Kumanan Wilson, a scientist at Ottawa Hospital in Canada, had been playing around with data on the levels of metabolites in newborn babies in 2014. He had a hunch that these substances – byproducts of, or ingredients, for metabolic processes – could yield some valuable information.

There was no dearth of data from Canada and the US to work with – almost every baby born in these countries in the past 50 years had their heel pricked to draw a few drops of blood that would be tested for a range of diseases. Also part of the records was the gestational age – the number of weeks at which the baby is born, as determined by ultrasound examinations. It was easy to correlate the levels of metabolites with the gestational age. But there was nothing new about that – normal levels of metabolites for different gestational ages were already known. Abnormalities in their levels were pointers to diseases like hypothyroidism and sickle cell disease.

Then it occurred to Wilson that the reverse might be more challenging, and perhaps instructive. Was it possible to determine the gestational age of a baby from the levels of metabolites? And which of the hundreds of metabolites found in a blood was best suited for this purpose?

“If the reverse approach worked, it could be crucial to global efforts to improve newborn health,” said Wilson. Lives of lakhs of newborn children in developing countries like India could be saved.

Dr Kumanan Wilson (left) and his colleague Dr Pranesh Chakraborty (right) in their lab at Ottawa Hospital.

Making progress

Every year nearly 3.5 million babies are born prematurely in India, by far the highest among all countries. The numbers are growing rapidly because of the increase in hypertension and diabetes among women, and the spread of IVF treatments. These babies are underdeveloped and at risk of infection. Of them more than 3 lakh die.

Spotty healthcare is partly to blame, but even when suitable facilities do exist, doctors face another problem. In most cases, especially in rural India, there is no way of establishing the babies’ gestational age. Few women have ultrasound examinations, and their recollection of their last menstrual period can be unreliable. So it’s impossible to tell whether a baby is just low birth weight or preterm. As a result of this, it’s hard to determine treatment protocols – which for preterm babies are specialised, involving antibiotics, and steroid injections to strengthen the organs.

To find the reverse relation between gestational ages and metabolites, Wilson and his colleague Steven Hawken decided to work with data for 1,59,215 babies born in Ontario between January 2012 and December 2014. The gestational age of these babies was known.

The two scientists first used the traditional method of determining gestational age using birth weight, sex and the order a child is born in their family (first, second and so on) to see how the results tallied with the known age.

A preterm baby. Photo courtesy: Kumanan Wilson.

Next they added haemoglobin levels to their algorithm. Oxygen levels in the womb are much lower than those outside, so the foetus has a different kind of haemoglobin known as foetal haemoglobin. Its levels change as the foetus grows. After birth the levels of foetal haemoglobin starts reducing, being replaced gradually by adult haemoglobin, till at six months after birth, when foetal haemoglobin disappears. Foetal haemoglobin levels measured at birth can therefore be a good indicator of gestational age.

The scientists then went further by including the levels of two hormones, thyroid-stimulating hormone and a variant of progesterone. Finally, in the last model, they threw in a slew of other metabolites like fatty acids, amino acids and enzymes.

The results were astonishing. As more metabolites were included, the accuracy of the algorithm increased. The biggest difference, according to Wilson, “where it’s of greatest benefit”, was in the case of preterm babies. Here the traditional method stumbled, but using the full complement of metabolites, the scientists were able to determine the gestational age (with a two-week margin of error) in more than 95% of the cases.

Field tests

Being able to tell whether a baby was born between 32 and 34 weeks or whether it was born at 37 weeks with low birth weight made the difference between life and death.

Even the accuracy of the model that included only foetal haemoglobin was 91%. The advantage with foetal haemoglobin, says Wilson, is that it can be taken from umbilical cord blood upon birth, instead of having to wait 48 hours – metabolites in a newborn fluctuate rapidly, stabilising only after this period – to take a blood sample. It’s also cheaper, requires fewer machines, and spares the nervous mother a pinprick to her newborn.

For this scientist of Sri Lankan descent whose family had migrated to Canada when he was three, the results were particularly gratifying. Analysing metabolites in newborns would also “have the surrogate benefit of establishing a protocol for reporting back clinically significant results” and detecting conditions like congenital hypothyroidism and sickle cell anaemia.

A heel prick test. Photo credit: USAF Photographic Archives/Wikimedia Commons [Public Domain].

While Wilson and his colleagues were working in Canada, another set of scientists working independently at the University of California and the University of Iowa had made similar discoveries. The papers were published simultaneously in 2017, getting them funding from the Bill and Melinda Gates Foundation to test their applicability in developing countries. This was imperative given the vast environmental and racial, ethnic differences between the populations where the work was done and where it could be of greatest use.

Wilson collaborated with a team at Stanford University to run field trials, tracking 1,661 pregnant women in Bangladesh and 420 in Zambia. The women were made to have prenatal ultrasounds to accurately determine the age of their foetuses. After birth, the blood sample taken from their babies were sent to Wilson’s lab in Canada, where the metabolites were analysed and the results put through the algorithm to see how they tallied with the known gestational age of the babies.

The papers detailing the results are going to be published soon, but the results “are very good”, says Wilson. So good that the team is moving ahead to a large-scale implementation of the method – the sites are going to be decided in a few months, but the team is going to focus on South Asia and Sub-Saharan Africa.

Research in India

This work coincides with a new Indian research with emphasis on preterm birth in India. An ambitious project initiated by the Science Ministry’s Department of Biotechnology in 2014 is now delivering its initial findings. While this project is studying a number of factors, for the most part, they are addressing the problem from the opposite end – trying to predict which women will give birth preterm.

Eight thousand women in early pregnancy are enrolled in this study, which follows them through childbirth to six weeks after that. Bhabatosh Das of the Translational Health Science and Technology Institute has been looking at the changing composition of the bacteria in the vagina of these women, while scientists at other institutions are exploring changes in the mother’s genome and protein levels.

According to Das, there are significant changes in the population of bacteria in the body, or the microbiome, as a pregnancy progresses. But what he is on the lookout for are differences in the microbiomes of women who give birth at term and those who give birth prematurely. He also wants to investigate whether there are differences in the way the microbiomes of these two groups change through pregnancy. Initial results, he says, do indicate compositional differences, but he can only be certain once he gets more preterm babies.

Identifying preterm babies is, however, just the first step in addressing the problem of premature birth – better systems still have to be created to save these children.

But it’s an indispensable step – to which Wilson has offered the first implementable solution.