Trying to predict the future of ex-preterm infants: who benefits from a brain MRI at term?


  • Invited Commentary for Pearce and Baardsnes, Term MRI for small preterm babies: do parents really want to know and why has nobody asked them? pages 1013–1015.

Annie Janvier, M.D., Ph.D., Department of Pediatrics and Clinical Ethics, University of Montreal, 3175 Chemin Côte-Sainte-Catherine, Montreal, H3T 1C5 Quebec.
Neonatologist and Clinical Ethicist, Sainte-Justine Hospital, 3175 Chemin Côte-Sainte-Catherine, Montreal, H3T 1C5 Quebec.
Tel: +514 345 4931, ext 3305 |
Fax: +514 345 4822 |


Three dimension


Complete genome hybridization


Magnetic resonance imaging


Neonatal intensive care unit


Negative predictive value


Positive predictive value

inline image

inline image

In this issue of the journal, Jason Baardness and Rebecca Pearce (1) discuss their trying experience as the parents of 25 week twins. Lily died after 7 days of NICU stay, and Maren had a very complicated course. We took care of their two daughters. Like most parents, they wanted to be informed about the chances of survival of their children and their potential long-term consequences. We remember counselling them during their daughters’ stormy course. Unlike most parents, they have both scientific backgrounds and understand statistics very well. We remember telling them that medical publications only gave probabilities and we could not predict with much certainty what would happen to their daughter; Maren was there, and she was the one who would eventually answer with absolute precision all of their questions about her long-term outcome.

There are a number of potentially appropriate indications to perform investigations just before discharge which are designed to predict future neurological or developmental outcomes of preterm babies: (i) to focus follow-up programmes; (ii) to select babies for early intervention programmes; (iii) to do quality control; (iv) to prepare parents for the future; (v) and for research to better understand the effects of prematurity on the brain. Is the term equivalent age cerebral magnetic resonance imaging (MRI) an appropriate tool for these purposes? A normal term MRI has a high negative predictive value (NPV) for cerebral palsy, but a much lower NPV for developmental delay (2,3). An abnormal MRI is much more troublesome, although there may be a statistically significant association with adverse outcomes, the positive predictive value (PPV) of most findings, for either cerebral palsy or developmental delay, in numerous studies is below 50%, (2) and in some studies as low as 10% (4). With such a low PPV, how much does it really help to know the results of a term MRI for a specific baby?

The predictive values are really not adequate to deny follow-up to those with normal scans nor to select patients to participate in early intervention programmes (5,6). If the term MRI is being performed as a screening test, it does not satisfy the WHO criteria for such a test. Does changing the probabilities with the MRI really help parents prepare for the future? Instead of knowing – for a 25-week infant like Maren – before the test that your baby has a statistical 20% risk of ‘major’ disability and 35% of ‘minor’ disability, and a 50% chance of not being disabled, after the test you learn that your baby has either a 30% or a 65% risk of disability, depending on the findings. Do such adjustments in risk estimates help parents? Neonatologists have a great deal of data about the survival and long-term outcomes of vulnerable preterm infants. These data come essentially from population studies. Percentages may inform parents, but parents want to know what will happen to their child (or twins, triplets), and not to a previous hundred similar children. While we can inform parents about the range of likely outcomes for their child, we cannot inform them about what exactly will happen to their child, and the term MRI does not substantially improve that prediction.

As for research, term MRI could help to understand the causes of disability and the trajectory of development in the preterm and provide a basis for research to reduce disability or the impacts of disability. We think it is highly likely that the research that is being performed around the term MRI will eventually improve outcomes. We suggest that the only reasonable analysis of the published data is that the MRI can be useful for research, but is not adequately predictive for any of the clinical care indications listed above. As a research tool, the MRI should be being performed within defined research protocols and with parental informed consent.

The MRI provides very detailed pictures of brain structure; it allows calculation of cerebral volumes and the detection of minor abnormalities that may not otherwise be evident. They can produce spectacular information about tract mapping and how it appears to differ in those born very preterm. There are more and more ‘precision’ technologies in medicine. Many parties may benefit from this precision: healthcare institutions, companies selling technology, insurance companies, healthcare professionals, researchers… Many other investigations are part of this precision-based medicine armamentarium: CGH (complete genome hybridization), near infrared spectroscopy, genetic tests on foetal cells in maternal blood or cord blood, 3D pregnancy ultrasounds, genome sequencing, investigations predicting cancer risks, PET scans, autism screening questionnaires, etc. Patients are now represented with numerous tests. Our fascination with precision-based medicine and with these phenomenally powerful tools and numerous tests or technologies seems to have replaced our evaluation of the baby – as a person within a family – and not represented as a series of tests. What is the meaning of such investigation for families? Other than the performance, sensitivity, specificity, reliability of these tests and their predictive values, a significant and critical aspect of precision-based tests have not been investigated: what is the impact on patients, parents and/or families? The meaning of these tests for the primary stakeholders needs to be investigated.

Despite long conversations with this couple, we had omitted to counsel them about the term equivalent age cerebral MRI, to inform them about what answers it could give and questions it could raise. That it did not replace the neonatal follow-up that Maren would nevertheless be enrolled in. That it was different from the many eye examinations she had been subjected to every week for a while; the eye examinations were performed to screen for something treatable, which has significant consequences if not treated. After the term MRI, we remember receiving a long letter Jason wrote. Jason is a highly organized, meticulous, calm and quiet person. His letter was everything we had not seen about Jason. The letter included many of the concerns that you can read in their article.

Their letter affected us. Before reading it, we had been sceptical that the term MRI was very helpful for an individual baby; afterwards, we realized that it may actually be harmful. In our opinion, routine term MRI for preterm infants should be performed in the research setting with valid parental informed consent, consent which must clarify the uncertainties that will remain afterwards and allow parents the option to refuse without coercion.

At scientific meetings, we are frequently disconcerted by the paucity of studies actively investigating how to improve – as opposed to describe – the long-term outcomes of preterm infants or their parents coping abilities. Every year, more investigations are presented that generally describe how abnormal, abnormally wired, and not exactly perfect ex-preterms are. Rarely do the studies describe what ex-preterms can in fact do and how we can help them and their parents. Today, if we type ‘MRI preterm brain’ on PubMed, we can find 642 articles, a considerable amount of research is being performed and enormous research funds have been expended. But we are unaware of any of these studies that have investigated whether the technology helps parents and families of preterm infants. We have not found one that even raises the possibility that MRI results may in fact harm some families. We are grateful that Maren and Lily have parents who had the courage to write their story, so we can learn from it and re-evaluate our practice.

Conflicts of Interest

Annie Janvier and Keith Barrington do not have any conflicts of interests.