Article Text

Download PDFPDF

Treatment of type I ROP with intravitreal bevacizumab or laser photocoagulation according to retinal zone
  1. B Mueller1,
  2. D J Salchow1,
  3. E Waffenschmidt1,
  4. A M Joussen1,
  5. G Schmalisch2,
  6. Ch Czernik2,
  7. Ch Bührer2,
  8. K U Schunk3,
  9. H J Girschick3,
  10. S Winterhalter1
  1. 1Department of Ophthalmology, Charité Universitätsmedizin Berlin, Berlin, Germany
  2. 2Department of Neonatology, Charité Universitätsmedizin Berlin, Berlin, Germany
  3. 3Department of Neonatology, Vivantes, Klinikum Am Friedrichshain, Berlin, Germany
  1. Correspondence to Dr Bert Mueller, Department of Ophthalmology, Charité Campus Virchow Klinikum, Augustenburger Platz 1, Berlin 13353, Germany; Bert.mueller{at}


Aims To investigate the outcome of intravitreal bevacizumab (IVB) compared with laser photocoagulation in type I retinopathy of prematurity (ROP).

Methods Case records of 54 consecutive very low birth weight (VLBW) infants with type I ROP (posterior ROP, n=33; peripheral zone II, n=21) who were treated either with IVB (n=37) or laser photocoagulation (n=17) between 2011 and 2015 were retrospectively evaluated.

Results Patients with posterior ROP displayed significantly faster regression of active ROP within 12 days (range 9–15 days) if treated with IVB compared with laser photocoagulation, where active ROP regressed within 57 days (range 28–63 days) (p>0.001). No difference was observed in peripheral zone II.

Five of seven patients (12%) who developed a recurrence in both eyes after IVB required additional laser photocoagulation within a mean of 12.7 weeks (11.3–15.6 weeks) after the previous treatment. After laser photocoagulation one patient with posterior ROP developed macular dragging and another patient developed a temporary exudative retinal detachment in both eyes. 12 months after treatment the spherical equivalent was not statistically significant different between IVB and laser photocoagulation in posterior ROP patients. However, IVB lead to a significant lower spherical equivalent in infants with posterior ROP (+0.37 dioptres, range −0.5 to +1.88 dioptres) compared with peripheral zone II (+3.0 dioptres range +2.0 to +4.0 dioptres, p<0.001).

Conclusions IVB leads to faster regression of active ROP in infants with posterior ROP compared with laser photocoagulation. Spherical equivalent after 12 months was comparable in those treated with IVB and laser photocoagulation, but it was significantly lower in posterior ROP than in peripheral zone II.

  • Retina
  • Neovascularisation
  • Child health (paediatrics)
  • Optics and Refraction

Statistics from


Retinopathy of prematurity (ROP) is one of the most common causes of visual disability and blindness in children worldwide.1

ROP is a vasoproliferative disorder associated with premature birth, which type I may require treatment to prevent macular dragging and tractional retinal detachment.2–4 Vascular endothelial growth factor (VEGF) dysregulation is a the key mechanism in the pathogenesis of ROP and a target for current treatment options using either laser photocoagulation of the peripheral avascular retina or intravitreal anti-VEGF-antibody treatment.5

In contrast to laser photocoagulation, anti-VEGF antibody blockade allows the development of further retinal vascularisation while inducing regression of vitreoretinal proliferations.6

The Bevacizumab Eliminates the Angiogenic Threat of Retinopathy of Prematurity (BEAT-ROP) trial showed that intravitreal bevacizumab (IVB) (a monoclonal anti-VEGF antibody) is superior to laser photocoagulation in zone I disease in terms of recurrence rate of ROP and the incidence of unfavourable outcome, defined as macular dragging and tractional retinal detachment. The best treatment for zone II ROP is still under debate with regard to retinal morphology and vascularisation, refraction, visual function and systemic side effects.4 ,7–9

The follow-up study of the BEAT-ROP trial so far demonstrates a significant lower incidence of high myopia in eyes with zone I ROP if treated with bevacizumab compared with laser, 2½ years after trial closure.7

Here we report our experiences in the treatment of type I ROP (ROP stage III with plus) with regard to both treatment modalities and zone involved, the immediate treatment effect and morphological, functional and refractive outcomes after 12–15 months.

Materials and methods

Patients characteristics

Fifty-four very low birthweight (VLBW) infants (defined as birth weight <1500 g) with type I ROP who received either IVB (n=37) or laser photocoagulation (n=17) between June 2011 and January 2015 were included in this retrospective study.

All parents provided written informed consent before ROP treatment. The Institutional Data Safety Committee of the Charité-Universitätsmedizin Berlin approved this study.

Definition of severity of ROP

ROP was defined according to the International Committee for Classification of ROP protocol.10 In addition to these criteria, zone II was clinically subdivided into posterior zone II (the area near the border to zone I displaying a notch within the horizontal raphe and the temporal vascular arcades) and peripheral zone II (the more peripheral rest of zone II).

According to the Early Treatment for Retinopathy of Prematurity (ETROP) protocol, treatment was indicated in patients with type I ROP (zone I and posterior zone II: progressing extraretinal proliferation of at least 3–4 clock hours with plus disease; peripheral zone II: stage III ROP in five continuous or eight cumulative clock hours with plus disease).3 ,11

The distribution of type I ROP with respect to retinal zone involved was zone I (n=4), posterior zone II (n=29) and peripheral zone II (n=21). Patients with zone I ROP were pooled together with posterior zone II ROP due to their low number and this group is referred to as ‘posterior ROP’.


Preterm infants with type I ROP in zone I and posterior zone II were preferably treated with IVB following the BEAT-ROP protocol (0.625 mg bevacizumab) under sedation with propofol.4 The injections were carried out bilaterally if necessary during the same treatment session. The ocular surface was anesthetised with unpreserved oxybuprocain eye-drops. Eyelids and conjunctiva were rinsed with octenidin for disinfection. A sterile eyelid speculum was inserted and bevacizumab 0.625 mg (total volume 0.025 mL) was injected into the vitreous cavity via the temporal lower quadrant of the pars plana.

To minimise the risk of infection, infants who carried multi-drug resistant bacteria (methicillin-resistant Staphylococcus aureus, extended-spectrum β-lactamase producing gram-negative rods), received laser treatment (n=5) in the beginning of the observational period. Both treatment options were offered to the parents as experience with IVB increased during the study period, supplemented with systemic and topic antimicrobial treatment.

Infants with peripheral zone II disease were treated with laser photocoagulation (OcuLight SLx, Iridex Corporation, Mountain View, California, USA, using the following laser-settings: 810 nm, energy 250–300 mW, pulse duration 250 ms) under general anaesthesia in the beginning of the reported period. Based on the clinical experiences with IVB we offered both treatment options (IVB and Laser) to the parents of infants with peripheral zone II ROP after 2013.

Follow-up and definition of recurrence

Patients were examined daily within the first week after IVB to monitor for signs of intraocular infection, then weekly until complete regression of the active ROP (regression of plus disease as well as of any proliferation present at time of treatment), followed by monthly visits within the first 3 months and then at least every 3 months up to 12–15 months of age.

Recurrences requiring additional treatment after IVB were defined as progressing extraretinal proliferation (stage III ROP) of at least 3 clock hours with the potential to exert traction on the retina.

Patients who received retinal laser photocoagulation were seen at least weekly up to complete regression of active ROP, then every 3 months up to 12–15 months of age.

Data collection

The following parameters were collected from records of patients in this study: gestational age, birth weight, sex, ROP stage at time of treatment, treatment modality and gestational age at date of intervention (in weeks). Outcome parameters were time to complete regression of active ROP, defined as regression of plus disease and resolution of active (perfused) vitreoretinal proliferation, or time to recurrence from day of treatment.

At age 12–15 months of age cycloplegic refraction (manual retinoscopy or automated refraction using the Retinomax K-plus, Nikon, Japan) was measured and the spherical equivalent noted. Standardised uncorrected visual acuity was measured using teller acuity cards (TAC II, Stereo Optical Co., Chicago, Illinois, USA) at a test distance of 55 cm.12 ,13 Monocular visual acuity of >4.7 cycle/degree (20/130 Snellen equivalent) was regarded as within normal limits for infants 12–15 months of age. Children with binocular visual acuity less than 0.63 cycle/degree (20/960 Snellen equivalent) were regarded as having severe visual impairment.

Retinal morphology with respect to foveal architecture (presence or absence of the foveal reflex) and peripheral vascularisation of zone II, presence or absence of macular dragging and retinal detachment was also collected from the records.

Statistical analysis

Data are reported as medians and IQRs or as absolute numbers and percentages. Comparison between groups was done using the Mann-Whitney U-test, Kruskal-Wallis test or χ2 test or Fisher’s exact test as appropriate. A multivariate analysis of variance (MANOVA) was used to investigate the effect of ‘retinal zone’ and ROP treatment and their interaction on the outcome parameters considering gestational and chronological age as covariates. Spearman rank correlation coefficient (RS) was calculated to investigate the relationship between refractive error and chronological age. Statistical analyses were performed using Statgraphics Centurion software (V.16.0, Statpoint, Herndon, Virginia, USA) and MedCalc Software (V.13.3.1, Mariakerke, Belgium). A p value less than 0.05 was considered statistically significant.



Patients' characteristics of the 54 VLBW infants are shown in table 1. Posterior and peripheral zone II ROP significantly differed with respect to sex and gestational age but not with respect to birth weight. ROP treatment was performed in infants with a median postmenstrual age of 25 (IQR 24.1–26.1) weeks. Chronological and gestational age were significantly higher in infants within peripheral zone II ROP (p<0.001 for both) at time of treatment.

Table 1

Comparison of patient characteristics between infants with posterior ROP (zone I and posterior zone II) and peripheral zone II ROP

Of the 33 infants (66 eyes) with posterior ROP, 28 (56 eyes) were treated with IVB and 5 infants (10 eyes) received laser photocoagulation (median number of laser spots per eye was 952, range 904–1601).

Of 21 infants with peripheral zone II disease IVB was performed in 9 (18 eyes) and 12 infants (24 eyes) received photocoagulation (median number of laser spots per eye was 742, range 462–910).

Effect of ROP treatment

The effect of ROP treatment on the time to complete regression of active ROP, recurrence rates, visual acuity and spherical equivalent at month 12–15 are shown in table 2.

Table 2

Outcome of ROP treatment with IVB and laser photocoagulation in infants with posterior ROP (zone I and posterior zone II) and peripheral zone II ROP

The time to regression after treatment differed significantly (p<0.001) between the groups. As shown in figure 1, the shortest time to regression was seen in infants with posterior ROP treated with IVB. Laser photocoagulation led to a significantly longer time to regression (p=0.002). However, this effect was only seen in infants with posterior ROP.

Figure 1

Time from treatment until complete regression of ROP in days, horizontal bars represent the median. Children with zone I ROP are highlighted in bold. IVB, intravitreal bevacizumab; ROP, retinopathy of prematurity.

ROP recurrences occurred in 7 (12%) infants treated with IVB at a median of 12.7 weeks (range 11.3–15.6) after treatment. The impact of ROP zone on likelihood of ROP recurrence was not statistically significant. In five patients additional laser treatment was required (median number of laser spots 462 per eye, range 336–675) before ROP regressed. In two cases (a pair of twins, posterior zone II disease) recurrent proliferation regressed spontaneously with the formation of vitreoretinal degeneration appearing like white with/without pressure. Recurrences in laser treated eyes were not observed.

One child (gestational age 23+3 weeks/birth weight 615 g/zone I ROP) died 3 weeks after IVB due to cardio-respiratory failure secondary to severe bronchopulmonary dysplasia.

Exudative retinal detachment including the posterior pole developed in both eyes of one patient (24+0 gestational age/470 g) with posterior zone II disease 1 week after uneventful laser photocoagulation (780 spots right eye/1040 spots left eye). The retinal detachment resolved within 2 weeks with topical and systemic corticosteroid treatment, but left residual subretinal exudates covering the posterior pole. In another infant with posterior zone II disease macular dragging occurred after laser treatment (1601 spots per eye) associated with atrophy of the retinal pigment epithelium.

One child (24+2 gestational age, 420 g) with zone I ROP developed culture-negative keratitis in one eye after bilateral IVB treatment. The keratitis did not respond to antimicrobial therapy and resulted in permanent corneal opacity. The course of the second eye was uneventful.

Refractive error

To investigate the effect of ROP severity and treatment modality on refractive error and visual acuity, the data of both eyes of 49 infants (90% of all infants enrolled in the study) 12–15 months after treatment were pooled. As shown in table 2, there are statistically significant differences in spherical equivalent (p<0.001) between the four patient groups (posterior ROP zone I and zone II pooled together) versus peripheral ROP.

As shown in figure 2, the spherical equivalent was not statistically significant different in infants with IVB compared with those treated with laser photocoagulation with posterior ROP and peripheral zone II ROP. However, infants treated with IVB therapy had a significantly lower spherical equivalent (more myopic refractive error) if they had posterior ROP compared with peripheral zone II ROP (p<0.001). The spherical equivalent was also significantly lower in laser treated infants with posterior zone II disease compared with IVB treatment in infants with the peripheral zone II (0.007). There was no significant difference in terms of visual acuity.

Figure 2

Refractive error (spherical equivalent) 12–15 months after ROP treatment, horizontal bars represent the median. Pooled data of right and left eyes per patient eyes with zone I ROP are highlighted in bold. IVB, intravitreal bevacizumab; ROP, retinopathy of prematurity.

Multivariate analysis of variance

At the time of treatment infants differed significantly in terms of gestation and chronological age. The results are shown in table 3 and confirm the results of the univariate analysis. After adjustment for gestational and chronological age, a strong effect of ROP treatment modality on the time to regression remains. This effect, however, was only seen in infants with posterior ROP (p=0.001). ROP treatment modality had no significant effect on visual acuity and refractive error. Both were affected by ROP zone, but the effect was only marginally significant (p=0.036 and p=0.023). Of the covariates, only chronological age had a significant effect on visual acuity (p=0.017) and refractive error (p=0.006). Furthermore, there was a moderate correlation between chronological age and refractive error (Rs=0.297, p=0.004).

Table 3

Results of a MANOVA with retinal zone and ROP treatment as main effects, and their interaction and gestational age and chronological age as covariates

Retinal morphology and visual acuity

Retinal vascularisation in zone II was delayed and incomplete in all cases with posterior ROP. Twelve to 15 months after treatment, funduscopy showed a structurally normal fovea in 94 of 98 eyes (95%). In the remaining eyes macula dragging and exudates of the posterior pole were seen (see above – section Effect of ROP treatment).

Four infants demonstrated optic atrophy due to intraventricular haemorrhages and/or hydrocephalus that required ventricular shunt surgery.

Five patients were lost to follow-up. One infant died (see above) and four moved or were hospitalised in a child care facility.

Final visual acuity did not differ between posterior zone II and peripheral zone II subgroups. When considering a resolution of 4.7 cycle/degree (20/130 Snellen equivalent) as appropriate for age 12–15 months, 16 patients (32%) had a visual acuity below this level and five (10%) were considered as visually severely impaired, due to sequelae of exudative retinal detachment or optic atrophy with intraventricular haemorrhage and/or hydrocephalus.


In this study, we compared IVB and laser photocoagulation as treatments for type I ROP in VLBW infants. Time to regression, recurrence rate and refractive error in relation to retinal zone and gestational age served as main outcome measures.

We found that IVB led to faster regression in the more immature ROP stages zone I and posterior zone II compared with conventional laser photocoagulation. Previous studies have shown that vitreous VEGF concentration depends on the size of avascular (ischaemic) retina and that intravitreal injection of anti-VEGF antibodies binds free VEGF in the vitreous and thus inhibits its action almost instantaneously.14 This rapid effect can be clinically observed within the first 2 days after injection when plus disease regresses and regular retinal vessels begin to cross the demarcation line or the ridge between the vascular retina centrally and the avascular retina peripherally.

Resolution of vitreoretinal neovascularisation can be observed in most cases within a few weeks after IVB. This is in contrast to laser photocoagulation that ablates the avascular retina while active VEGF molecules persist in the vitreous. This may temporarily lead to transient progression and may explain the higher rate of unfavourable outcomes in infants with posterior ROP.

We did not observe macular dragging after IVB. Macular dragging was observed in half of the cases with zone I and in 15% of the cases with posterior zone II in the BEAT-ROP study.4 Hwang et al15 reported similar results with macular dragging in 16% of cases with posterior ROP treated with laser, but did not observe it in patients treated with IVB.

In most cases, the duration of rapid vascularisation of the avascular retina is limited to 6–8 weeks after IVB. Thereafter, vascularisation often ceases and telangiectatic vessels along the vascular front may develop. This process may be triggered by an increase in vitreal VEGF concentration. Recurrences requiring additional treatment—either a second IVB or laser treatment—were observed after primary IVB in 6–19% of treated eyes within 9–19 weeks in the literature.4 ,15 ,16 This is consistent with our observation. These recurrences may also resolve spontaneously as was seen in a pair of twins in this study. Even 12 months after IVB, retinal zone II may not be fully vascularised, suggesting that the mechanism of retinal maturation require a certain intravitreal VEGF concentration, or a change of molecular pathways. It has been suggested, that this could be the period when the retinal vessels become fully enveloped by pericytes.17 Longitudinal measurements of intravitreal VEGF concentrations could help to answer this question, but might be difficult to perform in clinical practice. Persisting vascular abnormalities such as branching, shunting and capillary dropout have been described in fluorescein angiograms as late as 9 months after IVB.6 ,18

The persistence of large avascular retinal areas after IVB treatment for ROP may be of concern. In children with other ischaemic retinopathies such as familiar exsudative vitreoretinopathy (FEVR) and Coats disease, laser photocoagulation of the avascular or ischaemic areas would be recommended.19 ,20 Gene defects causing FEVR have also been detected in some patients with ROP suggesting that these disorders might share certain mechanisms regarding regulation and maturation of the retinal vascularisation and remodelling of the capillary network.21 ,22 This observation may explain late recurrences up to 2½ years after primary IVB therapy for ROP.16 ,23

However, eyes with prethreshold ROP (stage II with plus disease in posterior zone II and mild stage III with plus) may regress spontaneously and will behave similarly to IVB treated eyes in terms of vascularisation, without the IVB associated temporary vascular growth. These eyes may display even larger avascular retinal areas, which have not been reported to induce ROP recurrences. This might argue for delayed vascular maturation potentially controlled by pericyte coating as mentioned above.17

Whether these vascular abnormalities affect function (eg visual acuity and visual field), remains to be determined.

Nevertheless great care should be taken for tight follow-up visits in paediatric ophthalmology departments after discharging the patients from the neonatology units.

Our findings suggest that regression after IVB happens faster in posterior ROP than in peripheral zone II while no differences between laser and IVB has been observed. Possible explanation include the generally slower rate of progression in patients with peripheral zone II ROP compared with posterior ROP, the higher gestational age and the greater likelihood of spontaneous regression in peripheral zone II ROP.

Preterm infants with ROP are at increased risk to develop refractive errors, particularly myopia.24 Both the ETROP and cryotherapy for ROP trails reported an increased prevalence of very high myopia (>8 dioptres) following ablative treatment (laser in the former, cryotherapy in the latter) in comparison with eyes that had spontaneous regression of ROP. This is in contrast to myopia in posterior prethreshold ROP that is attributed to retinal immaturity.25 ,26

In the BEAT-ROP study, the occurrence of very high myopia (> −8 dioptres), 2½ years after treatment was associated with conventional laser treatment in both zone I and zone II subgroups. Nevertheless, low (−1 to −5 dioptres) to high myopia (−5 to −8 dioptres) was found in about half of the eyes treated with IVB in zone I ROP and in about a quarter of the eyes with zone II ROP; this may be attributed to retinal immaturity. Nearly half the eyes had refractive errors within the physiological range (−1 to +4 dioptres).7

In their series with up to 5 years of follow-up, Hwang et al15 also reported an increased prevalence of high myopia in eyes with posterior ROP, as well as in eyes that received conventional laser treatment compared with IVB.

Our study demonstrates that 12 months after treatment, 78% of preterm infants with type I ROP who received either IVB or laser were within age appropriate range for refractive errors (−0.5 to +3 dioptres). However, the majority of eyes developing moderate to high myopia had posterior zone II ROP irrespective of treatment modality. Because of the low number of eyes with zone I ROP in this study, a statistical differentiation between zone I and posterior zone II was not statistically meaningful.

The strengths of this study include the relatively large total sample size, a single group of investigators and the use of the same equipment and protocol throughout the study period. Aside from its retrospective nature, limitations of this study include the non-uniform distribution of the patients and the relatively small number of patients in some of the subgroups. To standardise findings, documentation of morphological parameters such as foveal reflex and peripheral vascularisation might be performed with retinal imaging wide-field camera systems and ocular coherence tomography in future research.


IVB seems to lead to faster regression of type I ROP in eyes with posterior ROP compared with laser photocoagulation and thus to a lower rate of unfavourable outcomes. Recurrences after IVB treatment may occur in about 12% of the cases 12–15 weeks after the first treatment requiring prolonged follow-up.

In the majority of cases with posterior ROP, the retina will not be fully vascularised after 12–15 months. The significance of this finding is yet to be determined.

Treatment modality did not affect the refractive error 12 months after treatment. However, the spherical equivalent was more myopic in eyes with posterior ROP compared with those with peripheral zone II ROP. This suggests that retinal immaturity has a prominent effect on refractive error in infants with ROP.



  • Contributors All authors and co-authors contributed substantially to the preparation of the paper by examining the patents, collecting and reviewing the data, calculating the statistics, writing and discussing the results and critically reviewing the paper.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • At a glance
    Keith Barton James Chodosh Jost Jonas