Down's Syndrome Biochemical Screening - Part 2 Flashcards
Can you use age alone as a screening test?
- At age 35 risk is 1 in 356.
- This represents approximately 5% of pregnancies.
- This means that is we just used age as a screening test and tested al those women who were 35 and older the overall detection rate would only be 30%.
- Although risk is higher in older women, far more younger women are pregnant.
What markers were proposed to combine with maternal age in order to improve the screening test compared to using age alone?
- At age 35 risk is 1 in 356.
- This represents approximately 5% of pregnancies.
- This means that is we just used age as a screening test and tested al those women who were 35 and older the overall detection rate would only be 30%.
- Although risk is higher in older women, far more younger women are pregnant.
- People started to look for markers that they could measure in a maternal blood sample = relatively straight forward and painless.
- People also started looking at ultrasound markers.
- Serum markers that have largely gained acceptance for use in Down syndrome screening include:
1) . Alpha-fetoprotein (AFP)
2) . Human Chorionic Gonadotrophin (HCG)
3) . Unconjugated estriol (UE3)
4) . Inhibin A
5) . Pregnancy Associated Protein A (PAPPA)
6) . Newer markers (ADAM12).
List the main serum markers that have largely gained acceptance for use in Down syndrome screening.
Serum markers that have largely gained acceptance for use in Down syndrome screening include:
1) . Alpha-fetoprotein (AFP) = often used as a tumour marker.
2) . Human Chorionic Gonadotrophin (HCG)
3) . Unconjugated estriol (UE3)
4) . Inhibin A
5) . Pregnancy Associated Protein A (PAPPA)
6) . Newer markers (ADAM12).
Are there usually clear cut-off values between marker serum marker values in affected and unaffected individuals?
- Ideally there would be very clear differences between serum marker values in individuals very likely to be affected and serum marker values in individuals very unlikely to be affected. With markers such as this we can create a clear cut-off value between the 2.
- In most cases however you get a situation where a particular marker value make the individual more likely to be in a particular group (i.e. affected or unaffected) but there is quite a lot of overlap between the two.
- We need some way of relating our serum marker concentration to the relative chance of the individual being in either the affected or unaffected group.
What is the function of Alpha-fetoprotein (AFP) in the foetus? How do AFP levels vary throughout pregnancy? How does this affect our assessment of Down syndrome risk based on AFP marker levels during pregnancy?
- Essentially functions as a transport protein in the foetus transporting molecules around the blood. Also has a semi-structural function in keeping fluids in the right place - basically the same function as albumin in the adult.
- AFP concentration goes up in the early stages of pregnancy and begins to tail off as the foetus makes albumin in the later stages of pregnancy.
- The foetus itself will loose AFP into the maternal fluid through the foetal urine. If we measure the concentration of AFP in amniotic fluid we would find that the concentrations largely follow those in the actual foetal blood itself.
- As you have the circulation between the amniotic fluid and the maternal circulation you end up with AFP levels rising in the maternal blood circulation so you can take a blood sample from the mother and it largely follows the same sort of time curves.
- If we were going to compare whether you were more or less likely to be in the unaffected or affected groups with a particular concentration of a serum marker, we would need to take into account the fact that these levels change during pregnancy.
Why do we use median values when looking at serum marker data rather than mean values? How do we use median and median log 10 values in screening?
- Median values are less skewed by high values than mean values.
- In a situation where we have skewed data the median and the mean are different. However, if we log all the results what we end up with is an approximation of normally distributed data and the median and the mean are the same. We can then work with multiples of the median.
- If we take median log values say for the AFP values for a large population of women at different gestations then we can plot the median log 10 values of AFP agains gestational age and get a straight line that shows us the median AFP value for a population at any given time in gestation.
- For an individual women we can look at her own median value on such a graph and work out how many times bigger her log 10 median value is compared to the background population. This is called the Multiple of the Median.
- NSC definition: The serum marker concentration for a pregnant woman, divided by the median concentration value for unaffected pregnancies of the same gestational age.
What is the Multiple of the Median?
NSC definition: The serum marker concentration for a pregnant woman, divided by the median concentration value for unaffected pregnancies of the same gestational age.
What would the MoM value be is an individuals measured value is 30 and median AFP for this gestational age is 30?
If measured value is 30 and median AFP for this GA is 30 then for this pregnancy 1.0 MoM.
What would the MoM value be is an individuals measured value is 75 and median AFP for this gestational age is 30?
If measured value is 75 and median AFP for this GA is 30 then for this pregnancy 2.5 MoM.
How are AFP values known to be affected on Downs syndrome pregnancies? If we have a graph showing median AFP values in both unaffected and affected pregnancies how can we work out how much more likely an individual with a particular MoM value is to be affected?
- AFP values are known to be lower in T21 affected pregnancies.
- The reason behind this is unknown and the difference was discovered entirely by chance.
- If we have a graph showing AFP values in Downs syndrome pregnancies and AFP values in unaffected pregnancies, for any given MoM value we can work out how much more likely it is that they will be affected by measuring the heights of the unaffected and affected graph lines for that MoM value and calculating the difference in height between the two lines = likelihood ratio.
- This can be used in Bayes calculations where the odds of something happening equal the likelihood ratio time the prior odds of something happening:
- Posterior Odds = Prior Odds x Likelihood ratio.
- For this screening:
Posterior odds = Age related odds (prior) x Likelihood ratio.
- It is relatively straight forward then to do this with more markers to calculate more likelihood ratios to continue to use the prior chance of something happening x the age alone odds. We can work these out using a computer.
In Bayes what simple calculation allows you to calculate posterior odds?
Posterior Odds = Prior Odds x Likelihood ratio.
Recap of Downs syndrome markers and risk:
- There is risk related to age alone.
- Marker levels change with gestational age.
- Affected pregnancies have different marker levels to unaffected.
- Can combine age and marker levels to create a risk. The same principle applies if you want to use autosomal markers rater than serum biochemistry markers.
Does the MoM principle apply to ultrasound markers?
- MoM principle also applies to USS measure.
- On average, larger NT in affected.
What factors are known to affect DS screening marker levels? What do we need to take into account to ensure that our MoM measurements are as accurate as possible?
Things that affect marker level:
- We need accurate dating of the pregnancy in order to ensure that the MoM value calculations are correct.
- Weight - basically if the foetus is the same size in everybody but the mother’s body is twice the mass of the average sized woman then the markers will be diluted out in that woman’s body. Usually have a median weight for population and will apply a percentage correction change based on the size of the mother.
- Smoking.
- Ethnicity.
- IVF treatment cycles - some of the drugs used in IVF treatments can alter some marker levels.
In the case of Bayes calculations for Downs syndrome screening what factors might affect the prior odds?
Things that affect the prior odds:
- Previous Downs pregnancy (increases the chance that it is not just a spontaneous change causing Downs syndrome). Will manipulate the prior odds to take that into account.
