Clinical biomarkers: Pre-meeting workshop Meeting for NIH/NICHD.
Summary of meeting (S Tong - Chair), held 5th April 2021.
Scope: "Looking at clinical biomarkers for use in pregnancy with a focus on detecting and distinguishing placental pathologies from population variance, where are we currently, and what needs to happen next?"
Themes arising from the meeting can perhaps be best grouped into 'challenges/considerations' and 'concepts in developing clinical biomarkers'.
1. Focus on clinical utility
- Let clinical need inform and guide development of new biomarker tests
- It was noted that researchers should be wary of chasing molecules merely because they have an association with a disease. The test being developed must have a vision of being clinically useful from the outset.
2. Population variations is a challenge to keep in mind
- Obesity, ethnicity, sex and regional differences can affect the performance of a biomarker test when applied to different populations.
- These should be taken into account when developing a biomarker test for clinical use.
- Validation of a biomarker test is absolutely critical
- Ideally, it is validated in different populations, given point 2.
o A good example is the Fetal Medicine Foundation first trimester test for preeclampsia developed by Professor Nicolaides and colleagues (King's College). It is currently offered in the UK, and other countries too (South Africa, Australia, Spain…). It is now being validated by Prof Liona Poon in Hong Kong/Asia. She is finding some differences in performance which she is attempting to overcome by reworking reference ranges of the test.
- Validation needs to be done in very large numbers, especially to convincingly overcome any questions about different population variances.
4. Tests can be – or ideally – multi-modal
- They could combine clinical information, biomarker levels in the circulation, and imaging findings (principally ultrasound).
- The FMF first trimester test does this.
5. Tests should be developed where specific test performance characteristic thresholds that would make it clinically useful should be defined.
- E.g. It should be decided by clinicians what the sensitivity (at set specificity or screen positive thresholds) would represent a likely clinically useful test. And the positive predictive values.
- Thresholds will likely be different for different diseases, the gestation when the test is being applied and the likely management options.
o E.g. FMF first trimester test has a positive predictive value of around 15% for preterm preeclampsia but this is acceptable given the baseline incidence of the diseases is <1%.
- In contrast, such test performance would be too low to screen for small for gestational age fetuses (where there is an increased stillbirth risk) as the baseline incidence is already 10%.
CONCEPTS IN DEVELOPING BIOMARKERS
1. Consider 'epochs' across pregnancy to develop biomarker test – 12, 20 and 36 wks gestation.
- Prof Nicolaides/Prof Poon sensibly advocate 12, 20 and 36 weeks as important milestones during pregnancy (and pregnancy care) when circulating biomarker test for diseases could be developed.
- At these gestations, ultrasounds are often done. and tests could be integrated with the ultrasound results.
- It is pragmatic as a 12 week gestation is able to pick early onset preeclampsia but not late onset (and is useful in triaging those who should be offered aspirin). Then doing further screening tests at 20 and 36 weeks gives biomarkers a better chance at performing well in screening diseases of later pregnancy, as biomarkers will generally perform better if done closer to disease onset (ie developing a 12 week test to predict diseases happening at the end of pregnancy is unlikely to succeed).
- A 36 week test is sensible as the disease incidence of fetal growth restriction and even preeclampsia rise acutely during the final weeks of pregnancy.
2. First trimester screening test for gestational diabetes.
- This provides an opportunity for earlier intervention, compared to waiting 26-28 weeks before diagnosing the condition where in fact the disease had probably been brewing for months.
3. Biomarker screening test to identify those at increased risk of diseases in later life.
- While challenging, it should be noted that this opportunity exists. It could be important given a lot of adult diseases may have some of their genesis arise while the person is still in utero (DOHAD).
4. 36 week biomarker test for placental insufficiency
- There are a few groups trying to develop such a test
- Its value is that it can identify fetuses' that may be at increased risk of stillbirth because of placental insufficiency (the incidence rises across late pregnancy).
- There is a clinical need – ultrasound and clinical examination fall short in identifying babies that are small because of placental insufficiency.
- Also, for fetuses identified as small, an alternative clinical management decision exists – time birth by inducing labor. This may save some babies from stillbirth.
5. A bold new initiative of a new major cohort collection.
- The group was nearly unanimous in proposing a major initiative of a multi-centre cohort collection.
- A VERY LARGE collection, a "pregnant moonshot".
- There there is deep phenotyping, agreed approaches to define diseases, agreed protocols for sample collection etc.
- It would be a multi-pronged collection (genetics, imaging, good clinical phenotyping, and the collection of samples for all manner of 'omics)
- It would need to be a very large population
- The samples could be shared across many researchers to hunt for new biomarkers
- It would require a very clever lead team. So that it is highly effective, and not 'just another cohort collection' that falls short of producing significant clinical impact.