Mank E, Sáenz de Pipaón M, Lapillonne A, Carnielli VP, Senterre T, Shamir R, van Toledo L, van Goudoever JB; FIT-04 Study Group. Efficacy and Safety of Enteral Recombinant Human Insulin in Preterm Infants: A Randomized Clinical Trial. JAMA Pediatr. 2022 May 1;176(5):452-460. doi: 10.1001/jamapediatrics.2022.0020. PMID: 35226099; PMCID: PMC8886453.
Dr. Ikhwan Halibullah
Neonatal Fellow, The Royal Women’s Hospital
Prof. Peter Davis
Neonatologist, The Royal Women’s Hospital
Professor, Department of Obstetrics and Gynaecology, The University of Melbourne
A/Prof. Brett Manley
Neonatologist, The Royal Women’s Hospital
Associate Professor, Department of Obstetrics and Gynaecology, The University of Melbourne
TYPE OF INVESTIGATION
In premature infants born between 26 weeks and 0 days and 32 weeks and 0 days of gestation and with a birth weight of 500 grams or more (Population), does the addition of recombinant enteral insulin at a low dose (400 μIU/mL milk) or high dose (2000 μIU/mL milk) (Intervention), compared with placebo (Comparison), reduce the time to full enteral feeding (Outcome)?
- Design: Prospective, randomised, multicentre
- Allocation: Computerised random allocation in ratio of 1:1:1, blocked and stratified by centre and gestational age at birth (26-28 weeks’, 29-32 weeks’)
- Blinding: The trial sponsor, parents and medical caregivers were blinded.
- Follow-up period: Until 3 months’ corrected age.
- Setting: 46 neonatal intensive care units throughout Europe, Israel, and the United States of America from October 9, 2016, to April 25, 2018.
- Gestational age 26 to 32 completed weeks at birth
- Birth weight ³ 500 grams
- Clinically stable and able to tolerate enteral feeding
- Expected to wean off parenteral nutrition during hospital stay
- Major congenital malformations
- Intrauterine growth restriction
- Chest compressions or resuscitation drugs given directly after birth
- Fraction of inspired oxygen >0.60
- Suspected infections
- Confirmed necrotising enterocolitis
- Receiving systemic insulin infusion
- Maternal diabetes requiring insulin
- Nil orally
- Already established on full enteral feeds
- Intervention: The investigational product contained either 400 μIU/mL milk of recombinant human (rh) insulin (low-dose) or 2000 μIU/mL milk of rh insulin (high-dose). Placebo was manufactured identically but without the inclusion of insulin. The study drug was added to the infant’s milk daily at standard dosage from day 5 postpartum and continued for 28 days or until a unit or hospital transfer was required.
- Primary outcome: Time to achieve full enteral feeds (FEF), defined as an enteral intake of ³150 mL/kg per day for 3 consecutive days.
- Secondary outcomes:
- Number and percentage of infants reaching FEF within 6, 8, and 10 days of intervention
- Time to achieve an enteral intake of ³120 mL/kg per day for 3 consecutive days
- Number of days receiving parenteral nutrition
- Growth velocity (weight, body length, and head circumference)
- Weight on study day 28
- Weight z-score on study day 28
- Change in weight z-score from birth to day 28
- Safety outcomes (until time of discharge)
- Serious adverse events including necrotising enterocolitis, clinically or culture-proven sepsis and mortality.
- Blood glucose tests were performed twice daily in the first 4 study days and thereafter on alternate days between 7 AM and 10 AM before enteral feeding.
- Insulin antibody serologic assessment at:
- Study day 28, or
- Day of hospital transfer, or
- Day of transfer to another unit within the hospital, or
- Day of discharge to home, AND
- At 3 months’ corrected age
- Analysis and Sample Size:
- To achieve a clinically important reduction in time to FEF of 1.4 days, with an alpha level of 0.05 and a power of 80%, 115 infants were required in each arm of the study. To account for twins and drop-out, planned recruitment size was 460 infants total. Analysis was by intention-to-treat.
- The study design included the conduct of a futility analysis by a third-party contracted research organisation as part of the safety analysis following randomisation of the first 225 participants. A conditional power of 35% was pre-specified, and if below pre-specified threshold of 35%, the study would be deemed futile and ceased.
- Patient follow-up:
- 110 infants were randomised to the low-dose group, 95 to the high-dose group and 98 to the placebo.
- The data and safety monitoring board recommended early cessation of the trial based on an unfavourable futility analysis. At that time, 303 infants had been randomised prior to cessation, and these infants were analysed within their randomised groups. However, only 261/303 (86%) reached the primary endpoint, due to cessation of sponsorship of trial (17/42, 40%), hospital transfer (15/42, 36%) and unspecified ‘other’ reasons (10/42, 24%).
Infant characteristics were similar across the three groups including demographics, perinatal factors, condition at birth, age at recruitment and initiation of feeds and type of feeds. Of the total sample, 177/303 (58%) received mixed feeding i.e., a combination of different types including mother’s own milk, donor human milk and preterm formula, while 96/303 (32%) received mother’s own milk exclusively, 9/303 (3%) donor human milk and 16/303 (5%) preterm formula.
For the primary outcome, the median (IQR) time to FEF was shorter in both intervention groups, 10.0 (7.0 – 21.8) days in the low-dose group and 10.0 (6.0 – 15.0) days in the high-dose group, compared with 14.0 (8.0 – 28.0) days in the placebo group. This result was statistically significant with a mean difference of time to FEF, between the intervention and placebo groups of 4.0 (95% CI 1.0 – 8.0) days (p = 0.03) for the low-dose group and 4.0 (95% CI 1.0 – 7.0) days (p = 0.001) for the high-dose group.
For the secondary outcomes, significantly more infants achieved FEF by study days 6, 8 and 10 for both intervention groups compared with placebo. The time to achieve ³120 ml/kg/day for 3 consecutive days was also significantly lower in both intervention groups than in the placebo group. The median number of days of parenteral nutrition was significantly lower in the high-dose group vs. placebo, but not in the low-dose group vs. placebo. There were no differences in weight gain, weight on study day 28, weight z-score on day 28, change in weight z-score, head circumference or body length.
Safety outcomes were only reported in the 293 infants who were included in the interim safety analysis. The proportion of infants with one or more serious adverse events was 16/108 (15%) in the low-dose group and 11/88 (13%) in the high-dose group, compared to 19/97 (20%) in the placebo group. No infants developed insulin antibodies. Statistical tests were not reported for these safety outcomes.
The study authors concluded that the enteral administration of rh insulin at two different dosages was safe and that it significantly reduced the time to achieve FEF in preterm infants born at 26-32 weeks’ completed gestation. Based on this, the authors support the use of rh insulin as a supplement to human milk and preterm formula.
Feed establishment is challenging for many preterm infants. While interventions such as probiotics have reduced the incidence of necrotising enterocolitis (1), no intervention has previously been shown to reduce feed intolerance effectively and safely in preterm infants. Erythromycin (2) and lactase (3) are ineffective and cisapride is associated with adverse effects (4).
It is encouraging that this study of a novel therapy reports promising results. Comparing two doses of recombinant human (rh) insulin supplementation with placebo, the authors report a statistically significant reduction in the time to achieve full enteral feeds. The size of this effect, a mean difference of 4.0 days (p = 0.03 [CI 1.0 – 8.0]), seems clinically important.
Despite the encouraging outcome, it is important to view these results within the context of a trial that was stopped early following a planned interim futility analysis. The investigators had prespecified that a conditional power of 35% was required for study continuation after the first 225 enrolments. As this futility criterion was satisfied, 303 infants were enrolled, short of the 345 specified in the trial protocol.
There was a slight imbalance in the numbers allocated to the three groups: 110 were randomised to the low-dose, 95 to high-dose, and 98 to placebo. Similarly, there were imbalances in allocation within countries, important due to potential variation in feeding practices. Dropout was substantial, with only 86% of randomised infants reaching the primary endpoint. The most common reason was the cessation of sponsorship. While the statistical plan was published as supplementary material, information on missing data handling has been redacted, leaving some questions about data integrity.
The interim futility analysis resulted in a misestimation of the eventual outcomes. There may be merits of stopping a trial early for futility, particularly to save resources (5). However, early termination reduced the precision of this study’s estimates of effectiveness.
The ethical implications of early trial cessation need to be carefully considered. In this study, some infants already exposed to the intervention had treatment termination without outcome assessment. In the absence of any indication of harm, the mid-treatment involuntary removal of participants, who have accepted the risks of being subjected to an investigational therapy, seems unfair.
The author’s conclusion that product safety has been demonstrated needs to be interpreted with caution, particularly as the product manufacturer funded the trial and contributed to the design and conduct of the study. Of the 293 infants included in the safety analysis, only 196 were exposed to the treatment. As the event rate of some serious events such as necrotising enterocolitis can be as low as 3% (6), this relatively small sample may not detect rare complications. Further, blood glucose levels were only measured at most twice a day prior to feeds. The potential of treatment-induced hypoglycaemia, even if brief, may have therefore not been adequately studied.
These positive results from a well-written report of a multi-centre trial are appealing. However, prudent clinicians should cautiously analyse its conduct and conclusions, before introducing a new therapy to their vulnerable patients.
Sharif S, Meader N, Oddie SJ, Rojas-Reyes MX, McGuire W. Probiotics to prevent necrotising enterocolitis in very preterm or very low birth weight infants. Cochrane Database Syst Rev. 2020;10(10):Cd005496.
Ng E, Shah VS. Erythromycin for the prevention and treatment of feeding intolerance in preterm infants. Cochrane Database Syst Rev. 2008(3):Cd001815.
Tan-Dy CR, Ohlsson A. Lactase treated feeds to promote growth and feeding tolerance in preterm infants. Cochrane Database Syst Rev. 2005(2):Cd004591.
Kohl M, Wuerdemann I, Clemen J, Iven H, Katalinic A, Moeller JC. Cisapride may improve feeding tolerance of preterm infants: a randomized placebo-controlled trial. Biol Neonate. 2005;88(4):270-5.
Schoenfeld DA. Pro/con clinical debate: It is acceptable to stop large multicentre randomized controlled trials at interim analysis for futility. Pro: Futility stopping can speed up the development of effective treatments. Crit Care. 2005;9(1):34-6.
Alsaied A, Islam N, Thalib L. Global incidence of Necrotizing Enterocolitis: a systematic review and Meta-analysis. BMC Pediatrics. 2020;20(1):344.
CONFLICTS OF INTEREST