## Information that Can Be Used to Classify A Variant

Geneticists use many types of information can be used to determine whether your variant changes disease risk:

- Comparing your variant to other variants
- Looking at how common your variant is in others with and without disease
- Seeing whether the variant might be present in the same gene in other species

It is very likely that this information has already been evaluated and that it is insufficient to classify your variant, hence the classification of “variant of uncertain significance.”

One of the most powerful and well-established ways to gain information about a genetic variant is called familial co-segregation, linkage analysis, or family analysis. This analysis consists of determining if a variant travels with a disease in a family more often than would be expected by chance. The information necessary to classify your variant may be present in your family.

## Basics of Using Family Information for Family Co-Segregation Analysis

Family analysis has been used for decades to identify disease genes. The main challenge of family analysis is investing the time and effort to gather a large enough family tree or pedigree to conclude that a variant and disease happen together more often than expected by chance. In genetics there are rules for calculating the probability that the disease and variant happen together by chance, and in medical genetics there are standards for the level of evidence that is necessary to classify a variant.

Here is an example to illustrate how geneticists calculate probability and what types of families contain the evidence necessary to reclassify a VUS. We will start with a diagram of the small family tree/pedigree we will use as example of how family analysis works. This may be similar to the family tree the genetic counselor drew at your genetics visit.

In this diagram the father is represented by a clear square, the mother is a clear circle, and there are three daughters that are each filled-in circles. The filled-in circles show that all three daughters have the same disease. The plus symbols show that they each inherited the same rare VUS from their father.

At first glance, one might think that it is clear that the disease and the variant are traveling together, but this might not necessarily be true. It is likely the family was tested because all three sisters have the same disease, so we need to calculate the probability that all three sisters have the same rare variant.

Each of us has two copies of most genes. We inherit one from our father and one from our mother. If one of the daughters was diagnosed with the VUS, then there would be a ^{1}⁄_{2} chance one sister inherited the same variant from her mother regardless of disease association, and a ^{1}⁄_{2} chance the other sister inherited it as well. So the probability that both sisters not initially tested have the same rare variant by chance is ^{1}⁄_{2} × ^{1}⁄_{2} or ^{1}⁄_{4} (25%).

On the other hand, there is also a chance that the variant is associated with disease and that it is not a coincidence that the variant is in all three sisters with the disease. This probability is more complicated and depends on the frequency of the disease in the general population. Geneticists use ratios of probabilities, called likelihood ratios, to determine if there is enough evidence to classify a variant.

## Evidence Standards for Variant Classification

Several organizations have recommended that for medical decision-making, a 99% chance that a variant is associated with disease is sufficient to classify a variant as pathogenic. This is equivalent to a likelihood ratio of 99 or more. If there is a 99% to 95% chance the variant causes disease a variant would be classified as likely pathogenic (likelihood ratio of 99 to 19). On the other hand, less than 1% chance of causality (likelihood ratio less than 0.01) would classify a variant as benign and 5% to 1% chance of causality (likelihood ratio of 0.05 to 0.01) would classify a variant as likely benign.

Classification | Likelihood Ratio for Causality | Probability of Causality |
---|---|---|

Pathogenic | Greater than 99 | Greater than 99% |

Likely Pathogenic | 19 to 99 | 95% to 99% |

Variant of Uncertain Significance (VUS) | 0.05 to 19 | 5% to 95% |

Likely Benign | 0.01 to 0.05 | 1% to 5% |

Benign | Less than 0.01 | Less than 1% |

## Probability of Causality is Different than Disease Risk

Most variants in disease risk genes either do or do not disrupt the function of the gene regardless of how they are classified.

Here is an example to illustrate the difference between probability of disease and probability of a variant being pathogenic or benign:

I like to pack light. I ask my friend, who is a weather reporter, if I should bring an umbrella on a trip to Pennsylvania next week. She says she does not know; it depends on where I am going. The probability of rain in Philadelphia is 60% and the probability of rain in Pittsburgh is 3%. When I tell her I am going to Blossburg, Pennsylvania, she says that she does not know where Blossburg is or what the weather might be. If we knew that the weather in Blossburg were like the weather in Philadelphia 95% of the time, then we would know that I could expect about a 60% chance of rain. If I knew that the weather in Blossburg were like the weather in Pittsburgh 95% of the time, then we would know that I could expect about a 3% chance of rain.

The term VUS is similar. It expresses uncertainty about knowing if a variant is pathogenic (with similar probability of disease as other pathogenic variants) or benign (with similar probability of disease in most people). The term VUS says nothing about the probability of an individual getting a disease. It may be high or low. Variant classification through family analysis is the biological equivalent of looking to see if the weather in Blossburg is more similar to the weather in Philadelphia or more similar to the weather in Pittsburgh. When we talk about the probability of a variant being pathogenic or benign, we are describing how similar it is to other pathogenic or benign variants, not giving a probability of the disease happening.

## Classifying a Pathogenic Variant

Getting back to our pedigree example, with additional details filled in such as the specific disease (breast cancer in this case) and the ages of diagnosis, we could calculate a likelihood ratio. For this family the likelihood ratio of co-segregation is 3.2 or a 76% chance that the variant is causing breast cancer. Our conclusion would be that this family shows some evidence that this variant is associated with the disease, but that there is not enough evidence to classify the variant. If this were considered to be enough evidence to make medical decisions then doctors acting on this would be wrong almost 25% of the time, or 1 in 4 families.

Adding more relatives will increase the probability that the variant is classified. If a variant is pathogenic then adding distant relatives that have both the variant and the disease is especially powerful because the probability that these relatives have the same variant by chance is very low. This diagram shows how adding a few cousins to a pedigree can potentially increase likelihood ratios to allow variant classification. Remember, a likelihood ratio over 20 is considered medically actionable.

In this diagram the individuals analyzed are in colored boxes and the likelihood ratios corresponding to analysis of just those individuals is in colored text. You can see how the likelihood ratios grow with more relatives and more distant relatives. Although every family is different, it usually takes identifying about 3 to 7 individuals that have both the disease and the variant to classify a variant as likely pathogenic. The more distant the relatives with the same phenotype are, the fewer it takes to classify a variant because more genetic events separate distant relatives.

## Classifying a Benign Variant

In one study 80% of variants initially classified as VUS eventually turned out to be benign. The information needed to classify a variant as benign is different in some ways. Knowing about more individuals that have the variant always increases the information about that variant. Unlike pathogenic variants, the distance of the relationship is not as important to classify a variant as benign, so knowing about lots of close relatives that carry your variant could add very useful information. Although every family is different, it usually takes about 4 to 8 individuals who do not have the disease but do have the variant to classify a variant as likely benign. The diagram below illustrates what segregation analysis might look like for a benign variant.

## Using Other Information to Help Classify a Variant

Geneticists who have experience in variant classification can combine pedigree information with other information about your variant to generate combined likelihood ratios, which may help classify your variant. For example, if your family analysis shows a likelihood ratio of 15, this might be combined with a likelihood ratio of 2 from other information to give a combined likelihood ratio of 30, which would be enough to classify your variant as likely pathogenic. Similarly, if there is a likelihood ratio of 0.15 from five unaffected individuals in your family who all have your VUS, this might be combined with a likelihood ratio of 0.25 from other information to give a combined likelihood ratio of 0.0375, which would be sufficient to classify your variant as likely benign.

Methods for calculating exact likelihood ratios for non-pedigree information have not been standardized. Many laboratories use decision rules that approximate likelihood of causality without numerical calculations. Different genetics laboratories also use different methods to combine information about variants.