Discussion
In this national study comprising preterm infants born at GA <24 weeks, the rate of ROP, ROP treatment, retreatment and unfavourable retinal outcomes were substantial. Of 399 infants born at GA <24 weeks, 91.5% developed any stage ROP and 65.8% of those developed severe ROP (stage 3 or more). Altogether 43.3% and 72.2% of those with severe ROP underwent ROP treatment. In 20 eyes of 13 infants (3.2%) ROP progressed to retinal detachment despite repeated treatment attempts. APROP was diagnosed in 22 (5.5%) infants and progressed to retinal detachment in six infants. Two infants developed unilateral endophthalmitis and phthisis after anti-VEGF injections.
Previous studies on ROP have included a limited number of infants born at GA <24 weeks. Due to different policies regarding active perinatal care and ROP treatment at these low GAs over time and between hospitals and countries, comparisons with other cohorts of extremely preterm infants must be made with caution. In a previous Swedish population-based study during 2004–2007, 53/58 (91.4%) infants with GA <24 weeks at birth developed some stage of ROP and 27/58 (46.5%) underwent treatment.6 Ishii et al reported that of Japanese infants born at GA <24 weeks in 2003–2005 (n=320), 27.5% required ROP treatment, while Miller et al reported in a much smaller group of patients (n=23) that the cumulative probability of receiving laser therapy was nearly 46.0% if born at GA <24 weeks in the USA, 2006–2008.13 15 In a study from England in 2006 (the EPICure study), Costeloe et al12 reported that 31.9% of infants born at GA <24 weeks received laser treatment for ROP (n=69). In the current study from Sweden from 2007 to 2018, 43.3% of the most immature infants were treated. The proportion of screened infants who were treated yearly varied during the study period, ranging from 20.8% of infants born in 2011 to 59.3% of infants born in 2009. We found no increasing in incidence of ROP treatment during the study period; however, the number of infants screened increased over time, which indicate increased survival rates but not decreased morbidity rates. We suspect that the impact of the new routines with increased oxygen saturation target levels from 88%–92% to 91%–95% that were implemented in most parts of Sweden during 2014 have probably affected treatment need. Holmström et al reported a nearly doubled increase in incidence of treatment for ROP in 2015 in one Swedish healthcare region, as compared to the years before the new oxygen routines (2008–2013), while the incidence remained stable in one region which did not implement the new oxygen routines.21
Survival rates may be influenced by many factors such as an obstetrician’s willingness to intervene to rescue the fetus and the neonatologist’s policy regarding initiation of neonatal intensive care as well as the quality of neonatal care for very immature infants.1 2 All these variables influence the health of the surviving infants. We found an increased number of infants screened and treated for ROP, and a trend of higher rates of retreatment in the later years of the study period that may be due to the increase in anti-VEGF treatment, which is known to have a high risk of recurrence. Thus, there has been an increasing workload for ophthalmologists performing ROP screening and treatment in Sweden in recent years. This is worrying, since there is a general shortage of ophthalmologists willing to perform ROP screening in this and many other countries.
In Sweden, the first intravitreal injection of anti-VEGF for ROP was performed in 2010. In 2015, anti-VEGF treatment became more frequently used as first choice of treatment for APROP and central ROP. In our cohort, the retreatment rate was 32.4% after laser and 69.0% after anti-VEGF, which to our knowledge has previously not been reported exclusively in infants with GA <24 weeks. Lower recurrence rates have been reported in cohorts comprising infants born with a wider range of GA.22 It is well known that recurrence is more common after anti-VEGF injections than after laser treatment.23–25 In the Rainbow trial, which formed the basis for approving the anti-VEGF drug ranibizumab for ROP, treatment recurrence rates were 18.9% after laser treatment and 31.1% after ranibizumab treatment.11 The high recurrence rate after anti-VEGF in our study may be partly due to the high percentage (5.5%) of infants with APROP, the most aggressive form of ROP with high risk for recurrence after treatment.26 27 In the current study, the median time to retreatment after first anti-VEGF injection was 8.3 weeks compared with 2.2 weeks after laser therapy. In the Rainbow trial, the time to recurrence after the first anti-VEGF treatment varied between 4.1 and 18.2 weeks (median 8.0 weeks).11 In our cohort, one infant was found to have peripheral vasoproliferation 73.3 weeks after ant-VEGF injection. Long-term follow-up is a logistic clinical problem, and there are no general guidelines for how long infants need to be followed after anti-VEGF injections.
Rate of recurrence after the first treatment varied among the seven centres where treatment was performed. Only two centres used anti-VEGF as first treatment in more than two patients and the proportion of retreated patients at those centres were 8/15 (53.3%) and 10/11 (90.9%), respectively. The retreatment rate after the first laser treatment varied from 7.1% to 63.0% among centres. In a recent national study, it was found that in 11/17 preterm infants who had become visually impaired, the ROP screening and/or treatment process had been suboptimal.7 In fact, in 10 of 17 visually impaired infants, the first laser treatment was considered suboptimal or untimely. These results are in line with the newly published study by Spandau et al,28 who demonstrated that treatment failure of type 1 ROP was due to inadequate laser treatment in 8/10 cases, for example, undertreatment, overtreatment or skip lesions. Hence, correct laser treatment is crucial and might be facilitated by using wide-angle photography after treatment, before finalising the procedure to identify areas of undertreatment or skip lesions (Gränse et al, in manuscript). All aspects of optimal screening including the ophthalmologist’s judgement of ROP staging, the presence of plus disease, APROP and need for treatment are essential to ensure timely treatment and ensuring the best possible outcome.29 30
In our cohort, 20 eyes of 13 infants had unfavourable retinal outcomes. These thirteen infants developed retinal detachment in one or both eyes, and two of those infants also developed endophthalmitis after anti-VEGF injection in one eye each. We reviewed these infants’ medical records and confirmed, in a majority of cases, the presence of treatment failures and/or incomplete follow-up in accordance with the findings reported by Norman et al.7
In this vulnerable group of infants, infants were subjected to a median of 14.0 ROP examinations (range 3–42). During the study period, a total of 6061 ROP examinations were performed. Altogether, 261 ROP treatments were performed in 173 infants. The ROP examinations are known to be stressful and painful for the infant. During and after ROP examination, fluctuant blood pressure, increased pulse rate, desaturation and increased need for oxygen supplementation occur.31 32 Laser therapy is regularly performed under general anaesthesia in Sweden, and most centres use general anaesthesia also for anti-VEGF injections. Infants undergoing laser treatment have been found to be at high risk of intraoperative and postoperative adverse events like hypotension, bradycardia and apnoea.33 Repeated anaesthetic procedures may affect the infants’ neurodevelopment, and there are concerns about the long-term effects of anti-VEGF treatment on neurodevelopmental outcome.34 Thus, it is crucial that the ROP screening and treatment procedures are as efficient and gentle as possible, especially as several authors have stressed that ROP is a biomarker for brain volumes at term and later neurodevelopmental outcomes.35 36
Strengths and limitations
The study has a retrospective design, which is a limitation. The strengths are: prospectively collected validated data with national coverage for more than 10 years, follow-up of infants when moved between hospitals, patient register structured protocols (SWEDROP) and file review of all neonatal diagnoses.
