Heel-prick test for prematurity

26 April 2019

Interview with 

Kumanan Wilson, Ottawa Hospital

NEWBORN-FEET

A young baby

Share

But first, to childbirth. The average human baby takes 40 weeks to develop from the time of conception to the time of birth. And in developed countries, a dating ultrasound scan, included as part of the antenatal screening process, establishes when the baby’s due. So if an infant arrives early, doctors can tell and look out for appropriate problems accordingly. But in poorer countries, this sort of information may not be available. And under these circumstances, if a baby’s smaller than it should be, is that because it’s premature, or is something else wrong? Previously we had no idea. Now, speaking with Chris Smith, Kumanan Wilson, from Ottawa Hospital, describes his possible solution…

Kumanan - We've developed a blood test that can be done on an infant soon after birth, that can tell us whether that child is premature or not. Premature births are one of the leading causes of death and disability in children around the world, and this is particularly a problem in low resource settings. Expectant mothers don't have access to the type of prenatal care we have in more developed settings, and they don't have access to prenatal ultrasound in particular. So when a child is born, we don't know whether that child is born too soon, or premature, or born on time.

Chris - So how did you actually do the study?

Kumanan - This study is a validation or a further test of something we developed on children born in Ontario. Our first study developed a mathematical model that could look at blood from a prick of the heel of a newborn infant, and could tell you how premature that baby is.

Chris - So when you take the heel prick sample, what do you look for in the blood in order to make that determination of whether that baby was born on time or not?

Kumanan - This blood sample we obtained from the child is what we do in a routine newborn screening program. And in that program we test for a variety of illnesses, such as cystic fibrosis, or thyroid disease, or things like sickle cell disease. For the purposes of the newborn screening program we look at things like amino acids, fatty acids, haemoglobin levels. We've known from past studies that these vary based on the gestational age of a child. What we were interested in finding out was, if we had the patterns of those chemicals, could we do the reverse, could we determine the gestational age of a child? And that's why we were able to successfully do in our Ontario sample of children.

Chris - This is what researchers dub "metabolomics", isn't it? Where we take a combination of markers, and we say, "when you see these markers present at these levels, this can be used to make various predictions".

Kumanan - That's correct. This is a form of metabolomics. In this case we're using a convenient sample of chemicals that we obtain from our newborn screening program. It is possible we can make this test better if we expand and look for chemicals that we aren't routinely collecting.

Chris - Now when you subject the samples to your analysis, how accurate is it? In other words, if you take your gold standard infants that have been ultrasound screened, do you know what their gestational age is when you take the sample? How out is your prediction versus reality?

Kumanan - We are able to predict the gestational age within one week. However the reality is, in the vast majority of our sample, the children were born at term, and birth weight alone would have been adequate to determine the gestational age of those infants. Where the added value of the metabolomics is in children who are small. In children who are born small, there is the risk of misclassifying them as premature, when they may be small for other reasons. For example, they may be sick.

Chris - So this is against a gold standard population of individuals born in a rich country, first world setting. Have you validated it against individuals who were born in the target territory? So third world settings, or resource poor settings. And also, might the levels have a degree less of confidence when you take into account ethnicity?

Kumanan - These are great questions. The first study we did after we developed our test in Ontario was to use our immigrant database to look at different ethnic groups. And we found that within our Ontario cohort, the test performed well amongst the various ethnic groups. The study that was published in the eLife journal then further validated this in a cohort in Bangladesh. This is a cohort that is part of an ongoing study. We were fortunate because they had dating ultrasounds done on all of the infants. And we found that the test we developed in Ontario performed very well in the infants born in Bangladesh. That increases our confidence that this approach can be used in multiple settings, in multiple low-resource settings, and can help us determine both the burden of preterm births, as well as help us evaluate interventions that are designed to reduce preterm birth.

Chris - And is that where you're going next with this then?

Kumanan - The next step is a partnership we have embarked upon with Stanford University to expand this approach to multiple lower-resource settings. In Zambia, Zimbabwe, Kenya, and another site in Bangladesh.

Chris - And to throw you a bit of a curve ball, are there any situations, or are there any diseases, acquired conditions, inborn conditions as well, which can throw your system? So in other words you get data that look promising for the individual, but actually you're being misled?

Kumanan - This is actually one of the questions we want to answer as we move forward. We do wonder if a stressed sick child at birth, that can distort their pattern of chemicals, and that might affect our assessment of gestational age. So that's a very important question for us to answer. We hope to explore that as we examine the samples from these multiple sites.

Comments

Add a comment