funded by 
Some of these problems are very specific, others have a more general character. Below we want to give some examples and provide suggestions for changes - some are revolutionary, others are minor adaptions - for further discussion.
Some of the chapters below may look very strange to some readers. They show theoretical cases which are not discussed in the ISCN manual. There might be correct descriptions or descriptions which cannot be recommended - but that will need some discussion.
Most problems seem to arise from the fact that ISCN deals with quite simple aberrations only - generally a chromosome will not experience more than one aberration. If a second aberration follows, there is the "der" notation which can - in theory - cope with dozens of aberrations introduced into one chromosome. But in practical use, it may already fail with the second aberration.
We could try to take a more "mathematical" view on cytogenetics. The introduction of aberrations could be something like an "algebra". But if we now tried to express that the inversion of an inverted fragment leads back to the original input, ISCN will likely fail. Of course, for everydays work of cytogeneticists, such a mathematical body is not necessary, but it would be a useful tool to detect the limitations of the nomenclature.
An even more important problem is education and teaching. There are plenty of detectable errors in the Mitelman database (detectable because the karyotype descriptions there do not comply with ISCN standards), and often it becomes clear that the researchers simply did not know how to denote their cytogenetic findings correctly. This means that the ISCN should be kept as simple as possible.
The ISCN manual referred to in the text is "ISCN 1995 - International System for Human Cytogenetic Nomenclature (1995)" by Felix Mitelman (Editor), Basel, 1995.
For comments and suggestions, please write to us: info@cydas.org.
Table of contents:
Examples:
A chromosome was found with band composition 2qter->2q31::2p21->2qter
Thus some calculation work is needed for re-combining the chromosomes and the bands to get the break points. They are no more seen at one single view. Thus, bands may be placed at the wrong position. Let us have a look at such an example from the Mitelman database:
From the publication by "Boomer
et al 2001, Leukemia", we look at cases 14 and 15:
Case 14: "46,XX,der(19)t(1;19)(q23;p13)"
Case 15: "46,XY,t(1;19)(p13;q23)/46,idem,..."
Between those two cases, the bands of the translocation between chromosomes
1 and 19 have changed their position. Since there is no band 19q23,
the first version seems to be the correct one.
The Mitelman database contains many karyotypes with wrong breakpoints. During an analysis performed in February 2004, we discovered 420 karyotypes with non-existing break points among ca. 22000 investigations from publications of 1995 or later.
We assume that many more wrong break point assignments stayed undiscovered because they do formally exist, while only misassignments leading to non-existing break points can be automatically detected.
We guess that the quality of karyotype data could be greatly enhanced when that separation of chromosomes and bands is given up.
Examples:
"t(9;22)(q34;q11)" would be changed into "t(9q34;22q11)".
"der(19)t(1;19)(q23;p13)" would be changed into "der(19)t(1q23;19p13)".
These examples show at the first view where which aberration did occur. We approve that many cytogeneticists may be interested first in which chromosomes are afflicted by the aberration and then at a second view in which bands are involved. But still we favor the changed version for overall clarity. Some visual emphasis like bold print in documents (examination reports, publications) could be optional and would abolish such objections.
Though one might argue that the two extra chromosomes might be derived from both original chromosomes instead of from one only, this is certainly not true for cases where a ploidy change happened after a primary gain / loss of a normal chromosome (e.g. "45,XX,-8" -> "90,XXXX,-8,-8").
We suggest the use of multiplicators also at this position.
Here we suggest a consistent use of multiplicator placing, i.e. the former example would be changed to "49,XY,+marx3". CyDAS happens to cope with both styles (and even other not suggested styles).
We suggest a consistent use of multiplicators for all types of aberrations, hence "dminx10~15"
On page 57 of the ISCN manual, a hardly understandable example is shown where the comma is used inside an aberration: "46,XX,der(1,3,11)t(1;3)(p32;q21)inv(1)(p22q21)t(1;11)(q25;q13)". Already there in the manual, this form is "not recommended".
No such case was found in the Mitelman database.
Apart from the fact that hardly anybody will understand the meaning of the above formula, its use would destroy the very first step of analysing an ISCN formula: splitting the formula at its commata in order to receive its consituent items.
We suggest that this use of comma be abolished.
The "or" can also be placed inside an aberration (e.g. "46,XX,add(19)(p13 or q13)"). Such a style was not discovered in the Mitelman database, and CyDAS cannot deal with it. The effort which would have to be undertaken to cope with the "or" at such a position would be enormous. We suggest that the "or" should be placed between aberrations only, i.e. the example above would be changed into "46,XX,add(19)(p13) or add(19)(q13)".
A greater problem arises if in an alternative interpretation a single aberration was split into two aberrations. Imagine that the primary interpretation of a karyotype is "46,XY,t(1;11;13)(p34;q14;q14)", while the alternative is "46,XY,t(1;13)(p34;q14),del(11)(q14)". "46,XY,?t(1;11;13)(p34;q14;q14)or?t(1;13)(p34;q14),?del(11)(q14)" will likely not be understood correctly (this case was taken from the Mitelman database; it may be wrongly interpreted). We have not found a useful suggestion for such cases.
For example, the band composition of "der(1)t(1;3)(p32;q21)inv(1)(p22q21)t(1;11)(q25;q13)"
is
"3qter->3q21::1p32->1p22::1q21->1p22::1q21->1q25::11q13->11qter".
We suggest to clearly discourage the use of the "short" system for such
highly aberrant chromosomes and recommend the "detailed" system for chromosomes
with more than two aberrations.
Example: After translocation t(1;2)(p31;p21) an inversion occurs at 1q22 and 2p23. The aberrant chromosome is denoted as "der(1)(2pter->2p23::1q22->1p31::2p21->2p23::1q22->1qter)" in the detailed system. We do not know a way to describe this derivative chromosome in the short system, since some way of denoting recursion would be required.
Hence, we suggest to emphasize the use of the detailed sytem whenever the description of a derivative chromosome in the short system raises doubts.
Example: The dic(5;7)(q22;p13) gets involved in t(3;7)(q21;q21). The resulting derivative dicentric chromosome is then denoted "der(5;7)t(5;7)(q22;p13)t(3;7)(q21;q21)" instead of "der(5;7)dic(5;7)(q22;p13)t(3;7)(q21;q21)".
We suggest that the symbol "t" denote translocations only, and the symbol "dic" be used for all cases of formation of dicentric chromosomes.
E.g. the der(22) of t(9;22;1)(q34;q11;p31) is der(22)(22pter->22q11::9q34->9qter), the same with the "normal" Philadelphia translocation t(9;22)(q34;q11).Still, it is noted as "der(22)t(9;22;1)(q34;q11;p31)" not as "der(22)t(9;22)(q34;q11)" here.
When a three way translocation has been safely identified, it looks quite logic to use it for the derivative chromosome. But that may cause problems: most derivative chromosomes whose description contains a three (or more) way translocation from the Mitelman database look bogus.
Examples:
Rock
et al 1993, Surg Neurol, Case 10: 44,XY,der(2)t(2;10;14)(q35;q23;q11),-10,-13,-14.
From the number of chromosomes lost and replaced, chromosome count is 43.
Next, how can one tell that the terminal fragment of chromosome 2 was translocated
onto chromosome 10, if no der(10) was found?
Jarosová et al 2003, Cancer Genet Cytogenet, Case 5 (many other cases look bogus, too): 46,XX,del(6)(q23),t(12;21)(p13;q22),der(20)t(12;21;20)(p13;q22;q1?3). The t(12;21;20)(p13;q22;q1?3) would give raise to der(12)(12qter->12p13::20q1?3->20qter), der(21)(21pter->21q22::12p13->12pter) and der(20)(20pter->20q1?3::21q22->22qter). No information is given on these der(12) and der(21). Also the extra t(12;21)(p13;q22) would give raise to a der(12) and a der(21). Furthermore, in leukemia the translocation of a translocation product is not unusual, but that would be described as a derivative chromosome with two separate translocations.
Furthermore, when two of the chromosomes involved have the same number, this notation will obscure which derivative chromosome is meant. E.g. is "der(1)t(1;1;3)(p22;q31;q21)" (1) "der(1)(3qter->3q21::1p22->1qter)" or (2) "der(1)(1pter->1q31::1p22->1pter)"? This problem is also due to a lack of differentiation between homologous chromosomes.
Sometimes, people have a totally different concept of a three way translocation. An example is described in our HowTo on ISCN Errors.
Example:
We start with a whole arm translocation between chromosomes 1
and 3:
46,XY,t(1;3)(p10;q10). Now let us introduce an
inversion in that derivative chromosome having the p-arm of chromosome
3
at 3p13 and 3p23.
The problem now is that we must be able to distinguish between the
two chromosomes which were the result of the balanced whole arm translocation:
the "der(1;3)(p10;q10)" describes only one of the chromosomes
and replaces both normal chromosomes 1 and 3, and the
"der(1;3)(q10;p10)" describes the other one of the chromosomes
and also replaces both normal chromosomes 1 and 3. Hence,
the formula (1) "46,XY,der(1;3)(p10;q10),der(1;3)(q10;p10)inv(3)(p13p23)"
is not correct, since it describes two derivative chromosomes stemming
from
unbalanced translocations, no normal chromosomes 1
and 3 would be present and chromosome count would be 44.
Also, the formula (2) "46,XY,+der(1;3)(p10;q10),der(1;3)(q10;p10)inv(3)(p13p23)"
and the formula (3) "46,XY,der(1;3)(p10;q10),+der(1;3)(q10;p10)inv(3)(p13p23)"
cannot be correct because both imply the gain of a derivative chromosome
while balanced aberrations only did occur.
Furthermore, the formula (4) "46,XY,der(1)t(1;3)(p10;q10),der(3)t(1;3)(p10;q10)inv(3)(p13p23)"
is not correct because the way the derivative chromosomes are shown here
implies that they received their centromers from one parent chromosome
only.
Though the aberrations in this example are still very simple, they cannot be denoted in the short system of the present version of the ISCN. Also, the detailed version ("46,XY,der(1;3)(1pter->1p10::3q10->3qter),der(1;3)(3pter->3p23::3p13->3p23::3p13->3p10::1q10->1qter)") is not acceptable because of the replacement problem like in formula (1).
We do not have a suggestion how this issue can be solved correctly.
The following notation can be calculated to the correct karyogram and can
be understood easily, but still implies gains and losses which never did
occur:
46,XY,-1,+der(1;3)(p10;q10),+der(1;3)(q10;p10)inv(3)(p13p23),-3
According to the ISCN manual (p. 69), also interstitial fragments may
be exchanged. When scanning several thousands of karyotypes for such a
translocation, only one was found: t(2;4)(p15q23;p14q24)
stemming from a publication from 1993, i.e. before ISCN 1995; and here,
a centromer containing fragment would have to be exchanged.
No karyotype showing a translocation of interstitial fragments between
three chromosomes was discovered.
The example given in the ISCN manual ("t(5;6)(q13q23;q15q23)")
could not be found in the Mitelman database.
Hence, doubts if such translocations do really exist may be advisable. If these doubts could be substantiated further, the nomenclature for the exchange of interstitial fragments could be abolished.
If such a chromosome faces another aberration, it must be denoted as a derivative chromosome. A terminal deletion in the above mentioned isodicentric chromosome would be written "der(7;7)idic(7)(p12)del(7)(q31)". In the der() clause there must be two chromosomes 7 because of their two centromeres in the derivative chromosome (p. 55). But a "der(7;7)" means that two normal chromosomes 7 have been replaced by that derivative chromosome, and that is definitely wrong.
The core problem is again the replacement rule or the rule for denominating the chromosome(s) in the der() clause. No problems would arise if centromeres of distinct origin only were to be denoted. See also: Differentiating homologous chromosomes
We could conclude that the ISCN formula is:
46,XX,der(9)t(9;22)(q34;q11),-22,+der(22;22)t(9;22)(q34;q11)idic(9)(q34)
But can an iso-dicentric at such a site be described with the symbol idic? There is no centromere of chromosome 9 available in the der(22), and it is the centromere of chromosome 22 which gets duplicated (again, there are some problems with the two centromeres of the same origin; see also "Differentiating between homolgous chromosomes"). No such example is given in the ISCN manual.
Example:
A fragment of chromosome 1 band 1p33 to 1p35
has been inserted into chromosome 2 at 2p16: ins(2;1)(p16;p33p35).
If the fragment was very short (from band 1p33 only), the description
typically used is "ins(2;1)(p16;p33)" instead of "ins(2;1)(p16;p33p33)".
This is inconsequent.
The inconsequence becomes clearer when we look at a different type of aberration, e.g. a deletion. If chromosome 1 has lost a fragment from 1p33 to 1p35, it is "del(1)(p33p35)". But if the fragment is very short - again from 1p33 only - the description cannot be "del(1)(p33)" because that describes a deletion starting from 1p33 upto (and including) the p terminus.
For a clear and consistent nomenclature, we suggest that such omissions be explicitly forbidden.
Though this looks pretty easy, it may cause confusion. Let us start with a karyotype having two different translocations between chromosome 1 and 2: "46,XX,t(1;3)(p31;p21),t(1;3)(q23;q33)". Now, the derivative chromosome 1 has been gained: "47,XX,t(1;3)(p31;p21),t(1;3)(q23;q33),+der(1)t(1;3)". Which ones are the break points??
We suggest that break points must always be repeated for reasons of clarity.
But a chromosome is a linear structure which starts at one end (e.g. p terminus) and extends to its other end (e.g. q terminus) via its centromere(s). It is not an anchor (a centromere) on which two linear structures (the arms) are fixed; that description becomes problematic by itself when chromosomes have more than one centromere.
From the point of view of the fragment composition of derivative chromosome, only the clear linear view is useful. And now, the definition of a direction towards a centromere or away from a centromere looses its meaning - a "plus strand" or "minus strand" definition, as used by the sequencing people, makes sense.
If we look at the examples given in the chapter on introducing duplications, interchromosomal insertions and intrachromosomal insertions into derivative chromosomes, it becomes immediately obvious that in a "direct" event the band named first can be found second in the derivative chromosome (i.e. the fragment was inverted from the linear view), and in an "inverted" event the first band can be found first in the derivative chromosome (i.e. the fragment was not inverted from the linear view), as well as agreeing definitions.
The ISCN standard definition becomes non-applicable if the starting chromosome for an aberration is a derivative chromosome. If we start with a paracentric inversion inv(1)(p21p33) and then do an intrachromosomal insertion with a fragment from the q-arm into the inverted fragment, how is then the orientation towards the centromere defined - towards a "virtual" centromere respective to the inverted fragment 1p23->1p33, or to the "real" centromere of the derivative chromosome?
If someone thinks that is obvious, just get a more complicated starting point. After an interchromosomal insertion ins(1;2)(q32;p12p24) (= der(1)(1pter->1q32::2p12->2p24::1q32->1qter), we form a dicentric chromosome der(1;3)ins(1;2)(q32;p12p24)dic(2;3)(p23;p21) = der(1;3)(1pter->1q32::2p12->2p23::3p21->3qter). What is then the orientation of the fragment "2p12->2p23" towards the centromere (which one)?
We suggest that it could be optionally added to the chromosome count field in angle brackets together with the ploidy level or alone, preceded by the symbol "mn".
Example:
"46-52<mn48,2n>,XX,+1,+2,+3,+4,+5,+6,+7,+8,-9,-10,-11,-12,-13[cp]"
or "46-52<mn48>,XX,+1,+2,+3,+4,+5,+6,+7,+8,-9,-10,-11,-12,-13[cp]"
We suggest that this dot be declared optional instead of compulsory.
We suggest that optional spaces may be allowed.
Typically, no information is given if the ring chromosome is monocentric or multicentric.
The symbol "dic r" for dicentric ring chromosomes is found only once in the Mitelman database, the symbol "trc r" is not found at all.
Does a "r(12;16)" replace chromosome 12 or chromosome 16 or both? If it were monocentric, the correct description would be "der(12)r(12;16)" or "der(16)r(12;16)" and the chromosome given in the der() clause was to be replaced; otherwise the description would be "dic r(12;16)" and both chromosomes would be replaced.
Sometimes, the band composition can be used to find the information: "r(1;9)(p12q44;p11q24)" has centromeres of both chromosome 1 and 9 thus the correct version was "dic r(1;9)(p12q44;p11q24)".
In a few cases, only one band instead of one fragment per chromosome
involved is given. Such cases might be typos, and a translocation be meant.
With "der(9)r(9;22;16)(q34;q11;p?)" (Morel
et al 2003, Cancer Genet Cytogenet, Case #4), a ring formed of the
der(9) of a t(9;16;22) was meant according to the abstract
of the article. This may indicate problems with describing rings derived
from derivative chromosomes.
Furthermore, the short system does not privide means for describing ring chromosomes derived from isochromosomes / iso-derivative chromosomes, e.g. ring formation in 11q13 and 11q23 in an i(11)(q10).
While any other aberration symbol can be used for a simple aberration, there are some cases where "r" can be used only in conbination with "der", i.e. when a ring chromosome is set up by more than one chromosome and is not dicentric. E.g., a "r(1;3)(p36q23;q21q27)" by itself is wrong, it must be "der(1)r(1;3)(p36q23;q21q27)".
The chromosome introducing the centromere need not be the first chromosome named, e.g. "der(?15)r(13;15;15)(q?21q?32;p1?1q?21;q?22q?26)" where the second chromosome introduces the centromere.
Next, the description of dicentric derivative ring chromosomes gets confusing, e.g. "der(1;5)dicr(1;3;5)(p36q23;q21q27;?)".
We do not have a suggestion for solving these problems in the short system. The detailed notation seems to be advisable for complex ring chromosomes and thus should be encouraged.
More complicated is the description of the derivative chromosome originating from that event. It has two centromeres, thus in its der() clause two chromosomes have to be named (pp. 35, 55): "der(1;1)(1pter->1q21::1p21->1qter)". And whenever two chromosomes are named in a der() clause, two normal chromosomes are replaced with this one derivative chromosome, thus implying a loss of large amounts of chromosomal material while in fact a pericentromeric region was gained. No problems would arise when only centromeres of distinct origins would be denoted in the der() clause. With duplications, the CyDAS software internally tries to keep track of the origins of the centromeres and then denotes centromeres of distinct origin - in contradiction to the ISCN manual - only; when introducing further aberrations into the chromosome, the track keeping does not always work.
See also: Differentiating homologous chromosomes
For example, a "der(1;3)(1pter->1q32::3p14->3q21::1q32->1qter)" could be interpreted as a "der(1;3)ins(1;3)(q32;p14q21)" (note the obligatory der() clause because the remaining parts of chromosome 3 will be lost). The ISCN manual does not describe such cases. Maybe this interpretation is not accepted.
For example, an "ins(1;3)(p35;q25q27)" is by itself an interchromosomal insertion. But when we start with a derivative chromosome, e.g. "der(1)t(1;3)(q32;q21) = der(1)(1pter->1q32::3q21->3qter)", the insertion of "der(1)t(1;3)(q32;q21)ins(1;3)(p35;q25q27)" can also be intrachromosomal, thus leading to "der(1)(1pter->1p35::3q27->3q25::1p35->1q32::3q21->3q25::3q27->3qter)", and of course interchromosomal "der(1)(1pter->1p35::3q27->3q25::1p35->1q32::3q21->3qter)".
See also: Differentiating between homologous chromosomes
From a theoretical point of view, such regions could also reside at
them, e.g. "der(1)del(1)(p22)hsr(1)(p22)". Is such a description
correct? As we can deduct from the last paragraph of that chapter, there
is no definitive distinction of "homogeneously staining regions", "abnormally
banded regions" and "regions of unknown origin". Hence, that derivative
chromosome in the example above could be "add(1)(p22)".
After the number of the chromosome, the symbol "b" for "base pair" and the number of the base pair follow.
If the point of rearrangement in a Philadelphia chromosome has been sequenced and found to correspond to base 130785212 of chromosome 9 and base 21852311 in chromosome 22, the derivative chromosome 22 could be described as "der(22)(22pter->22b21852311::9b130785212->9qter)". A resolution at kilobase level could be "der(22)(22pter->22b21852k::9b130785k->9qter)", and at Megabase level der(22)(22pter->22b21M::9b130M->9qter), the translocations could be described as "t(9;22)(9b130785212;22b21852311)", or "t(9;22)(9b130785k;22b21852k)", or "t(9;22)(9b130M;22b21M)", resp.
By mapping the probes of FISH to their corresponding positions on the chromosome, this type of nomenclature could help unifying the distinct styles.
But this denomination cannot be applied for chromosomes containing two or more bands originating from bands with the same desigantion. In a "der(1)dup(1)(q12q24)", the location of "1q22" is not clear.
It is not a complete description because many cases remain where no result can be given. Furthermore, several examples can be found throughout the ISCN manual where that rule is not observed:
Hence, a new rule was implemented. Its hallmark is that the pericentric region is ordered in plus-strand orientation (i.e. the centromeric region is parallel to the centromeric region of a normal chromosome). If several centromeres are present, the centromere with the lowest number is relevant. If on both sides of the centromere are segments of the same chromosomal arm (e.g. when the chromosome is derived from an isomerization), the longer segment is relevant (i.e. a longer p arm segment will look upwards, while a longer q arm segment will look downwards).
As of writing, there were no new complaints on confusingly ordered chromosomal bands.
Hence we suggest to update that chapter to the newer rule.
Underlining is not a plain text compatible method, hence no such data can be found in databases. Furthermore, it is not applicable in case of triploidy or trisomy.
We suggest to add a single character after the chromosome number, in alphabetic ordering, e.g. 46,XX,der(13;21A)(q10;q10),add(21B)(q21) to distinguish between such chromosomes. For triploidy (constitutional or acquired), a "C" could be used, and for farther chromosomes accordingly, e.g. 47,XX,der(13;21A)(q10;q10),add(21B)(q21),+21C.
This would also help to cope with some rearrangements at the centromeres.
A chromosome with band composition (1pter->1q21::1p21->1qter)
could originate from a dup(1)(p21q21) or a dic(1A;1B)(p21;q21).
If the rule for denoting the chromosomes involved in a derivative chromosome
would then be adjusted so that only centromeres of distinct origins be
named, many problems with derivative chromosomes could be solved:
the dup would translate into a der(1A)(1Apter->1Aq21::1Ap21->1Aqter)
while the dic would become a der(1A;1B)(1Apter->1Aq21::1Bp21->1Bqter).
The number of chromosomes replaced by the derivative chromosomes would
be clear.
See also:
An ISCN formula does not only describe the state of a karyotype, but typically tries to give a hypothesis how that state did arise, i.e. the development of that state.
E.g. when on a chromosome #9 material from chromosome #22 from 22q11 to 22qter is found to replace the region from 9q34 to 9qter, and vice versa, the description is typically "46,XX,t(9;22)(q34;q11)" thus indicating a step from a normal karyotype to the aberrant karyotype. The description of the state only, without implying any steps to get there, would be "46,XX,der(9)(9pter->9q34::22q11->22qter),der(22)(22pter->22q11::9q34->9qter)" which generally is not used.
Furthermore, sometimes the ISCN tries to show the method with which a karyotype was analysed. If a normal karyotype was found with FISH analysis, the description can become extremely long - and many physicians will conclude that that's a bad situation for the patient: he has so many aberrations. With FISH, also normal findings are shown, and not only aberrant findings.
Sometimes, FISH results do not only confirm banding results, but alter them. A few examples are given in the ISCN manual (pp. 95ff):
CGH shows aberrant findings only, but uses a different way to denote the aberrations.
Imagine the speed of car would be given in different ways depending
whether it was measured with the speedometer of the car or with radar or
with other methods - everybody would call that inconvenient. But that's
actually the way the current ISCN deals with chromosome findings.
Hence, only a few symbols need to be integrated into the standard ISCN:
The symbol "sep" (=separated) means that two bands which normally reside on the same chromosome were found on different chromosomes. FISH: colocalistion missing where expected.
The symbol "con" (=connected) means that two bands which normally do not reside close to each other came close to each other. FISH: colocalisation found where not expected.
Integration of CGH-Symbols: "enh", "amp", "dim"
Further symbols:
loss of heterocygosity