Are the predisposing genetic and triggering factors of aHUS fully catalogued and understood and will it help to know how variable are the risks of these between individuals?
That is a question from the aHUS Patients Research Agenda.
Yes, and what is the answer?
To the first part , genetic mutation in complement components, they are are being gathered and the genetics is being explored. Similarly new triggers are being found. To the second part it is early days, but when aHUS is found on onset it varies between individuals and so is unpredictable to a large extent.
So all Complement genetic mutations/ variants are being gathered.
Well not all have been captured and not all in one place. There are many databases and not all share data with each other. And not all variants which are found are published in articles. so may not reach any of those databases.
OK but for those that are collected they know how they cause aHUS in different people?.
Firstly not all genetic mutations in Complement components cause aHUS. Each mutation found is classified.
There are five classes : Pathogenic, Likely Pathogenic , Uncertain significance, Likely Benign , Benign.
Yes disease causing. Something that would cause the component not to work as normal. Either it cannot do its control job or there is too much of it to overwhelm the controls of the complement system.Conversely benign mutations do not cause aHUS. Those in between have no, or only some possible,certainty for causing aHUS.
Who classifies them and how do they decide the classification.
It would require a geneticist and there two ways to determine classification.
OK what two ways are used?
One is a functional test of a mutated component done in the laboratory in a test tube ( “in vitro” ). In a specimen of blood the mutated component is observed to see whether it does its job or not. If it does it, it is classified as benign, if it doesn’t it is determined pathogenic.
And the other way?
A computerised genetic model is used based on certain algorythms and rarity of mutations. It then predicts whether it is pathogenic or not.
The first method would seem to be more accurate.
It may be but it is more expensive to do . Predictive modeling is more commonplace in the world of genetics not just for aHUS . But both methods have their drawbacks.
So there are concerns about the accuracy of classifications.
Yes and an example will help explain it. The example involves Complement Factor H , CFH. Mutations of CFH are implicated in around 1 in 5 cases of aHUS. It is a key controller of Complement’s alternative pathway activity. Patients with CFH mutations frequently have the worst prognosis. So proper classification is important for patient management.
Agreed but tell us more.
What I am going to tell you is based on the work of an American/ Spanish research group led by Hector Martin Merinero from Madrid. The science in their work methods is quite difficult for lay people to fully understand so this can only be the gist of what they discovered. Their full report can be read using a link at the end of this article*.
What did they do?
The group set out to establish how much more accurate the testing and predictive modeling could be and eliminate the uncertainty of both.
They looked at 105 CFH mutations. 26 that had been previously classified and 79 mutations that had not been. They then did laboratory tests of expression (visible) and function (what is does or does not do when active) of each mutated CFH . They also used the prediction model method i.e. . CADD -Combined Annotation Dependent Depletion ( version 1.6). The mutations were in different bits of CFH and were created in a lab for the study.
Different bits of CFH ?
Yes. Picture CFH as a string of 20 beads, each bead is bit or in doctor’s language a “short consensus repeat” or SCR. The string has two ends (terminals) and a middle. If the beads were numbered 1 to 20, then beads 1 to 4 would be at the “N terminal” and 19 and 20 at the “C terminal”. SCRs 5 to 18 are regarded as the mid-region.
Also there even differences within the bits. Some SCRs are involved in the binding with other parts of Complement like C3 and CFI, or the “sialic acid” at the surfaces of blood vessels, to control Complement activity. Mutations within them are more problematical for Complement control. Other bits of SCRs were not involved in the control process as such.
They allotted the mutations to each bit (SCR) they investigated and also to the amino acids which made up each bit of the SCR. Imagine the each bead being made up of ribbon like strands. Each strand being composed of amino acids.
Sounds complicated, let’s accept the ” N” and ”C” and SCRs roles , but what did they find?
Yes it is. First of all, of the 26 previously classified mutations they found that 24 were classified correctly but In this study there was no uncertainty. They were either pathogenic or not.
And what about the 79 previously unclassified?
Well, knowing that study methodology was working they classified 52 as non aHUS causing and 27 as pathogenic for aHUS. They also observed how the mutations were scattered across the 20 SCRs . 20 were in the N Terminal SCRs1-4, 32 were in the mid region SCRs 5-18 , and 27 were in the C terminal SCRs 19 and 20. There were relatively more in the SCRs of the C terminal of CFH and more of them were pathogenic. An aHUS hot spot!
How many were pathogenic in each part?
8 of the 20 in the N terminal were pathogenic. 19 were pathogenic of the 27 in C terminal. And none of 32 in the mid section were found to be pathogenic.
What did the computer model say about these genetic mutations.
There were differences. 5 in the N terminal, 3 which were predicted pathogenic were benign when tested, and 2 predicted benign were found to be pathogenic. In the mid section 6 previously classified as pathogenic were found to be benign .In the C terminal there were 7 discrepancies , 2 previously classed pathogenic were benign and 5 predicted benign were pathogenic.
So there were incorrect predictions and maybe there will still be.
Yes the predictive modeling method is not fully reliable and must be treated as such but the researchers did find a way to improve predictions by adjusting the scoring method threshold for CFH mutations . The normal score threshold “19” was found to be suitable for N terminal and Mid section SCR mutations but not for C terminal mutations where a threshold of 10 was found to be a better level. An above 10 predictor score was more accurate for classifying disease causing variants.
Do I understand it right in vitro functional testing of CFH mutation pathogenicity is more accurate but is expensive to do and such patients’ costs are not always payable. So predictive modelling is more cost effective and most but not all pathogenic mutations can be classified correctly. But the researchers have improved predictive modeling classification by identifying and recommending a revised CFH specific threshold for C terminal mutations.
That’s about it. Oh and the research group also found that measuring the CFH levels in the patients blood also helped with the classifying CFH C terminal mutations process.
More accurate aHUS pathogenicity awareness can then help in individualised treatment
Yes that is the important outcome from the findings by this research group, but more work will be needed. There are not only a lot more CFH mutations than 105, but also other components of the Complement System have their own pathogenic mutations too. A definite “yes or no” answer to the pathogenicity of all aHUS Complement mutations is still some way off.
There is an on line database which anyone can access which is hosted by University College London (UCL) and can be accessed at this link HERE. Patients who may know their specific aHUS pathogenic mutation may find more information about it if it is in there. More information about the UCL complement genetic database can be found HERE
*The full article which inspired this post can be read at this link HERE
Article No. 531