Jump to comment:

• Response to: Response to "Preoperative aqueous humour flare values do not predict proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment"
  Verena C Mulder, Jan Tode, Elon HC van Dijk, Jan C van Meurs and Felix Treumer
  Published on: 16 May 2017
• Response to "Preoperative aqueous humour flare values do not predict proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment"
  Friederike Schaub, Bernd Kirchhof and Sascha Fauser
  Published on: 7 February 2017

• Published on: 16 May 2017

  Response to: Response to "Preoperative aqueous humour flare values do not predict proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment"

  • Verena C Mulder, PharmD The Rotterdam Eye Hospital/Rotterdam Ophthalmic Institute, Rotterdam, The Netherlands
  • Other Contributors:
    • Jan Tode, MD
    • Elon HC van Dijk, MD
    • Jan C van Meurs, MD
    • Felix Treumer, MD

  Thank you for your interest in our publication entitled "Preoperative aqueous humour flare values do not predict proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment".

  As per request, we would like to provide more details on our protocol.

  As described in our discussion, centre 1 used the mean of ten correct measurements making sure these measurements did not differ more than 2 standard deviations from each other. In centre 2, seven correct measurements were recorded of which the highest and lowest value were discarded leaving an average of five measurements. A correct measurement meant that the background readings did not differ more than 15% (indicated by the code ‘BG’ on the output) and that single “cell/C” measurements were replaced by an additional measurement. In addition, measurements with a small signal to noise ratio (indicated by the code ‘s/n’) were avoided as much as possible. However, with low flare values this was not always feasible. The flare meters were located in a room with blinds (centre 1) and a room without windows (centre 2); computer screens and lights were turned off during measurements. Both flare meters were calibrated monthly to assure correct readings. We therefore believe that the included mean values are artefact free.

  Despite the exclusion of patients with additional conditions such as AMD, CRVO and preoperative PVR grade C or higher, we did end up with patients with a preopera...

  Show More

  Thank you for your interest in our publication entitled "Preoperative aqueous humour flare values do not predict proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment".

  As per request, we would like to provide more details on our protocol.

  As described in our discussion, centre 1 used the mean of ten correct measurements making sure these measurements did not differ more than 2 standard deviations from each other. In centre 2, seven correct measurements were recorded of which the highest and lowest value were discarded leaving an average of five measurements. A correct measurement meant that the background readings did not differ more than 15% (indicated by the code ‘BG’ on the output) and that single “cell/C” measurements were replaced by an additional measurement. In addition, measurements with a small signal to noise ratio (indicated by the code ‘s/n’) were avoided as much as possible. However, with low flare values this was not always feasible. The flare meters were located in a room with blinds (centre 1) and a room without windows (centre 2); computer screens and lights were turned off during measurements. Both flare meters were calibrated monthly to assure correct readings. We therefore believe that the included mean values are artefact free.

  Despite the exclusion of patients with additional conditions such as AMD, CRVO and preoperative PVR grade C or higher, we did end up with patients with a preoperative flare value above 100 pc/ms. These measurements were accurate because they showed small SD’s (e.g. mean 263.8± SD 2.8; mean 196.5±SD 3.5; mean 162.0±SD 6.0) and flare was also visible upon slit lamp examination. Therefore, because these values do occur in patients with a rhegmatogenous retinal detachment we feel that inclusion of these measurements is mandatory.

  However, excluding patients with a preoperative flare value above 100 pc/ms from our regression and ROC analyses – as is suggested by dr. Schaub and colleagues – did not significantly improve the results from either centre 1 or centre 2. Logistic regression centre 1: OR 1.013; p = 0.446, centre 2: OR 1.035; p = 0.028. Sensitivity and specificity at a cut-off of 15 pc/ms dropped to 78 and 39% in centre 1, and 36% and 77% in centre 2 (p = 0.051).

  A possible explanation for the discrepancy with the previous reports is our low prevalence of postoperative PVR (7.5% and 6.2% respectively) compared to the previous reports. All previous reports included consecutive patients and found prevalences of 10.3%(1), 14.9%(2) and 20% (3). Both the positive predictive value (PPV) and the odds ratio (OR) are affected by the prevalence. This means that even with comparable sensitivity and specificity the PPV and the OR will be lower when the prevalence is lower. In addition, different reports used different definitions of postoperative PVR. We used reoperation due to epiretinal membranes and/or subretinal strands within six months of initial surgery, as did Schröder et al.(1) Hoerster et al. based their findings on PVR grade C at 3 months postoperatively. While ten patients developed PVR only four (6%) needed a reoperation.(2) It is unclear from the report whether all four patients had a flare value > 15 pc/ms but if this were the case this would give a PPV of 20% and OR of 8 instead of the reported 40% and 30. Conart et al. used PVR grade B and C at 6 months as the outcome. Moreover, they included 20% patients with preoperative PVR grade C which influenced the postoperative number of patients with PVR. Both studies included flare values > 100 pc/ms which confirms that high values are not uncommon.(2,3)

  The predictive value of the flare value thus depends on the patient group that is targeted. Our results indicate that there is a large overlap in flare values which makes selecting those patient at high risk for developing postoperative PVR (according to our definition) inaccurate.(4) Dependent on the implications of a positive test result (flare value > 15pc/ms) we should decide whether a false discovery rate of 60-90% is acceptable.

  We agree with dr. Schaub and collegues that identifying patients with an increased risk of PVR would be of great benefit in studies of the treatment of PVR by “enriching” the study population. Unfortunately, in our study, we have not been able to validate preoperative flare measurements for this purpose.
  Recently, however, we have shown that postoperative flare measurements may have that potential (Mulder et al, non published data, presented at the NOG Maastricht, March 29 2017), and we encourage other groups to validate these findings.

  References
  1 Schroder S, Muether PS, Caramoy A, et al. Anterior chamber aqueous flare is a strong predictor for proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment. Retina 2012;32(1):38-42.
  2 Hoerster R, Hermann MM, Rosentreter A, et al. Profibrotic cytokines in aqueous humour correlate with aqueous flare in patients with rhegmatogenous retinal detachment. Br.J.Ophthalmol. 2013;97(4):450-453.
  3 Conart JB, Kurun S, Ameloot F, et al. Validity of aqueous flare measurement in predicting proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment. Acta Ophthalmol. 2016;.
  4 Mulder VC, Kluft C, van Etten PG, et al. Higher vitreous concentrations of dabigatran after repeated oral administration. Acta Ophthalmol. 2017;.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 7 February 2017

  Response to "Preoperative aqueous humour flare values do not predict proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment"

  • Friederike Schaub, MD Department of Ophthalmology, University Hospital of Cologne, Cologne, Germany
  • Other Contributors:
    • Bernd Kirchhof, MD
    • Sascha Fauser, MD

  We noticed the article entitled "Preoperative aqueous humour flare values do not predict proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment" by Mulder and associates with interest.(1)

  Several studies have been published concluding that elevated aqueous flare values seem to be associated with increased risk for PVR redetachment.(2-4) Schroeder et al reported that values >15 photon counts per milliseconds (pc/ms) increases the risk for PVR 16-fold.(4) Hoerster et al showed that the odds ratio for PVR development with preoperative flare values >15pc/ms was 30.7 (p=0.0001) with a sensitivity of 80% and specificity of 79%.(3) Conart et al verified these findings (OR 12.3, p<0.0001 for later PVR in flare values >15 pc/ms).(2)

  In contrast Mulder et al concluded on their data compilation that laser flare measurements are inaccurate in predicting PVR.(1) Logistic regression analyses showed a significant increase in odds with increasing flare at least for the second centre (1) supporting the notion that high flare measurements herald PVR. However, the large variation precluded sufficient sensitivity and specificity to separate between groups. We assume the reason for the large variation is that high-level outliers were included. For center 2 only the highest and the lowest values were excluded, no information is provided for center 1. Values of 100pc/ms, here up to 312pc/ms, are uncommon for the low-level type of i...

  Show More

  We noticed the article entitled "Preoperative aqueous humour flare values do not predict proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment" by Mulder and associates with interest.(1)

  Several studies have been published concluding that elevated aqueous flare values seem to be associated with increased risk for PVR redetachment.(2-4) Schroeder et al reported that values >15 photon counts per milliseconds (pc/ms) increases the risk for PVR 16-fold.(4) Hoerster et al showed that the odds ratio for PVR development with preoperative flare values >15pc/ms was 30.7 (p=0.0001) with a sensitivity of 80% and specificity of 79%.(3) Conart et al verified these findings (OR 12.3, p<0.0001 for later PVR in flare values >15 pc/ms).(2)

  In contrast Mulder et al concluded on their data compilation that laser flare measurements are inaccurate in predicting PVR.(1) Logistic regression analyses showed a significant increase in odds with increasing flare at least for the second centre (1) supporting the notion that high flare measurements herald PVR. However, the large variation precluded sufficient sensitivity and specificity to separate between groups. We assume the reason for the large variation is that high-level outliers were included. For center 2 only the highest and the lowest values were excluded, no information is provided for center 1. Values of 100pc/ms, here up to 312pc/ms, are uncommon for the low-level type of inflammation in primary rhegmatogenous retinal detachment. Therefore, we challenge laser flare values beyond 100pc/ms. Measurements become easily disturbed by background lighting or cells leading to the necessity to exclude measurements falsified by artefacts.

  It would be interesting to know details on the protocol the authors followed to exclude artefacts. Furthermore, we would appreciate an explanation for the unusual variation of the values and the discrepant data towards previous reports.(2,4)

  Reference List

  1. Mulder VC, Tode J, van Dijk EH, Purtskhvanidze K, Roider J, Van Meurs JC et al. Preoperative aqueous humour flare values do not predict proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment. Br.J.Ophthalmol 2017. doi: 10.1136/bjophthalmol-2016-309134.
  2. Conart JB, Kurun S, Ameloot F, Trechot F, Leroy B, Berrod JP. Validity of aqueous flare measurement in predicting proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment. Acta Ophthalmol 2016.
  3. Hoerster R, Hermann MM, Rosentreter A, Muether PS, Kirchhof B, Fauser S. Profibrotic cytokines in aqueous humour correlate with aqueous flare in patients with rhegmatogenous retinal detachment. Br.J.Ophthalmol 2013;97:450-3.
  4. Schroder S, Muether PS, Caramoy A, Hahn M, Abdel-Salam M, Diestelhorst M et al. Anterior chamber aqueous flare is a strong predictor for proliferative vitreoretinopathy in patients with rhegmatogenous retinal detachment. Retina 2012;32:38-42.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top