INFANTS WITH birth weights less

CLINICAL SCIENCES of Retinopathy of Prematurity Michael X. Repka, MD; Earl A. Palmer, MD; Betty Tung, MS; for the Cryotherapy for Retinopathy of Prematurity Cooperative Group Objective: To report the timing of involution of acute retinopathy of prematurity (ROP). Design: An analysis of prospective retinal observational data recorded at infants eye examinations. Participants: Infants from the Multicenter Trial of Cryotherapy for Retinopathy of Prematurity (CRYO-ROP) had birth weights less than 1251 g and served as subjects. The study population included 766 children who were examined in 5 of the 23 study centers and who developed at least 1 clock hour of acute ROP, stages 1 through 3. One eye from each patient was randomly chosen for analysis. Main Outcome Measures: Investigators documented the location, extent, and severity of ROP during serial retinal examinations. The onset of the ROP s involution was determined from a review of these data, applying a set of predetermined criteria. Results: Acute-phase ROP began to involute at a mean of 38.6 weeks postmenstrual age. In 9% of patients, the ROP began to involute before 44 weeks of postmenstrual age. Acute ROP that demonstrated involution by moving from zone II to zone III was associated with an unfavorable outcome in 2 (1%) of 2 cases. Retinopathy of prematurity that was present only in zone III during a child s serial retinal examinations was never associated with the development of a partial or total retinal detachment. Conclusions: The onset of involution of acute retinopathyofprematuritycorrelatesbetterwithpostmenstrualrather than with chronological age. This is reminiscent of the reported similar correlation of postmenstrual age to the time of onset of prethreshold and threshold ROP. Zone III ROP was nearly always associated with a favorable outcome. Arch Ophthalmol. 2;118:645-649 From the Wilmer Ophthalmological Institute and the Department of Pediatrics, Johns Hopkins University School of Medicine, Baltimore, Md (Dr Repka); the Casey Eye Institute and the Department of Pediatrics, Oregon Health Sciences University, Portland (Dr Palmer); and the School of Public Health, University of Texas Health Science Center at Houston, (Ms Tung). Members of the Cryotherapy for Retinopathy of Prematurity Cooperative Group are listed in Arch Ophthalmol. 1996; 114:417-424. The authors have no proprietary interest in any of the instruments used in this study. INFANTS WITH birth weights less than 1251 g are at substantial risk for developing retinopathy of prematurity (ROP). In the Multicenter Trial of Cryotherapy for Retinopathy of Prematurity (CRYO- ROP), investigators found that 65.8% of such infants developed some stage of acute ROP and 6% developed threshold ROP, allowing entry into the randomized trial of retinal ablative cryotherapy in 1 eye. 1 The CRYO-ROP Cooperative Group has delineated the time course for the development of acute threshold ROP. 1 The timing of vascular events in the immature retina, specifically, the development of both ROP and threshold ROP, correlates closely with postmenstrual age. 1 This finding has implicated the infant s maturity, rather than postnatal environmental influences, in governing the time of occurrence of these vascular events. Important to physicians caring for extremely low-birth-weight infants is the timing of the onset of involution and ultimate resolution of ROP. Such information may be vital in determining the most costeffective strategy for scheduling serial retinal examinations. Though such information cannot eliminate the need for serial examinations of the individual infants, it may provide guidance for physicians who schedule and perform and for public health planners who estimate the manpower requirements for these examinations. The purpose of this study is to demonstrate the timing of the onset of involution of acute ROP. Secondary analyses consider the effect of gestational age, birth weight, and severity of ROP on the timing of involution. An additional analysis in this report concerns the outcome of peripheral zone III disease. We sought to determine if ROP in zone III that represented the involution of disease previously present in zones I and II could progress into an unfavorable outcome. We speculated that these latter patients have a worse prognosis than those whose initial diagnosis was acute ROP in zone III. RESULTS INVOLUTION There were 766 patients whose selected eye developed acute ROP in at least 1 sector of 1 zone in this sample of data from the 5 Natural History Clinical Centers of the 23 CRYO-ROP study centers. Twenty-two eyes (3%) never manifested involution through 3 months postterm. Twenty-one of these 645

SUBJECTS AND METHODS All subjects were participants in the CRYO-ROP. 2 The investigational review board at each participating institution approved the CRYO-ROP study protocol. Written informed consent for the sequential examinations that provided data for this outcome study was obtained from each subjects parent or guardian. All the infants weighed less than 1251 g at birth and had serial eye examinations, started by age 49 days, to document the presence and character of acute ROP during the neonatal period. Serial research examinations were performed about every 2 weeks (1 week for prethreshold ROP) until involution was clinically documented or the retinal vascularization had reached into zone III (vascularization was complete in the nasal meridians). The allowance to discontinue serial examinations under these conditions was placed in the protocol because zone III cases were not eligible for randomization in the cryotherapy protocol. A follow-up retinal examination was performed 3 months after the estimated due date of 4 weeks of postmenstrual age. age was defined as the chronologic age in weeks plus the gestational age at birth. Gestational age was assigned by the neonatologist caring for the patient, which represented a best estimate based on menstrual history, obstetrical data (including ultrasonography when available), and neonatal physical assessments. This study includes research data from children born prematurely who were evaluated at 5 of the 23 study centers. Those centers are the Natural History Clinical Centers, an identified subgroup of the CRYO-ROP clinical centers. One eye from each patient was randomly designated for these analyses. This was done for 2 reasons: (1) the data for the 2 eyes cannot be considered to be independent and (2) some patients were treated and thus did not follow a natural course. For patients who developed bilateral threshold disease, the control eye was used for the analysis. This eye had been randomly chosen to not receive cryotherapy in the CRYO-ROP study. For infants with unilateral threshold disease, the fellow nonrandomized eye was used. There were 128 patients enrolled at these 5 centers. Of these patients, 766 infants developed at least 1 clock hour of any stage of acute ROP in any zone and are the subjects of this report. Acute ROP was defined and documented according to the international classification of ROP. 3 The investigators recorded the details of each retinal examination from 34 separate 3 sectors of the eye: 12 in zone I, 12 in zone II, and in zone III. The term sector as used in this report includes data from 1 clock hour of 1 zone of the retina. Threshold ROP was defined as 5 contiguous or 8 cumulative clock hours of stage 3+ ROP in zone I or II. Special prethreshold disease was defined as stage 3+ ROP with fewer than the defined number of sectors of involvement needed for the threshold category. Prethreshold ROP was defined as any zone I or II disease with stage 2+ or stage 3. AlthoughCRYO-ROPinvestigatorsweredirectedtoidentify regressing ROP during their study examinations of each infant, it turned out that this was not always feasible, as eyes had either a partial or total retinal detachment reported at the 3-month assessment of outcome, while 1 eye was reported to have regressing ROP, but was never previously reported to have had acute ROP. Nineteen patients (2%) were unavailable for the 3-month examination and their timing of involution could not be estimated. Thus, in 725 records (95%) the time of onset of involution could be determined. The time of involution could be determined after reviewing only the serial retinal examinations in 675 patients (88%). The onset of involution could be estimated in an additional 5 patients (7%) by using the results from the 3-month outcome retinal examination. The interval from the last serial retinal examination to the 3-month examination was an average of 46.7 days in those infants. Results of the analyses examining the timing of onset of involution were very similar whether we included or excluded the children in whom we had to use the 3-month examination. Therefore, we elected to report the data for all 725 eyes. generally began between 34 and 46 weeks (5%-95%) of postmenstrual age. The mean time of onset was 38.6 weeks of postmenstrual age. A plot of time of onset of involution peaks between 36 and 4 weeks (Figure 1). A separate plot of the mean chronological age at the onset of involution shows a similar shape, with a mean of 11.6 weeks of chronological age (Figure 2). The cumulative percentage of patients who showed the onset of involution is given in Figure 3. The ROP began to involute by 4 weeks of postmenstrual age in 74% and by 44 weeks of postmenstrual age in 9%. The time of onset of involution based on postmenstrual age was analyzed in terms of zone of ROP, birth weight, gestational age, and severity of ROP (Tables 1, 2, 3, and 4). The zone of ROP did not substantially affect the timing of involution when related to postmenstrual age (Table 1). began at virtually the same mean postmenstrual age for each zone of disease. The analysis by birth weight showed that the onset of involution for each of the 3 birth weight groupings fell within the same postmenstrual week (Table 2). In general, the chronological age of onset of involution correlated inversely with birth weight. This is consistent with the observation that the onset of involution best correlates with postmenstrual age. Analysis by gestational age showed a slightly earlier onset of involution for lower gestational ages (Table 3). Children with gestational ages younger than 25 weeks had onset of involution on average 1 week earlier than the average patient. TheanalysisofseverityofROPbyzoneofdiseasefound no clinically important difference in timing (Table 4). An analysis of zone II disease by severity found involution to begin earliest with the mildest disease (stage 1: mean, 37.4 weeks) and latest with the most severe disease (stage 3+ at 9-12 hours: mean, 4.2 weeks). Though this trend is significant (P.1), it should be viewed with caution. The short interval is probably not clinically significant, and may have been influenced by minor differences in the pattern of follow-up of these infants, such as children with posterior disease or higher stages of ROP being seen more frequently. 646

regressing ROP can often be identified only retrospectively, through review of the serial examinations. Therefore, we developed several criteria to define the onset of involution. This decision algorithm was tested on a sample of serial retinal examination data from CRYO-ROP study infants taken from other centers not included in this study. The testing of the algorithm led to a refined set of criteria by which we define the involution of ROP. Meeting any one of the following criteria was sufficient for documentation of involution: (1) The use of research codes 8 or 9 (these identify regression or regressed disease) when there was a history of acute ROP in the same retinal zone at an earlier examination. (2) A reduction by 2 or more sectors in the number of sectors of the highest stage of acute disease recorded in a given zone from one examination to the next; for example, a change from 6 sectors of stage 2 disease in zone II to 4 sectors of stage 2 disease in zone II. (3) Documented completion of vascularization in all sectors of a zone that had previously had acute ROP; for example, a change from 4 sectors of stage 1 in zone II to 4 sectors of stage 1 in zone III. The development of stages 4 or 5 acute ROP was considered progression of disease. The specific time of involution could not be determined for these patients. If a patient would demonstrate 2 separate episodes of involution, the timing of the second episode was recorded for this analysis. Two of the authors independently applied this set of refined criteria for involution to the findings of the sequential retinal examinations. They each analyzed the data from all 766 eyes in the Natural History Clinical Centers to determine the presence of involution. The age of onset of involution was defined as the mean age from the 2 serial retinal examinations that showed acute ROP at the earlier examination and definite involution at a subsequent examination. The second examination was not always one of the serial retinal examinations; for some patients, it happened that the serial examinations were discontinued by the research protocol prior to the onset of definitive involution. Whenever acute ROP did not seem to be involuting at the time of the last serial retinal examination, the results of the 3-month outcome examination were used. Using the 3-month examination generally resulted in a larger gap between the last serial examination and the 3-month examination than that between the serial examinations. Use of these data leads to a slightly older age for onset of involution. Data from the research forms were also tabulated by gestational age, birth weight, zone, and severity of ROP to determine the mean time of onset of involution with respect to each of these factors. Two supplementary analyses of the involution data specific to zone III ROP were included in this study. The first of these concerned the frequency of the initial development of acute ROP in zone III, which then progressed to stage 4 or worse, rather than involuting. The second analysis asked how often ROP in either zone I or II subsequently developed into zone III ROP (with complete vascularization of zones I and II). The 3-month anatomical outcome of all these eyes was tabulated as favorable or unfavorable. Unfavorable outcomes were considered either a retinal detachment or a retinal fold through the macula. ZONE III ROP A secondary analysis assessed the outcome of 135 eyes that had acute ROP identified only in zone III throughout their entire series of retinal examinations (Table 5). For the eyes in this analysis, acute ROP was never seen in a lower zone, although incomplete vessels in zones I or II may have been reported. Not one of these 135 eyes had an unfavorable outcome. Plus disease (dilatation and tortuosity of the posterior retinal vessels) was identified in 1 eye. In this child s eye, the plus disease was associated with stage 2 ROP. Five eyes were reported to have zone III stage 3 ROP, and none were described as having plus disease or experiencing an unfavorable anatomical outcome. Next, we investigated zone III ROP cases in which any stage of acute disease was observed during earlier examinations in either zone I or II (Table 6). The conversion of ROP from zone I or II to zone III would meet our definition of involution. There were 8 zone I eyes in our cohort. Only 1 eye subsequently demonstrated zone III ROP, and this eye experienced a favorable anatomical outcome. 4 There were 2 eyes with zone II ROP that were subsequently reported to have evolved into zone III ROP. Two of these eyes had an unfavorable outcome, both documented with cicatricial grade III ROP. 5 COMMENT is a favorable event in the clinical course of the vascular disease for all infants who develop acute ROP. Information about the timing of this event has not previously been available in the literature describing ROP, although involution must occur in most infants who develop ROP. The time of onset of involution could not be directly ascertained from the CRYO-ROP database, because many cases of ROP involute inconspicuously, so that it is possible only in retrospect to recognize its occurrence. We therefore developed a set of criteria to define the onset of involution. Similarly, the time of the completion of involution is difficult to document in all cases for at least 2 reasons. First, it was not required for CRYO-ROP patients to be evaluated up to the completion of resolution and full vascularization of their retinas. Second, the retinal periphery may never fully vascularize after the development and involution of acute ROP. 6 Such factors may hinder any future studies from definitively documenting the completion of vascularization. The CRYO-ROP Cooperative Group has previously reported that the timing of acute ROP events correlates best with postconceptional maturity. 1 It was noted that the onset of stage 1 disease occurs at a median time of 34.3 weeks of postmenstrual age, prethreshold disease at a median of 36.1weeks,andthresholddiseaseatamedianof36.9weeks. Inthisreport,involutionofROPwasfoundtobeginatnearly a constant postmenstrual age, irrespective of most clinical and retinal factors studied. The peak time for the onset of involution was found to be during the 38th postmenstrual week, with 73% of eyes demonstrating onset between 35 weeks and 41 weeks. These times for the onset of involution were derived from the average of ages at 2 visits that 647

Patients With Onset of, % 16 14 12 8 6 4 2 Table 1. Time of Onset of by Zone of Acute Retinopathy of Prematurity (ROP) Zone of Acute ROP I 8 39.1 (35.3-44.4) 13.1 (9.3-18.6) II 575 38.3 (33.9-45.3) 11.6 (6.7-19.1) III 135 39.8 (34.-5.6) 11.4 (5.7-22.3) Indeterminate 7 39.5 (35.6-48.7).1 (6.9-19.7) 32 34 36 38 4 42 44 46 48 5 52 54 56 Figure 1. The onset of involution by postmenstrual age. The time for the onset of involution is spread across a wide range of ages, but peaks during postmenstrual weeks 36 to 4. Patients With Onset of, % 14 12 8 6 4 2 5 7 9 11 13 15 17 19 21 23 25 27 29 Figure 2. The onset of involution by chronological age. The onset of involution peaks between 8 and 12 weeks of age (mean age of onset, 11.6 weeks). Patients With Onset of, Cumulative % 9 8 7 6 5 4 3 2 31 33 35 37 39 41 43 45 47 49 51 53 55 57 Figure 3. Cumulative percentage of patients demonstrating onset of involution by postmenstrual age in weeks. demonstrated involution. If we were to use the second age as the time of onset of involution, the mean time would rise by 1.7 weeks, to a mean of 4.3 weeks. We believe that the derived value represents the more clinically relevant value. Plots of the timing of onset of involution (Figures 1 and 2) closely mimic the shape of the plots of the timing of onset of prethreshold and threshold acute ROP when compared with either postmenstrual age or chronological age, although they shifted to slightly older ages. 1 The onset of involution peaks about 2 weeks after the Table 2. Time of Onset of by Birth Weight Birth Weight, g (5%-95%* 5 1 39. 16. 5-749 163 38.8 (34.-45.3) 13.4 (8.6-2.3) 75-999 325 38.3 (33.7-45.9) 11.5 (6.9-18.1) -125 236 38.9 (34.3-48.1).4 (6.-19.1) Table 3. Time of Onset of by Gestational Age Gestational 25 77 37.8 (33.7-44.9) 13.9 (9.9-2.9) 25-27 363 38.3 (33.6-46.3) 12.1 (7.3-19.9) 28-31 263 39.2 (34.6-46.4).4 (5.9-17.9) 32 22 4.5 (37.3-45.4) 8.1 (5.3-13.4) peak onset of prethreshold disease and slightly earlier after the peak onset of threshold disease. These data suggest that the onset of involution represents one more event in the course of ROP, in which postmenstrual age is of key importance in establishing milestones of retinal development. This lends further support to the earlier speculation by CRYO-ROP Cooperative Group, in which they hypothesized that the development of acute ROP is controlled by a predetermined maturational sequence. 1 Many clinicians consider the advancement of retinal blood vessels from zone II to zone III in an eye where there has been no acute ROP to be indicative of a good outcome for that eye. This was found to be true for the patients in this study. For such children, this retinal vascular milestone might be considered as similar to involution. For the CRYO-ROP cohort, 5% of such eyes achieve this milestone by the 36 weeks of postmenstrual age. 1 This is nearly 3 weeks earlier than the timing of involution in eyes with acute ROP described in this study, suggesting that the development of acute ROP may cause a delay in the normal vascularization timetable. 648

Table 4. Time of Onset of by Severity of Retinopathy of Prematurity (ROP) Severity of ROP Zone I, any ROP 8 39.1 (35.3-44.4) 13.1 (9.3-18.6) Zone II, stage 3+, 33 4.2 (35.9-46.9) 14.3 (.9-19.9) 9-12 h Zone II, stage 3+, 13 39.7 (33.7-44.1) 13. (9.9-18.1) 5-8 h Zone II, stage 3+, 15 38.7 (34.3-43.) 12.6 (.3-15.) 1-4 h Zone II, stage 3, 9 4.2 (34.6-48.3) 13.9 (8.4-21.) not plus Zone II, stage 2 23 38. (33.4-44.7) 11.5 (7.-18.1) Zone II, stage 1 194 37.4 (33.7-45.4). (5.7-18.1) Zone III, any ROP 135 39.8 (34.-5.6) 11.4 (5.7-22.3) Indeterminate zone 7 39.5 (35.6-48.7).1 (6.9-19.7) Table 5. Detailed Severity of Zone III Retinopathy of Prematurity (ROP) Eyes* Maximum ROP Stage No Plus With Plus Total 1 98 98 2 31 1 32 3 5 5 All Stages 134 1 135 *Zone III ROP eyes with no previous examinations with acute ROP in zones I or II. None of these eyes experienced an unfavorable outcome. Table 6. Anatomical Outcome of Zone III Retinopathy of Prematurity (ROP) Eyes That Had Acute ROP in a Lower Zone Recorded During an Earlier Retinal Examination Favorable, No. (%) Unfavorable, No. (%) Zone I 1 () () Zone II only 198 (99) 2 (1) Total 199 (99) 2 (1) The outcome of zone III disease was a secondary goal of this report. Zone III, as defined in by the CRYO-ROP Cooperative Group, was not a geographic location, but rather consistent with the approach stated by the international Committee for the Classification of ROP. 3 Thus, vessels and/or ROP were defined as zone III if the vessels in the 2 nasal sectors reached within 1 disc diameter of the ora serrata and there was no ROP in the 2 nasal sectors. When we defined patterns of resolution of ROP, we noted that one pattern is the centrifugal progression of retinal vascular growth into zone III. In an earlier publication of the CRYO-ROP study, we reported that zone III ROP occurred in 12% of all cases studied. 4 However, it is important to recognize that the cases reported as zone III included only those eyes in which no ROP had been observed previously in zones I or II. Among those eyes, the outcome of zone III ROP was found to be favorable more than 99% of the time. It was even felt inconclusive as to whether zone III ROP ever leads to unfavorable anatomical outcome. Furthermore, we found no eye in this study with acute ROP identified only in zone III during serial examinations advancing to an unfavorable outcome. However, clinicians have reported seeing unfavorable outcomes from zone III ROP. 7 For the present analysis, we hypothesized that zone I or II ROP that involutes by progressing out to zone III could then lead to an unfavorable outcome, which would account for these unfavorable outcomes. We reviewed the CRYO-ROP data to find out how often this scenario occurs. Of the 21 eyes with zone I or II disease that progressed to zone III ROP, only 2 had an unfavorable outcome with cicatricial grade III. Thus, the advancement of acute ROP from zone II to III is associated with involution in nearly every case. These data strongly support the recommendation to monitor zone III ROP without treatment, when zone III is considered as in the CRYO- ROP study. This report demonstrates that the onset of involution correlates well with postmenstrual age. Most children showed evidence of involution of acute ROP by 44 weeks of postmenstrual age. Thus, clinicians can expect to decrease the frequency of and, in some cases, discontinue examinations of children with a history of acute ROP, once involution has been carefully documented or after 44 weeks of postmenstrual age. For children who never develop acute ROP with vascularization confirmed to be reaching into zone III, it would also be reasonable to discontinue serial examinations. The examiner, however, must be certain that the normal vascularization has reached the nasal ora serrata for any temporal ROP to be in zone III. Accepted for publication August 3, 1999. This study was supported by National Eye Institute Cooperative Agreement U EY5874, National Institutes of Health, Bethesda, Md (Drs Repka and Palmer and Ms Tung). Corresponding author: Michael X. Repka, MD, 233 Wilmer, Johns Hopkins Hospital, 6 N Wolfe St, Baltimore, MD 21287-928 (e-mail: mrepka@jhmi.edu). REFERENCES 1. Palmer EA, Flynn JT, Hardy RJ, et al, for the Cryotherapy for Retinopathy of Prematurity Cooperative Group. Incidence and early course of retinopathy of prematurity. Ophthalmology. 1991;98:1628-164. 2. Multicenter Trial of Cryotherapy for Retinopathy of Prematurity (CRYO-ROP) Cooperative Group: Manual of Procedures. Springfield, Va: National Technical Information Service, US Dept of Commerce; 1985. Publication PB88 16353. 3. The Committee for the Classification of Retinopathy of Prematurity. An international classification for retinopathy of prematurity. Arch Ophthalmol. 1984;2: 113-1134. 4. Cryotherapy for Retinopathy of Prematurity Cooperative Group. The natural ocular outcome of premature birth and retinopathy: status at 1 year. Arch Ophthalmol. 1994;112:93-912. 5. Reese AB, King MJ, Owens WC. A classification of retrolental fibroplasia. Am J Ophthalmol. 1953;36:1333-1335. 6. Mintz-Hittner HA, Kretzer FL. Postnatal retinal vascularization in former preterm infants with retinopathy of prematurity. Ophthalmology. 1994;1:548-558. 7. Carden SM, Good WV. Macular folds and poor vision associated with zone III retinopathy of prematurity. Am J Ophthalmol. 1998;126:46-462. 649