Acute posterior vitreous detachment (PVD) commonly results in vitreous hemorrhage (VH) and retinal tears.1 Previous studies have reported a 10% to 15% incidence of retinal tears with PVD alone.2–6 However, in the presence of a VH, that number grows to between 54% and 91%.7–9 Because the presence of a VH in patients with hemorrhagic posterior vitreous detachments (HPVDs) makes it more difficult to see the fundus, retinal tears may be missed by even the most careful clinician. Detection of retinal tears is critical in these patients in order to prevent severe vision loss related to complications such as retinal detachment (RD).7,8
Standard protocol for patients with both PVD and VH has been close follow-up with indirect ophthalmoscopy, scleral depression, and ultrasonography. Early pars plana vitrectomy (PPV) for PVD with dense fundus-obscuring VH is another approach that has been recommended for some patients, as it has been shown to be safe and effective and may lower the risk of severe, permanent vision loss by preventing complications such as macula-off RD.7–9 Although there are reports in the literature on both conservative serial monitoring and surgical intervention with vitrectomy, it is not clear which method results in better patient outcomes. Here in, we report the results of our cohort study comparing observation versus PPV for the management of HPVD. Our findings suggest that early intervention with PPV may lower the risk of RD, increase detection of occult retinal tears obscured by VH, and allow quicker resolution of VH.
Patients and Methods
Study Design and Participants
In this retrospective cohort study, data from 109 eyes of 105 consecutive patients with HPVD at the University of Alabama at Birmingham, Retina Consultants of Alabama, from April 2014 to February 2018, were extracted from medical records. All eyes underwent a complete posterior segment exam, including scleral depression with indirect ophthalmoscopy. Patients with retinal breaks underwent laser treatment in the office. Since all patients had a HPVD, they were given the choice of PPV or observation. Both the surgical and observation groups completed one to three follow-up visits to assess patient progress and outcomes.
Exclusion criteria included history of any vascular retinal disease, uveitis, neovascular age-related macular degeneration in either eye, previous intraocular surgery other than uncomplicated cataract or refractive surgery, and history of recent trauma. Inclusion criteria included initial visual acuity (VA) of 20/400 or better. Institutional review board (IRB) approval was obtained prior to collection of patient data from the University of Alabama-Birmingham IRB. This study was HIPAA compliant and adhered to the tenets of the Declaration of Helsinki.
Variables and Definitions
Patient data collected included age, sex, race (white, African-American, other), eye involved (left, right), lens status (phakic, pseudophakic), best-corrected VA (BCVA) at initial presentation, initial intraocular pressure (IOP), presence of retinal tear found on exam (yes, no), last IOP, BCVA, VH resolution (yes, no), and the development of RD (yes, no). BCVA was assessed using the Snellen chart and was defined as the best VA in the study eye from any follow-up exam. The last IOP was the last pressure taken upon follow-up examination. For the 66 eyes that elected surgery, additional data were collected including: early (≤ 14 days) or delayed (> 14 days) PPV, presence of retinal tears found during surgery not seen on initial exam, surgical complications, and other surgical procedures performed. Main outcome measures were rate of retinal detachment, time to BCVA, and time to VH resolution.
All patients in the surgery group underwent a standard three-port, widefield, 25-gauge PPV. Vitreous based shaving and scleral depression were used in all patients to remove VH near the pars plana to identify all possible retinal breaks. All retinal breaks were treated with endolaser. No patient underwent gas tamponade.
Categorical and continuous variables were compared using Fisher's exact test and Wilcoxon ranked-sums test, respectively. Time to BCVA (days) and time to vitreous hemorrhage resolution (days), since the initial exam, were assessed between the two groups using the Kaplan Meier method and the Wilcoxon test. P values less than .05 were considered statistically significant. All analyses were conducted in SAS Version 9.4 (SAS Institute, Cary, NC).
In all, 109 patients had HPVD, of whom 64 (66 eyes) underwent PPV and 41 (43 eyes) opted for observational treatment (Table 1). The median follow-up time was 45.5 days (interquartile range [IQR]: 17 days to 112 days) for PPV patients and 65 days (IQR: 32 days to 111 days) for observational patients. The mean ages differed, with observational patients being slightly younger (mean: 62.65 years ± 8.96 years) compared with those electing surgery (mean: 66.01 years ± 7.09 years; P = .0315). Males represented the majority of patients in the observational group (53.5%) and the surgical group (60.9%, P = .5495). There was no significant difference in race between the two treatment groups, with 90% or more of patients being white (P = .4373). Among those electing PPV, 69.1% were phakic whereas 53.2% of those under observation were phakic (P = .1536). Among PPV patients, 69.2% underwent early PPV.
Demographic and Eye Outcomes of 109 Eyes (105 Patients) With Hemorrhagic Posterior Vitreous Detachment Stratified by Treatment, PPV, or Observation: 2014–2018
Upon initial exam, the median BCVA for the observation group was 20/30 (IQR: 20/25 to 20/40), which did not differ significantly from the PPV group median of 20/30 (IQR: 20/25 to 20/50; P = .4825). The median BCVA (at follow-up) was 20/20 (IQR: 20/20 to 20/25) and 20/25 (IQR: 20/20 to 20/30) for the observational and PPV treatment groups, respectively (P = .5494). The median time to BCVA (Figure 1) was 30 days (IQR: 0 days to 70 days) for the observation group and 14 days (IQR: 0 days to 49 days) for the PPV group (P = .5936). The IOP on initial exam did not vary for the observational or PPV groups, with median IOPs of 14 mm Hg (IQR: 13 mm Hg to 17 mm Hg) and 15 (IQR: 13 mm Hg to 18 mm Hg; P = .8396). Similarly, the last follow-up IOP did not differ by treatment (P = .8738). Retinal tears were common with 33.3% of the PPV and 44.2% of observational patients having a tear on initial exam (P = .3128). Among the surgical group, 36.4% of patients were found to have additional retinal tears intraoperatively and were treated with laser; in this case, additional tears refers to tears that were not detected during an in-office visit, but were discovered during surgery. Patients may have had both preoperatively detected tears as well as additional tears detected during surgery. The median time to VH resolution was significantly shorter for those electing PPV (14 days; IQR: 7 days to 35 days) compared to those who were observed (58.5 days; IQR: 30 days to 91 days; P < .0001) (Figure 2). One out of 64 (1.5%) of patients undergoing PPV experienced a transient vitreous hemorrhage, which resolved. RDs, the only complication observed, occurred more frequently among observational patients (11.6%) compared to 1.5% of surgical patients (P = .0344). The only RD in the PPV group was from a new retinal break occurring 1 year from the original surgery date. Patients with a RD had significantly worse median BCVA (20/72.5; IQR: 25 to 70) compared to those without RD (20/25; IQR: 20 to 25; P = .0331). Those patients who underwent early PPV had a median BCVA of 20/25 (IQR: 20 to 30), and those with late PPV had a median BCVA of 20/20 (IQR: 20 to 25; P = .0022). There was no cataract progression in either group.
Time to best-corrected visual acuity (BCVA) by surgical status. PPV = posterior vitreous detachment
Time to resolution of vitreous hemorrhage (VH) by surgical status. PPV = posterior vitreous detachment
Acute PVD in the setting of a VH often results in complications such as retinal tears and, if left untreated, RDs.1–3 Previous studies on this topic vary widely, although all studies highlight the importance of close follow-up given the heightened rate of RD in these patients.7,9,10 However, current literature is unclear as to whether observation or vitrectomy results in better outcomes for patients. Results in this study suggest time to VH resolution is shorter for patients undergoing PPV, and rates of RD are lower in patients undergoing PPV. Also, the findings herein suggest that BCVA outcomes are good (20/25 or better) in both groups with time to attainment to VA being similar in both groups. Although, there was a statistical difference in BCVA in early versus late PPV (which was not explainable), this difference was very small and was not clinically significant. Of note, there was no difference in cataract progression between the two groups.
Previous studies have confirmed that the risk of developing a RD among HPVD patients persists anywhere from 2 to 6 weeks post initial examination. Therefore, standard protocol recommends close follow-up 4 to 6 weeks after the initial diagnosis.6,8 Because of the high risk of retinal tears in these patients (54% to 91%7,8,11–17), it is clear that at the very least, patients should be closely monitored for new or missed tears following diagnosis of HPVD. In a study reporting on the incidence of RDs and visual outcomes in eyes presenting with HPVD, Sarrafizadeh et al.9 state that close follow-up with clinical examination and ultrasonography is necessary. This conservative approach ultimately resulted in 39% of patients needing retinal detachment repair. Furthermore, Rabinowitz et al.18 and DiBernardo et al.19 concluded that the accuracy of ultrasonography in determining the presence of retinal tears is variable, pointing to the possible need for a more proactive method to identify hidden retinal tears and prevent permanent vision loss resulting from complications due to occult retinal tears.18,19 Although the conservative approach in our study resulted in only 11.6% of patients developing RDs, the more proactive approach (PPV) had a rate of RD of only 1.5%. Several prior studies have suggested that PPV should be considered in these patients due to the high risk of permanent vision loss due to retinal detachment.7,9,20 Our findings of less RDs and consequent vision loss (mean BCVA of 20/25 in the non-RD patients vs. 20/72.5 in the RD patients) in patients who undergo surgery for HPVD are in agreement with these studies. Coffee et al.8 state that acute symptomatic retinal breaks resulting from HPVD have a high risk of RD with the potential of permanent vision loss. The underlying cause of these RDs, of course, are occult tears not visible through the VH. The case for vitrectomy in these patients is quite compelling, in our opinion, when one considers that more than one-third (36.4%) of the eyes taken to surgery in our cohort had tears not seen preoperatively. We believe that the observation group in our series developed more RDs due to retinal breaks not identified because of VH obscuring an adequate view of the fundus. Since RD occurred more frequently among those opting for observation compared to PPV, and since the vision loss associated with RD is often great, surgery may be the best treatment option for patients with VH significant enough to inhibit an excellent view of the retinal periphery during a scleral depressed examination.
As seen in this study, it is certainly logical that the rate of VH resolution would be faster for patient's undergoing vitrectomy for HPVD as opposed to those who are observed carefully. It is interesting in this study that the time to attainment of BCVA was not shorter in the PPV group, which had faster VH resolution. The authors do not have a definitive explanation for both groups achieving BCVA at similar times. It is possible, however, that patients in the observation group may have had less severe vitreous hemorrhages than the surgery group, allowing for the similar timeframe of attainment of BCVA. This finding is reassuring for those patients who are observed, suggesting that patients can obtain visual improvement in a reasonable span of time. Moreover, the BCVA was similar for both groups, suggesting that both treatment options are still reasonable. Although the results in this study suggest PPV for HPVD may prevent RD in a significant number of patients, many patients cannot make it to the operating room (OR) with acceptable risk after the initial exam/diagnosis of HPVD because of systemic health concerns. Moreover, logistical, economic, and insurance situations may make it difficult to take a patient to the OR for HPVD. In such cases, our data still suggest that patients do quite well overall.
The limitations in this study include its retrospective nature, that vitreous hemorrhages cannot be graded in a retrospective fashion, that new tears noted during PPV possibly could have be iatrogenic, the fact that patients were able to choose treatment with physician influence, the variable follow-up period among the treatment groups, and VA measurement using the Snellen chart. Nevertheless, meaningful conclusions can be drawn from the results that have practical implications for any ophthalmologist who sees patients with HPVD. Regarding the inability to grade the severity of vitreous hemorrhages retrospectively, the authors, while recognizing this limitation, point out the fact that there were still more detachments in the observation group. One can hypothesize that, if anything, the patients taken to surgery would be more likely to have worse VHs than the control group. In the presence of a dense VH with no view, the surgeons in our group gave the patients the option of observation versus vitrectomy, but if there were any physician bias brought into the discussion with the patient, the physician would be more likely to suggest vitrectomy for the worse hemorrhages. The fact that there were still more detachments in the observation group (the group that theoretically would have less severe hemorrhages), strengthens the conclusions of our study, suggesting that vitrectomy should certainly be a consideration in HPVD patients. Additionally, with regard to iatrogenic tears, none of the new tears discovered were mentioned as iatrogenic in any of the operative dictations. Also, although we cannot definitively prove it from our retrospective dataset, the surgeons at Retina Consultants of Alabama/The University of Alabama at Birmingham who performed these cases (most of whom have 20+ years of experience) all agree that these tears were not ones created at the time of vitrectomy. Rather, they were occult tears discovered as the blood was cleared during vitrectomy. Furthermore, this is the only study to our knowledge directly comparing observation and PPV in patients with HPVD.
In conclusion, PPV for HPVD resulted in a shorter time to VH resolution, the ability to find additional retinal breaks intraoperatively, and fewer retinal detachments. Moreover, patients in both the observation and PPV groups had similar and good visual outcomes overall. In the era of modern small gauge vitrectomy surgery, PPV may result in fewer complications and better patient outcomes for patients with HPVD as compared to observation alone. Further research including a prospective clinical trial with larger sample sizes would further elucidate the role of PPV in the setting of HPVD.