta (partial hydatidiform mole), small fetus
Digyny — two maternal setsfailure to extrude a polar bodyTiny fibrotic placenta, severe growth restriction

Same chromosome count, same genes, opposite : the parental origin of a set carries information beyond its sequence. §10 returns to this.

is the common case. Roughly half of first-trimester losses are chromosomally abnormal — autosomal is the largest class at around 35–40% of karyotyped losses, then polyploidy near 10% and monosomy X near 5–10%. Aneuploidy is the leading known cause of human pregnancy loss.

3. Three failure modes, and how to tell them apart

graph TD A["Bivalent on the MI spindle"] -->|"homologues fail to separate"| MI["<b>MI nondisjunction</b><br/>gamete gets one chromatid<br/>from <i>each</i> homologue"] A -->|"normal disjunction"| B["Sister chromatids on the MII spindle"] B -->|"sisters fail to separate"| MII["<b>MII nondisjunction</b><br/>gamete gets <i>two copies</i><br/>of one homologue"] B -->|"normal"| N["Euploid gamete"] MI --> Z["Trisomic zygote"] MII --> Z N --> Z2["Euploid zygote"] Z2 -->|"mitotic nondisjunction<br/>or anaphase lag"| MOS["<b>Mosaic</b> — two cell lines<br/>in one individual"]

They leave different signatures. Take a mother heterozygous A/B at a marker close enough to the 21 that between them is rare, and a father contributing C:

ErrorWhat the egg carriesZygote at the marker
MI one chromatid of each homologueA B C — three , heterozygous
MII nondisjunctiontwo sister chromatids of one homologueA A C — two alleles, maternally homozygous
Post-zygotic mitotica normal eggA C in some cells, A A C or A C C in others

This is a genotyping assay, not a microscopy assay, and it is why the epidemiology is known. Two caveats. The inference assumes no crossover between marker and centromere — one crossover swaps the signatures, so real studies genotype a panel and reconstruct the recombination pattern. And the "MII" signature is now known to arise substantially from premature separation of sister chromatids during MI, which then segregate independently at MII. Read the two classes as centromere-heterozygous and centromere-homozygous; the mapping to MI and MII is a model laid on top.

For 21 the distribution is lopsided: about 90% of cases are maternal, and roughly 70–80% of those carry the MI signature.

Why MI fails. Chapter 09 established that a is held on the MI spindle by chiasmata plus sister-chromatid cohesion distal to them — the crossover is a physical clamp, not only an information shuffle. That predicts the failure mode: a bivalent with no crossover, or with its only crossover jammed against the centromere, has nothing holding it together and segregates at random. Confirmed — MI-origin trisomy 21 shows markedly reduced or absent recombination on the nondisjoined chromosome, and centromere-homozygous cases show an excess of pericentromeric exchanges, which entangle centromeres rather than clamping arms. Recombination placement, fixed decades before the error, is a risk factor for it.

4. Which autosomal trisomies survive

EponymApprox. live-birth frequencySurvival
21Down~1 in 700–800Median survival now into the sixth decade
18Edwards~1 in 5,000–8,000Majority die in the first year
13Patau~1 in 10,000–20,000Majority die in the first year

The obvious hypothesis — small are survivable — is wrong, and the counterexample is decisive:

ChromosomeLengthProtein-coding genesDensityTrisomy at term?
21~47 Mb~220~4.7 /Mbyes
18~80 Mb~270~3.4 /Mbyes
13~114 Mb~320~2.8 /Mbyes
19~59 Mb~1,400~24 /Mbnever

Chromosome 19 is smaller than 13 and carries four times as many genes. The survivable trisomies are the gene-poorest. Chromosomes 13 and 21 are also acrocentric — their short arms are rDNA repeat arrays already present in high copy number, so trisomy adds no meaningful dosage there.

"Survivable" is still relative: roughly a third of trisomy 21 conceptuses detected at chorionic villus sampling are lost before term. And there is no single "Down syndrome gene" — there are ~220 genes at 1.5×, and the is the sum over the dosage-sensitive subset.

5. Sex chromosome aneuploidy: tolerated, and why

FrequencyTypical presentation
45,X (Turner)~1 in 2,000–2,500 female birthsShort stature, ovarian insufficiency, cardiac and renal anomalies; normal intellect
47,XXY (Klinefelter)~1 in 500–1,000 male birthsTall, testicular failure and infertility; often undiagnosed
47,XXX~1 in 1,000 female birthsFrequently no clinical presentation
47,XYY~1 in 1,000 male birthsFrequently no clinical presentation

An extra X is compatible with an unremarkable life; an extra 19 is not compatible with birth. Two structural facts account for the gap.

buffers X dosage. Cells silence all but one X (Ch 13), and the rule is "keep one, silence the rest" — so 47,XXX silences two. Dosage is corrected for free, by machinery that had to exist anyway.

The Y is nearly empty. A few dozen protein-coding genes, most in multi-copy testis-specific families, against roughly 800 on the X. An extra Y adds almost no dosage-sensitive material, which is why 47,XYY is usually silent.

The buffering is incomplete, and the residue is the . About 15% of genes escape inactivation reliably and another ~10% variably; escapees include the pseudoautosomal regions, active on both X and Y. Escapees with a Y homologue — the pseudoautosomal genes plus a handful of X–Y pairs such as KDM6A/UTY — have a normal dose of two, in both sexes, so 45,X, with a single X, is genuinely haploinsufficient for them; SHOX, in the pseudoautosomal region, accounts for much of the short stature. Escapees without a Y partner are simply expressed more highly in XX than in XY cells, and that residual excess is what surfaces as the mild features of 47,XXX and 47,XXY.

45,X is also the sharpest lesson in ascertainment: about 99% of 45,X conceptuses miscarry. The live-born population is a selected 1%, much of it cryptically mosaic. Statements about "the 45,X phenotype" are statements about survivors.

6. The maternal age effect, derived

Maternal ageApprox. live-birth risk of 21
20~1 in 1,500
25~1 in 1,300
30~1 in 900
35~1 in 350
40~1 in 100
45~1 in 25–30

(Model-dependent past 40; these are live-birth rates, and rates at conception are higher throughout because of selective loss.)

Two facts generate the whole curve.

  1. Human oocytes enter prophase I in fetal life and arrest there, resuming only at ovulation. An egg ovulated at 43 has been sitting mid-division for four decades.
  2. Sister-chromatid cohesion is provided by loaded before that arrest, with no measurable replenishment afterwards.

Cohesin in an oocyte is therefore a decaying resource with no maintenance process — a cache populated once, never invalidated, never refreshed. The consequences chain:

Three predictions, all borne out: the effect should be maternal-specific (spermatogonia divide continuously and reload cohesin every cycle — and indeed the is on point , not , Ch 16); it should accelerate; and it should appear in oocytes and embryos long before live births. Preimplantation testing puts embryo aneuploidy near 20–30% in the early thirties and above 50% past about 40.

7. Mosaicism

Post-zygotic errors produce two or more cell lines in one individual, by three routes: mitotic (one trisomic and one monosomic daughter), anaphase lag (a fails to reach a pole and is lost), and rescue (a trisomic zygote drops one of the three — which rescues the embryo and creates the hazard in §10).

8. Structural rearrangements and their meiotic behaviour

TypeWhat changedCarrierMeiotic hazard
Deletion / duplicationsegment lost or gainedunbalanced—
Paracentric segment flipped, outsidebalanceddicentric + acentric products
Pericentric inversionsegment flipped, centromere insidebalancedduplication/deficiency products
Reciprocal two swap endsbalancedunbalanced segregants
two acrocentrics fuse at the centromerebalanced (45 chromosomes)unbalanced segregants
Isochromosomeone arm duplicated, the other lostunbalanced—
Ringboth ends lost, chromosome circularisesunbalanced, unstablebreaks each mitosis →

Inversions

Pairing is matching. To align an inverted segment with its normal partner, one of them must loop — the configuration is exactly the shape an inversion makes on a dot plot:

normal      A  B  C  D  E  F  G
inverted    A  B  F  E  D  C  G      (C–F inverted)

pairing configuration — the inversion loop:

            A   B          G
             \ /  C-D-E-F \ /
              X   | | | |  X
             / \  C-D-E-F / \
            A   B          G

            each rung pairs a marker with its own homologue; the lower
            strand is the inverted chromosome, read F→C along its length

outside the loop is harmless. Inside it, the consequence depends on the centromere.

Paracentric — centromere outside the loop. A crossover inside yields one chromatid with two centromeres and one with none. The dicentric is pulled both ways and breaks; the acentric goes nowhere and is lost. Both products self-destruct before they can become a pregnancy, so paracentric carriers have near-normal reproductive outcomes.

Pericentric — centromere inside the loop. The crossover products each have one centromere and look viable, but each carries a duplication of one flanking region and a deletion of the other. These do form conceptuses. Risk rises with the size of the inverted segment: a bigger loop captures more crossovers, and it leaves smaller — therefore more survivable — flanking imbalances. Empirically the risk of an unbalanced liveborn is negligible below about 30% of chromosome length, small at 30–50%, and highest above 50%.

Because loop crossovers destroy their products, the observed recombination rate across an inversion collapses to near zero. Inversions are recombination suppressors, which is why they persist as long (Ch 29).

Reciprocal translocations

Four chromosomes pairing pairwise by homology form a cross-shaped quadrivalent, which can segregate several ways:

Which travel togetherGametes
Alternatethe two normals; the two derivativesbalanced (normal or carrier)
Adjacent-1one normal + one non-homologous derivativeunbalanced
Adjacent-2homologous centromeres to the same pole (rare)unbalanced
3:1 (tertiary or interchange )three chromosomes to one pole, one to the otherunbalanced; the route to supernumerary-derivative syndromes such as Emanuel syndrome from t(11;22)(q23;q11)

Only alternate segregation gives balanced gametes, so the unbalanced fraction is high — and gamete studies confirm it. Live-birth risk is much lower because unbalanced conceptuses are mostly lost early. The reproductive is usually recurrent miscarriage rather than an affected child, and roughly 2–5% of couples with recurrent pregnancy loss have a balanced rearrangement in one partner. That is the strongest indication for karyotyping an adult.

Robertsonian translocations

The five acrocentrics — 13, 14, 15, 21, 22 — carry near-identical rDNA arrays on their short arms, giving non-allelic (Ch 18) a substrate for fusing two of them at the centromere. The short arms are lost; nothing of consequence goes with them. Carriers have 45 chromosomes and are phenotypically normal. Robertsonians occur in roughly 1 in 1,000 people, t(13;14) accounting for about three-quarters. The worked example segregates one in full.

When "balanced" isn't harmless

Roughly 6% of de novo apparently balanced translocations carry a phenotype, from three causes: a breakpoint landing inside a gene (a acting as a loss-of-function ); position effect, where the gene is intact but severed from its (Ch 50); and cryptic imbalance at the breakpoint, where the event was never balanced and the simply could not see the missing 200 kb. Inherited balanced rearrangements carry a phenotype far less often, for a plain ascertainment reason: a parent well enough to reproduce transmitted them.

9. Copy number variation is normal

The clinical framing above will mislead you if left alone. Structural variation is a normal feature of human genomes. Roughly 4.8–9.5% of the genome is copy-number variable among healthy people, around 100 genes can be homozygously deleted with no apparent consequence, and CNVs account for more differing base pairs between two people than all their SNVs combined. The is a sample, not a schema (Ch 45): a deletion relative to GRCh38 may be the majority . Pathogenic CNVs are the tail of that distribution — enriched exactly where you would predict, over dosage-sensitive genes and across intervals large enough to catch several at once.

Genomic disorders and NAHR

Some CNVs recur at identical breakpoints in unrelated people. That is non-allelic between — near-identical blocks tens to hundreds of kb long, about 5% of GRCh38 and closer to 7% of the complete T2T assembly. When the machinery matches a segdup to its paralogue instead of its allele, the crossover deletes or everything between them.

That predicts deletions and duplications as a reciprocal pair at comparable rates. They are:

RegionDeletionDuplication
17p12, ~1.4 Mb, contains PMP22HNPP — episodic pressure palsies, mildCMT1A — the commonest inherited neuropathy
7q11.23, ~1.5–1.8 Mb, ~26–28 genes incl. ELNWilliams–Beuren syndrome7q11.23 duplication syndrome
22q11.2, ~3 Mb between LCR22A and LCR22D22q11.2 deletion syndrome, ~1 in 4,000–6,00022q11.2 duplication, milder, often inherited

PMP22 is the cleanest dosage experiment in human genetics: one copy gives a mild neuropathy, two is normal, three gives a severe one — same protein sequence throughout. Note also the asymmetry down the table. Deletions are usually — not always — worse than the reciprocal duplication: 7q11.23 and 22q11.2 both run that way, while 17p12 runs the other, because PMP22 overexpression is the more damaging direction. That general asymmetry is why CNV interpretation weights deletions more heavily (Ch 55).

Chromothripsis

The default model of complex rearrangement is progressive: one break, then another, over many divisions. violates it — tens to hundreds of breakpoints on one or a few, generated in a single catastrophic event and re-ligated in random order and orientation by non-homologous end joining (Ch 17). The mechanism is a micronucleus: a chromosome that lags at anaphase is packaged in its own defective envelope, where the DNA is pulverised, and the fragments are stitched back when it rejoins the main nucleus. causes chromothripsis and chromothripsis causes aneuploidy.

The evidence is statistical, and the reasoning is the hypothesis-testing logic of S4. Under progressive rearrangement, copy number takes many values and breakpoints scatter. Under a single event you expect copy number oscillating between just two states along the chromosome, breakpoints clustered far more tightly than uniform, join orientations uniformly distributed over the four possibilities, and retained heterozygosity in surviving fragments — no time for a second hit. Applied to the 2,658 pan-cancer whole genomes of PCAWG, that signature appears in about 29% of tumours on high-confidence calls and 40% when weaker calls are counted — exceeding 50% in melanoma, glioblastoma and lung adenocarcinoma, and reaching 77% of osteosarcomas and effectively all liposarcomas. It also occurs in the , producing congenital disease from one event in one gamete.

10. Uniparental disomy

Both homologues from one parent, none from the other — heterodisomy if the two different homologues of one parent (an MI error, rescued), isodisomy if two copies of the same one (MII or mitotic).

graph LR T["Trisomic zygote<br/>2 maternal + 1 paternal"] -->|"lose one chromosome<br/><b>trisomy rescue</b>"| R{"which one?"} R -->|"the paternal — p = 1/3"| U["<b>Maternal UPD</b>"] R -->|"a maternal — p = 2/3"| N["Normal biparental disomy"] M["Monosomic zygote"] -->|"duplicate the single copy<br/><b>monosomy rescue</b>"| I["<b>Isodisomy</b>"]

Note the arithmetic: rescuing a leaves UPD one time in three. Trisomy is common, so UPD is not rare. Two reasons it matters despite normal sequence content:

Imprinted loci break. Where expression depends on parental origin rather than sequence (Ch 23), two maternal copies of a paternally-expressed gene give zero expression from two intact copies. 15q11–q13 is the standard case: Prader–Willi syndrome from loss of the paternal contribution — usually a paternal deletion, but roughly 20–30% from maternal UPD15 — and Angelman syndrome from loss of the maternal contribution, a few per cent of it paternal UPD15. Same locus, opposite parental origin, different syndromes.

Isodisomy exposes . Two copies of one homologue means homozygosity across a whole , so a single carrier parent's recessive pathogenic variant becomes homozygous in the child — recessive disease with one carrier parent, breaking the logic of Ch 15.

11. Detecting it: a resolution ladder

MethodResolutionDetectsBlind to
G-banded ~5–10 Mbeverything visible, including balanced events and ; needs dividing cellsanything smaller than a band
FISHprobe-limited, ~100 kb upone pre-specified locus; works on interphase nucleieverything you did not choose a probe for
Chromosomal microarray~10–50 kb genome-wideany imbalance; SNP arrays also give → UPD, , triploidyall balanced rearrangements
Short-read WGSbreakpoints to base pairbalanced and unbalanced; CNV from read depth, junctions from split/discordant reads — where happens
Long-read WGS / optical mappinglong reads: SVs from ~50 bp up, breakpoints to base pair; optical mapping: ~500 bp–5 kb up, breakpoints approximatesegdup-resident and repeat-mediated events; phasingcost, throughput, low-level mosaicism

Two consequences. Microarray replaced karyotype for developmental delay, taking diagnostic yield from roughly 3% to 15–20% — but only for imbalance. A couple with recurrent miscarriage and a balanced will have a normal microarray and an abnormal karyotype. Choose by hypothesis, not by resolution. And short reads are weakest exactly where recurrent CNVs live: NAHR needs long near-identical repeats, and short- cannot place reads uniquely in long near-identical repeats (Ch 42). That is not a gap, it is a blind spot aligned with the pathology.


Common misconceptions

What people thinkWhat's actually true
21 survives because 21 is smallestIt survives because chromosome 21 is gene-poorest. Chromosome 19 is smaller than 13, and trisomy 19 is never seen at term
is a structureless accidentIt is highly structured: ~90% maternal, mostly MI, concentrated on with absent or badly placed crossovers, rising steeply with maternal age
The is "old eggs" degrading generallyIt is specifically loaded in fetal life and never replenished, decaying over decades of prophase-I arrest. The prediction that it must be maternal-specific is what makes it a mechanism rather than a slogan
A balanced carrier has a mild version of the syndromeThey have no dosage abnormality and are typically healthy. Their risk is reproductive: unbalanced gametes, recurrent loss, and a chance of an unbalanced child
A normal microarray rules out a chromosomal causeMicroarrays see imbalance only. Balanced translocations and — the findings that explain recurrent pregnancy loss — are invisible to them
CNVs are pathogenic by nature4.8–9.5% of the genome is copy-number variable in healthy people, and CNVs account for more differing base pairs between two genomes than SNVs do. The pathogenic ones are a tail, not the distribution
Complex rearrangements accumulate gradually makes tens to hundreds of breakpoints in one event; two-state copy number, clustered breakpoints and join orientations spread evenly across all four possibilities distinguish it from gradual accumulation
UPD is harmless because all the sequence is thereIt silences imprinted loci, whose expression depends on parental origin, and isodisomy makes a whole chromosome homozygous, exposing one parent's variants
just means a milder versionSometimes. It also makes otherwise-lethal survivable, and it makes tissue choice decisive — a normal blood karyotype excludes nothing elsewhere

Worked example: familial Down syndrome, karyotype to recurrence risk

A child is born with Down syndrome. The :

46,XX,der(14;21)(q10;q10),+21

Step 1 — read it. 46 , but this is not free 21. Count the 21 long arms: one on the derivative, plus two free 21s = three copies of 21q. Trisomy 21 by sequence content, disguised by a normal chromosome count. Roughly 3–4% of Down syndrome arises this way.

Step 2 — find the carrier. About a quarter of such cases are inherited. The mother is:

45,XX,der(14;21)(q10;q10)

One derivative replacing one 14 and one 21, plus a normal 14 and a normal 21. Copy number for both 14q and 21q is two. Balanced, phenotypically normal.

Step 3 — pair them. Three objects share and form a trivalent:

        normal 14  ══════════════
                   ||||||||||||||          14q homology
   der(14;21)      ══════════════╤═════════════
                                 |||||||||||||  21q homology
        normal 21                ═════════════

Step 4 — enumerate. A trivalent segregates 2:1 — two of the three objects to one pole, one to the other. Three choices of which object travels alone, times two poles, gives six gamete classes. (The 3:0 alternative, which would give a doubly trisomic gamete and a nullisomic one, is vanishingly rare.)

Gamete14q dose21q doseFertilised by a normal gamete
der alone11Balanced carrier (45 chromosomes)
normal 14 + normal 2111Normal (46 chromosomes)
der + normal 2112 Down syndrome — the proband
normal 14 alone10monosomy 21 — lethal, early loss
der + normal 1421trisomy 14 — lethal, early loss
normal 21 alone01monosomy 14 — lethal, early loss

Step 5 — theory versus observation. Three of six are lethal and never present as pregnancies. Of the three viable classes one is Down syndrome: a naive 1/3 recurrence risk.

Observed recurrence for a female der(14;21) carrier is 10–15%; for a male carrier under 5%, often quoted as 1–2%. The gap is the teaching point, and it decomposes into three things. is not uniform — the trivalent does not distribute its three objects equally over the six outcomes. Selection continues after conception, so amniocentesis-stage risk exceeds term risk. And gametic selection differs by sex: spermatogenesis filters unbalanced products far more harshly than oogenesis, which is the male/female asymmetry.

Step 6 — the special case. If the derivative were der(21;21), every viable gamete carries two copies of 21q. Recurrence risk is 100%, and the options are donor gametes or preimplantation testing. Worth deriving rather than looking up.

Connections