Editor:Editor’s comment Author’s response Manuscript

changesDr. Schachat: Of course, PPV for these eyes has long be considered to be taboo. I assume due to fear of a greater chance of what? Death or a surrogate marker I assume, for death or at least morbidity, orbital recurrence. What should the safety measure be? Which of these? Note that if you want to worry about both, it gets a good deal more complicated since you would be looking at a composite outcome. See for example, the amusing JAMA paper on composite outcomes, there is no free lunch by Detsky

Dr. Schachat, thank you very much for your interest thoughtful comments, which have refined and enriched in our work.

We think describe death is as the most clinically relevant and serious outcome of retinoblastoma. It is dichotomous and unambiguous. Our primary outcome was survival relative to the eye that received PPV (PPV-ERS). Secondary outcomes were eye salvage and visual acuity following PPV. This is not a composite outcome as discussed in Detsky’s Commentary, but a single primary outcome and 2 secondary outcomes. We elected toalso report both overall survival (death from any cause) and PPV eye related survival (PPV-ERS) which looks at death due to tumor metastasis from the PPV eye. Overall survival in our study may be considered a composite outcome of both deaths related to tumor metastasis from either eye or death from sepsis. Given that the primary objective of this study is to evaluate the safety of PPV, we think PPVERS is the most relevant outcome.

We also independentlyOur secondary outcomes were studied recurrence free survival, eye salvage rate and vision which are also important outcome measures of interest to RB clinicians.

Abstract: lines 53-56

Then, once you decide what your marker is for safety, if it is orbital recurrence, it would take an N of 300 to rule out a 1% rate, an N of 100 to find a 3% rate, etc. What would an acceptable rate be? Thank you so much.

In this study, we reported the overall survival rate and PPV-ERS with 95% confidence interval.

Our study includes N 245 eyes/225 patients. It is hard to tell what an acceptable survival rate isOur primary outcome is PPV-ERS. There is no RCT to that has compared survival with different treatment modalities. There is no “acceptable” rate of death from retinoblastoma. Currently, patients may receive In addition to different treatments for different indications. Of course,, there are also social factors that affect survival which differ between settings. For example, sSome of our died dead patients may have been alive had their parents not abandoned treatment upon tumor recurrence, for example patient 21 in our report [Zhao J, Li Q, et al. Pars Plana Vitrectomy and Endoresection of Refractory Intraocular Retinoblastoma. Ophthalmology. 2018;125(2):320-2. Epub 2017/11/21. doi: 10.1016/j.ophtha.2017.10.015. PubMed PMID: 29153457].. We hope to simply report our survival outcome and let the reader judge what is acceptable to them. Lines 442–447, Results, Survivorship

Reviewer 1:

Reviewer’s commentAuthor’s responseManuscript changesThe first main point is in regards to the surgical procedure. This needs to be presented with much more clarity and detail. Please add more detail on port placement. Was UBM imaging done to prevent the ports being placed through active tumor? Were the ports placed away from tumors in particular manners. Was a peritomy done? Was the sclerostomy site treated with cryotherapy? It states later that Melphalan was irrigated onto the eye - please include here and approximate frequency. Was melaphalan placed subconjunctivally? 'risk of retinal detachment' is subjective - do you mean everytime a retinectomy was done (97% of cases?). when was the lens also removed? Only if tumor was seen in the ac? What does 'extensive tumor removed' actually mean? Were any vitreous samples taken for analysis either at the beginning or the end of the procedure? Please be more granular about when a post PPV dose of intravitreal melphalan was given

We thank the reviewer for his/her thoughtful comments and interest in our manuscript.the important suggestion for more procedural details, now incorporated in Methods.

Per reviewer’s recommendation we have described the PPV procedure in more detail.

A dilated fundus exam with scleral depression was performed prior to PPV. This allowed us to ensure there was no tumor at the sclerotomy sites. The three- port insertion sites were not fixed, we pre-selected pars plana sites that were tumor free. UBM and B-scan were used in eyes with vitreous opacities but was were not routinely performed on all cases.

A peritomy was not done prior to port insertion because we used transconjunctival sclerotomies with 23-or 25-guage system. A peritomy was performed After all active tumors were removed, the. We exposed the scleral surface to was washed it with melphalan-containing irrigation fluid.

TNo, the sclerotomy sites were not treated with cryotherapy.

Melphalan was irrigated onto the ocular surface roughly every 3-5 minutes to prevent corneal drying while providing chemoprophylaxis. Intra-operatively, the any exposed subconjunctival space was irrigated with melphalan after all active tumors were removedPPV before closing. At the end of the procedure, after the sclerotomy sites and conjunctiva were sutured, we also performed a subconjunctival melphalan injection to provide sustained chemoprophylaxis.

Silicone oil was applied whenever a partial retinectomy greater than 3-disc diameters was performed This is was a general ruleguideline. The final decision is was at the operator’s discretion based on how well the retina was stabilized with laser. We did not use silicone oil for eyes with complete retinectomy.

The lens was removed in eyes with cataract or anterior segment seeding or invasion.

Several injections of pProphylactic intravitreal melphalan were were recommended for eyes with diffuse vitreous seedings for as chemoprophylaxis in case of residual vitreous seeds remained in the eye following PPV. Some parents did not follow this recommendation for various reasons (mainly including because they lived far away). This is a retrospective review, not a prospective clinical trial.

We did not collect any vitreous samples for analysis because offor two reasons. First, the intravitreal melphalan irrigation is tumoricidal thereby increasing the chance of false negative result with regard to remaining intraocular tumor activity. Secondly, the vitreous cutter also disrupts the cellular integrity of retinoblastomas. For the above two reasons, we felt concluded the clinical yield of vitreous sample analysis would be low. Perhaps, Tthis topic suggestion could be further explored in future studies. Revised linesMethods: Lines 154-182The second main point is that at points the authors appear to present this as a last effort to save an eye and avoid enucleation (and avoid parents refusing further care) - which is valid - but when the reviewer looks closely at the data it becomes quite clear that all treatments were not exhausted before a PPV was done. Both the number of systemic and intra-arterial chemotherapy cycles are less than a usual standard course, it does not appear that many children received both which would be standard in other places (eg failure of intravenous followed by IAC). VERY few eyes here (6%) received intravitreal melphalan injections despite the fact that it has been shown to effectively augment the salvage of even very advanced eyes with extensive seeding. Thus it appears that PPV were instead planned as part of the salvage regimen or in the very least there was not a high threshold to move to PPV and in fact this was done without completing full courses of chemotherapy. Thisdoes not invalidate results, particularly in the setting of Zhao's paper that prolonged treatment for advanced eyes may increase risk but this needs to be made clear in this manuscript and currently is not.

We agree with the reviewer about theThank you for suggesting that we clarify the indications for PPV. importance of clarifying the indication for PPV.

Per reviewer’s recommendation we included a treatment consort diagram to clarify the pathway to PPV (figure 1). Furthermore, we clarified the definition of residual active disease and recurrent disease in the method section.

The reviewer is correct in pointing out that most patients with residual active disease did not receive the standard course of chemotherapy prior to PPV. We evaluated tumor response at each EUA. If the tumor showed significant regression, we would continued with additional chemotherapy. However, if the tumor showed new growth or no significant regression, we would engage in joint decision making

with parents to decide whether to attempt another cycle of systemic chemotherapy or to pursue intra-arterial chemotherapy, PPV or enucleation. We now clarify ied this point in the method and discussion sections. As such, we observe a variation in the number of cycles of pre-PPV chemotherapy in this retrospective study. We clarify the indications for PPV that were applied to a variety of clinical situations.there was variation in the number of cycles of pre-PPV chemotherapy. The reviewer is also correct in noting that only a small proportion of eyes was were treated with IAC in the period of our study (2013-14). We now included the apparent reasons in our explanation in the introduction. From 2013-2014, only two centers in China offered IAC. We started our IAC program in 2015. We routinely discussed IAC as a treatment option with and if chosen by the parents, we madeour patients and prompt referral were madeto the centers offering IAC. There were also some patients who were referred to us for PPV after poor response to IAC. However,Since the cost of one cycle of IAC cost is about the same as the PPV procedure. Therefore, the long travel distance and high cost deterred many parents from pursing IAC. We acknowledge thatd IVC is the main primary treatment modality for this cohortstudy.

With regard to intravitreal chemotherapy. If the primary tumor regressed following systemic or intraarterial chemotherapy but the eye had vitreous seeds, then we would attemptour standard therapy was intravitreal chemotherapy to treat the seeds. If the vitreous seeds responded to intravitreal injectionchemotherapy, , then, PPV would was not be necessary and the eye would was not be included in this study. If intravitreal injection failed, PPV may be usedwas considered to treat residual seeds. However,Generally in eyes with concurrent active primary tumor and vitreous seeds, we proceeded suggested with PPV, which has potential to . We felt intravitreal chemotherapy may treat the vitreous seeds, but the primary tumor would still need to be treated. In this case we opt for PPV because it simultaneous treat the both primary tumor and the seeds. This retrospective review of patients treated with PPV is not designed to compare efficacy of treatment modalities. We summarized this in the method section.

In this study, PPV was used when there was evidence of poor response to chemotherapy either in the form of residual tumor or recurrent tumor. We have not exhausted full courses of systemic or intra-arterial chemotherapy prior to progressing to PPV for residual active disease. We hope this is now clear in the manuscript.

Most patients with recurrent disease have completed a prior 6 cycles of systemic chemotherapy to establish initial complete regression; PPV was considered after tumor recurrence. Added new figure 1. The original figure 1 is now included as supplementary file.

Definition of residual and recurrent tumors:Revised lines 117-122

No standard course of chemotherapy:Added lines 132-138, 351-355

IAC:Added lines 122-126

Intravitreal chemo:Added lines 126-129

Line 86, make clear any chemo (IAC or intravenous)

R1 Revised as suggestedLine 88: “…following systemic and/or intra-arterial chemotherapy….”Revised line 887that 37.6% of eyes required a complete retinectomy for salvage and are NLP should be in the abstract, it classifies the intensity of disease and of the treatment modalityRevised as suggestedR1 Line 60-61: “The 37.6% of eyes that required a complete retinectomy for globe salvage had no light perception following PPV.”Added lines 59-61Added lines 60-61Please make clear what the authors mean by persistent tumor. In many eyes persistent tumor <3mm can be treated with laser consolidation. Persistent seeding can be treated with intravitreal melaphalan injections. If truly any persistent was an indication for PPV please clarify this. If it was persistent retinal tumor failing local consolidation please make that clear. There should be more clarity regarding indication for PPV here so that other centers have a comparable idea of how to apply these results to their practices.

We introduced a new Fig 1 to clarify standard treatments before PPV in this study. We appreciate reviewer’s attention to detail. Yes, residual active and recurrent tumor < 2-disc diameters were primarily treated with laser or cryotherapy. R1 Lines 119-122, Methods section: “Residual active tumor was defined as persistent active tumor despite primary chemotherapy and focal laser/cryotherapy consolidation (>3 mm any dimension, no significant regression, or vitreous seeds). This is now noted in the new figure 1 and discussed in the method section.

As noted earlier, our approach to intravitreal melphalan is now summarized in the method section.

Small tumor:Added lines 129-131

Intravitreal chemo:Added lines 126-129

Persistent seeding can be treated with intravitreal melaphalan injections. If truly any persistent was an indication for PPV please clarify this. If it was persistent retinal tumor failing local consolidation please make that clear. There should be more clarity regarding indication for PPV here so that other centers have a comparable idea of how to apply these results to their practices.IVit melphalan was used in standard consolidation after chemotherapy; PPV was used as described when repeated IVit melphalan was ineffective.R1 Lines 133-136: “Residual active or recurrent tumor manifest as only vitreous seeds without evident active retinal tumor was treated with intravitreal chemotherapy. Simultaneous presentation of refractory or residual vitreous seeds and active retinal tumor were treated with PPV without prior intravitreal chemotherapy.”

CONTRAINDICATIONS to vitrectomy must be clearly discussed

In this retrospective study, The relative contraindications for PPV followed clinical needs now explicit in are obscured optic disc and foveal invasion. The absolute contraindications to PPV are evidence of optic nerve invasion and extrascleral tumor invasion. R1 Lines 139-142: “Absolute contraindications to PPV were evidence of optic nerve invasion and extrascleral tumor invasion evident on clinical exam, CT or MRI. Relative contraindications were obscured optic disc, foveal invasion or extensive retinal invasion requiring complete retinectomy.” Now summarized in Fig 1 standard treatments prior to PPV.

PPV is not recommended in eyes with foveal invasion or extensive tumor invasion because of expected poor visual outcome. The yield of PPV is low from a risk to benefit ratio perspective. We think good cosmesis can also be achieved through prothesis. PPV is not advised in eye with obscured optic nerve because of potential optic nerve invasion which is not treated by PPV. PPV also cannot be safely performed in eyes with extra-scleral invasion. However, obscured optic nerve does not necessarily means optic nerve invasion, therefore it is a relative contraindication. Eyes with confirmed retrobulbar optic nerve invasion on imaging is an absolute contraindication.

Contraindications:Added lines 145-150.

There should be a section regarding the estimated total dose of melaphalan these eyes are exposed to including the number of times a post PPV dose of intravireal melphalan was given.

Thank you for this good suggestion. R1 Lines 168–174Methods:“After PPV sclerotomy sites and scleral surface were washed with 5 μg/ml melphalan irrigation fluid Overall, approximately 500 mL (25 mg melphalan) was washed through the eye and surgical field. At end of surgery, 20 μg melphalan was injected subconjunctival in the regions of the ports. Total exposure of the eye was approximately 20–30 µg melphalan. Eyes with diffuse vitreous seeds received 2–4 cycles of adjuvant intravitreal melphalan (20 μg/ 0.05 ml) starting one month post-PPV.”We agree it is important to include the dose and amount of melphalan used during infusion and injections. However, we felt the infusion fluid dosing should be listed separately from the intravitreal injections. The eye is only exposed to the infusion melphalan for the duration of PPV (~1.5hr). It is replaced by silicone oil or gas after the procedure. In contrast, intravitreal injections stay in the eye for much longer and have more sustained effect inside the eye. Adding the two doses together may be misleading.

Infusion melphalan amount and dose:Added lines 178-180Please add more information regarding post PPV recurrences that did not require enucleation - how many of these were there? 3 patients required more than one PPV? For what indication (line 208)

Details of post PPV recurrences are provided:R1 Lines 145–176: RESULTS/SUBHEADING Tumor Recurrence Following PPVPer reviewer’s suggestion, we included detail regarding treatment and patient outcome following post-PPV recurrence

The indication for additional PPV were was to treat tumor recurrence.Recurrences:Added lines 252-286The use of 'neoadjuvant' should be considered in light of the main comment #2 above - was this chemotherapy every intended to be curative? Why did some children receive adjuvant therapy? This is not clear.

In this cohort study of 245 eyes, primary chemotherapy with focal therapy consolidaton was intended to be curative, as standard care and was NOT neoadjuvant in preparation for PPV. PPV was exclusively used as a second line therapy when first line had failed, after for either residual active or recurrent tumor. Fig 2 shows exclusion from this study of 14 eyes that had primary PPV.Chemotherapy was not used as a neoadjuvant therapy in the sense of shrinking the tumor prior to PPV. We have started doing this in more recent years as we become more experienced, but this was not the case for the 2013-2014 cohort. We hope this is now clear with the new figure 1 and clarification on the indications for PPV.We have replaced the term “neoadjuvant” with “Pre-PPV” wherever it is usedto describe the treatments prior to PPV (See Fig 1).

As mentioned in the discussionSince, the PPV only treats intraocular tumor but does not offer any systemic chemoprotection for microscopic metastasis, . We further added in the discussion that anterior segment is a common site of recurrence because of limited access with PPV, optic nerve invasion is not treated by PPV and the potential for possible microscopic choroidal seedings leading to recurrences. Therefore, adjuvant systemic chemotherapy may providewas given to children considered at risk for failure of PPV to control all tumor the necessary chemoprophylaxis. After In the present paper wewe reviewed the data in 2017, we see observed a not-significant but potential survival benefit with from adjuvant chemotherapy after PPV. (R1 lines 230–233 “Patients who had adjuvant systemic chemotherapy had 100% (95% CI: 93.5%–100%) 5-year PPV-ERS, not different than those without adjuvant chemotherapy (5-year PPV-ERS, 100% vs 94.8%; P = .102).” therefore, we now routinely offered 2 cycles of adjuvant systemic chemotherapy following PPV. However, from 2013-2014, it was only suggested to parents as an option as we did not have evidence to back up this practice.

Adjuvant chemotherapy:Added liens 376-382Line 175, only 6.5% received IVM despite the significant seeding burden. The authors should state that PPV was the chosen indication for treatment of seeding in this cohort and not IVM

As mentioned, we now explained our approach to intravitreal chemotherapy in the method section.R1 lines 133–136: “Residual active or recurrent tumor manifest as only vitreous seeds without

evident active retinal tumor was treated with intravitreal chemotherapy. Simultaneous presentation of refractory or residual vitreous seeds and active retinal tumor were treated with PPV without pre-PPV intravitreal chemotherapy.”Intravitreal chemo:Added lines 126-129

Line 204Can the authors postulate why there was extension from the non-PPV eye in 3 cases where a child was being extensively followed? That brings up concerns regarding too few cycles of systemic chemotherapy as well as follow-up.

Yes, 3 casesThree patients had extraocular extension from the non-PPV eye.

Patient #1 was diagnosed with bilateral D/D retinoblastoma and. Patient was initially treated with six cycles of systemic chemotherapy. The left eye was successfully treated while the right eye had residual active tumor that was treated with PPV. The right eye developed recurrence 11 months following PPV and was enucleated. Five months after the enucleation, there were was no evidence of active tumor in the left eye, right orbit, and head MRI ruled out any orbital tumor in either eyes. However, 3 months later, the left eye developed extensive tumor recurrence and MRI showed tumor extension into the left orbit, left shoulder and right knee. No tumor in the right orbit. The patient died 3 months later. We think this may be a case of aggressive tumor biology, going from no evidence of active tumor to systemic metastasis in a short 3 months.

Patient #2 was diagnosed with bilateral C/D retinoblastoma. The patient initially received After 7 cycles of systemic chemotherapy, there was residual active tumors remained in both eyes. The left eye had tumor invading the macula, retinal detachment and was enucleated. The right eye received 5 additional cycles of systemic chemotherapy to attempt eye salvage. However, 3 months after last cycle of systemic chemotherapy, the right eye developed massive vitreous seedings. The right eye then received 2 additional cycles intravitreal chemotherapy. However,Since there was were still refractory seeds,s. PPV was performed. Sadly, 1One month following PPV (or 7 months after last systemic chemotherapy), the left orbit developed tumor recurrence and patient died shortly soon after from brain metastasis.

Patient #3 was diagnosed with bilateral C/E retinoblastoma. The patient was initiated on systemic chemotherapy. However, After only 1 cycle of systemic chemotherapytreatment, the patient was lost to follow-up, and. The patient returneds 1 year later. It turned out; theThe child was had been receiving traditional Chinese herbal medicine for 1 year. Upon examination, the left eye developed showed proptosis and MRI identified tumor extension approaching the optic chiasm from the left optic nerve. The tumor in the right eye had increased in size however, the macula was not invadedinvolved. No tumor invasion seen on at the right optic nerve. The parents declined any systemic chemotherapy, radiation or any curative therapy; however, they wished to ensure the child sees before he dies. We explained to the parents that PPV would not save the child’s life because of the extraocular disease; however, parents insistedrequested PPV ensure the child continued to see before he dies.. With ethics board approval from ethics board, we eventually offered PPV for the right eye to to this child to preserve vision under the exceptional circumstance. At the time, we estimated the child would have anywhere from a few months to 1 year of life. The child was then lost to follow-up following PPV. We later phoned the child’s local hospital and found out tThe child died 3 months following PPV from brain metastasis from the non-PPV eye.

In retrospective, we think These three deaths were not due to insufficient chemotherapy. as oppose to bad tumor biology, danger of orbital recurrence and issue of treatment compliance. Although, these three cases are interesting informativein their own rights, we do not think they should be discussed in detail in the current manuscript.

Line 206: if the extension site cannot be determined for 3 children then the sclerostomy site cannot be ruled out and the authors should acknowledge that.

Included per reviewer’s recommendationR1 lines 333-336: “Given the lack of histopathologic evidence and 3 cases of metastatic death with unknown extension sites, the present study cannot definitively exclude tumor seeding through the sclerotomy sites.”Sclerotomy site cannot be ruled out:Added lines 339-341Please include a section regarding histopathology of the post PPV enucleated globes with specific attention to tumor at the sclerostomy site histopathologically (not just port sites analyzed clinically)

During the pathology sectioning, the Pathological examination did not specifically contain sections of containing the sclerotomy sites was not obtained for this cohort of patients. Therefore, we cannot comment on the histopathology of the sclerotomy sites. We acknowledged this in the discussion. We wish to discuss about the histopathology in detail in future papers. R1 lines 332-3: “Histopathologic evaluation of sclerotomy sites was not performed because the standard histopathology sections did not all contain the sclerotomy sites.”

We included the AJCC histopathology staging of the enucleated eyes in the result section.Sclerotomy site not sectioned:Added lines 337-339

AJCC histopathology stagingAdded lines 296-298Line 231: visionFrom my calculation 55% of the eyes in this study are NLP (92+31+14) either from complete retinectomy(92) or enucleation. This needs to be stated. The remainder of the paragraph is very transparent and quite good.

Included per reviewer’s suggestionR1 lines 310-1:“Overall, 55.9% of eyes are NLP because of complete retinectomy or poor vision following partial retinectomy or enucleation.”% with NLP:Added lines 316-317Line 261 - the question of not dissecting the conjunctiva and going through the conj and sclera together may in fact increase the risk without the other safety measures of continuously irrigated melaphalan and this should be pointed out. In many places even post therapy the ports would not be placed through the conj and sclera together and instead a peritomy would be made to prevent fluid from moving into the subconjunctival space - significant surgically induced chemosis is known to occur during even small port PPV which without melphalan could be an issue. Please address.

The transconjunctival sclerotomy with insertion of a cannula allows intraocular content to be directly channeled to the outer conjunctival surface. The cannula act as a physical barrier to insulates the subconjunctival space and scleral wall from tumor seedings. We would agree that cannula is not full proof. For this reason, we also wash the subconjunctival space with melphalan and give subconjunctival injection.

Safety precautions to reduce risk of subconjunctival seeding:Added lines 169-171, 177-178.Line 272 please define IRSS stage I for no RB physicians

Included per reviewer’s suggestionRevised line 359Line 285 a line about the other eyes non PPV advancing and leading to orbital relapse is needed here

As discuss earlier, we do not think the orbital extension of the non-PPV eye is related to insufficient use of systemic chemotherapy. The discussion of those deaths would touch on other topics that is not directly relevant to this already lengthy paper.

That the recurrent eyes did worse merits further discussion - this also goes along with major point #2 above. Seems in fact that the eyes wherein this treatment was planned (or more planned) did better.

All 245 eyes in this study are planned PPV in the sense that we were expecting tumor inside the eye and employed safety-enhanced technique for each eye.

We think the survival difference between PPV for residual active tumor and recurrent tumor were more related to parent’s attitude toward enucleation. Those with recurrent tumor spend more time, money and were more committed in saving an eye. This is well illustrated by two patients who received 3 PPV. Both repeatedly refused enucleation and the children died with an un-enucleated diseased eye.

Recurrent vs residual survival disparity: added lines 386-392 Line 306 first, it does not seem that all of these eyes failed standard of care and second with 55% of eyes NLP, it is neither correct nor fair to say a large proportion with good potential for vision. That is over emphasizing these results, which stand on their own. Please modify this sentence.

We agree with the reviewer to let the results to speak for themselves. We deleted the sentence.Deleted original line 306Line 317 please remove 'not observed' unless all eyes can be analyzed clinically and histoapthologically.

Revised per reviewer’s recommendationDeleted original line 317Line 322 uniform surgical protocolsThis last paragraph needs to be very clear that attempts to do this without safety measures, continual melphalan infusion, and by a dedicated and highly trained surgical retina and RB team and appropriate consent from the family has potential to be dangerous. In fact, I'd prefer the authors to directly state: In all of these eyes the safest approach is to enucleate the eye, however, in certain cases where the parents refuse this approach, a safety enhanced PPV may be considered to salvage the eye in a

multidisciplinary center. In this paragraph please reiterate the contraindications to vitrectomy that were added to the methods. This is also important for other centers to understand and to have available to them in the case of parents refusing enucleation and requesting vitrectomy instead.Please include median age at time of vitrectomyWe agree with the reviewer and implemented the suggested changes.

We included median age at PPV.Conclusion:Revised lines 425-435

Age at PPV:Added lines 204-205

Please consider adding a treatment consort - the reviewer realizes these are complex eyes but the stepwise progression of care is not clear in this manuscriptAdded new figure 1 per reviewer’s suggestion

Added new figure 1The Kaplan meier curves are very low resolution and hard to read but it appears there is quite a large loss to fu in the first year. That needs to be addressed in the legend as well as in the results and in discussion of mortality. If these children truly are lost to fu after 1 year there are hypothetically other unknown cases of mets.

Instead of 4 graphs per figure, we have reduced each figure to include only two graphs. We expect the text to be larger and more readable now.

We agree with the reviewer that there is a large number of loss to follow-up patients in the first year after PPV. However, the loss rate is lower compare to the pre-PPV era. In our setting, we routinely see parents refuse enucleation after eye salvage therapy has been exhausted and abandon treatment altogether. Instead of letting patients be lost with active tumor inside their eye, PPV likely have saved the lives of some children who failed chemotherapy.

With that said, we acknowledge that there may be additional unknown cases of metastasis in the lost to follow-up cases. We added an estimate of worst-case scenario by assuming all 21 lost children were died.

Finally, the number at risk row under the KM graph already show how many children were followed at given time intervals.

Unfortunately, most hospitals seeing retinoblastoma patients in China are located in major urban centers. As such, follow-up is costly both in time and money for many families who live in rural areas. Through the publication of this paper, we hope to emphasize to more parents the importance of post-PPV follow-up given 30% of eyes develop tumor recurrence after the procedure.

Revised figures 4 and 5

Worst-case scenario:Added lines 233-235

Discussion of lost to follow-upAdded lines 400-407

Figure 4 is great, please include an eye with a complete or a large retinectomy as well. It is imperative that the authors are transparent about the outcomes of these eyes and this photo is clearly from one of the best outcomes (which should also be included, but not as the only photo)

This is an excellent suggestion. We added a new figure 6 as recommended by the reviewer, which describes a child who had complete retinectomy. The eye subsequently developed anterior chamber recurrence and was enucleated.Added new figure 6.

Given a limit of 6 attachments (tables + figures), we moved original figure 1, table 1 and table 2 as supplementary files

Reviewer 2

Reviewer’s commentAuthor’s responseManuscript changesPage 8, line 108 and 111: Definitions of "residual active disease" and "recurrent disease" are not mutually exclusive with recurrent disease indistinguishable from residual active disease if the interval between last therapy and control without treatment is only one month. The authors have to specify the interval between exams. In addition residual active tumor definition is not stringent enough as it includes true resistant/persistent disease as well as complete response following chemo-reduction therapy. The tumorectomy, performed concomitantly to PPV, offers an ideal way to determine a posteriori the active or inactive nature of the tumor sample. Please explain why the histo-pathology of the tumor was not done? Also the authors should specify where the persistent disease is located as well as the non-calcified nature of the tumors in order to exclude all type I regression. If the active residual disease after systemic chemotherapy is intravitreal or sub-retinal, it is amatter of adjusting the route of administration to target the disease correctly before moving to PPV, i.e. intravitreal chemo for vitreous seeds and intra-arterial for sub-retinal seeds.In her paper Francis J. is providing a usuful definition preventing the confusion between resistant and recurrent disease: "Treatments occurring after 3 months from the completion of initial treatment were defined as recurrence or persistence. Recurrence was defined as new tumor activity visible on ophthalmoscopic exam, including regrowth from regressed tumor and/or new subretinal or vitreous seeds. Persistence was defined as stable, non-calcified tumor that demonstrated no growth but was treated at the the clinician's discretion." (cf Francis JH, Roosipu N, Levin AM, Brodie SE, Dunke, IJ, Gobin YP, Abramson DH, 2018. Current treatment and treatment-free follow-up at a monthly examination.of bilateral retinoblastoma: the impact of intraarterial and intravitreous chemotherapy. Neoplasia 20, 757-763)

We thank the reviewer for the thoughtful and detailed responses. The aim of this manuscript is to provide an overview of survival, eye salvage and vision outcome following PPV. Unfortunately, we have not documented all details of these children’s clinical care. We agree many details the reviewers asked for are important. We have included some additional details that we collected, but for questions we cannot address with the current cohort, we will write about them in future works.

First, we agree with the reviewer that the frequency of follow-up is important in distinguishing between residual active and recurrent tumor. We now included a follow-up schedule in the manuscript.

We have reviewed the manuscript by Francis et al. and revised our definition of residual active disease and recurrent disease to be more precise.

We did not collect any vitreous samples for analysis because of two reasons. First, the intravitreal melphalan irrigation is tumoricidal thereby increasing the chance of false negative result with regard to tumor activity. Secondly, the vitreous cutter also disrupts the cellular integrity of retinoblastomas. For the above two reasons, we felt the clinical yield of vitreous sample analysis would be low.

Unfortunately, we did not consistently document the location of tumor or presence of subretinal seeds at the time of PPV.

We revised the definition of residual active disease which refers to non-calcified and partially calcified tumor. Complete calcification (type 1 regression) is not considered residual active disease.

Only a small proportion of this cohort were treated with IAC (16.3%) due to issue with accessibility. We now included our explanation in the introduction with regard to IAC. From 2013-2014, only two centers in China offered IAC. We started our IAC program in 2015. We routinely discussed IAC as a treatment option with our patients and prompt referral were made. There were also some patients who were referred to us for PPV after poor response to IAC. However, one cycle of IAC cost about the same as the PPV procedure. The long travel distance and high cost deterred many parents from pursing IAC. We now acknowledge in the manuscript that the predominant primary treatment modality in this cohort was systemic chemotherapy.

With regard to intravitreal chemotherapy. If the primary retinal tumor regressed following systemic or intraarterial chemotherapy but the eye has vitreous seeds, then we would attempt intravitreal chemotherapy to treat the seeds. PPV may be used if intravitreal chemotherapy fails. In the scenario where there was retinal tumor that was refractory to systemic or intra-arterial chemotherapy and there were vitreous seeds, we felt intravitreal chemotherapy may treat the vitreous seeds, but the primary tumor would still need to be treated. In this case we opted for PPV because it simultaneous treat the primary tumor and the seeds. We now added this information in the method section.

Follow-up schedule: Added lines 132-136

Definition of residual active and recurrent disease:Revised lines 117-122

IAC:Added lines 122-126

Intravitreal chemo:Added lines 126-129

Why exclude 14 primary PPV, they are of immense interest for the estimate of mortality.

The indication of primary PPV patients were mainly due to concomitant presence of ocular diseases that obscures view of fundus and retinoblastoma. In 2013-2014 time period, the main indication for PPV was to serve as a second line therapy for residual or recurrent tumor following chemotherapy. All 245 eyes included in this study are for this indication. We hope to write about primary PPV in a separate paper in the future.

Page 9, line 123: patient's choice should read parent's choice

Agreed. Revised per reviewer’s suggestion. Revised line 150

Page 9, line 124: what does "incompletely responds" means? Is it at the retinal, subretinal or vitreal level? Regression type II and III are not necessarily incomplete response.

We revised definition of residual active disease to include noncalcified or partially calcified tumor. We think the current phrasing is inclusive of subretinal, retinal, epiretinal and vitreous tumors. We also acknowledged that there is a component of clinician’s discretion.

Definition of residual active and recurrent disease:Revised lines 117-122

Page 9, line 125: what does "full course of chemotherapy" exactly means in the light of the comments regarding page 11, line 173/174 (cf below)?

We agree the phrase “full course of chemotherapy” is misleading and may be interpreted as delivering six cycles of chemotherapy, which is not the case for many eyes in this cohort. We wished to convey that chemotherapy was not given as an adjunct therapy immediately prior to PPV. The indication of chemotherapy was curative for all 245 eyes. We have now removed the phrase “full course of chemotherapy”.

Furthermore, we clarified in the manuscript that tumor response to chemotherapy was assessed at each EUA. If the tumor shows good regression, we would continue more cycles. If tumor does not show significant regression, progressed on serial EUA or developed vitreous opacities, we engage in joint decision making with the parents as whether to pursue additional systemic chemotherapy, IAC, PPV or enucleation with discussion on the risk and benefit of each option. Deleted original line 125

No fixed cycles of chemo:Added lines 132-138, 352-355

Page 9, line 127: in case of severe bone marrow suppression, did the authors consider to do intra-arterial or another appropriate targeted mode of chemotherapy before moving to PPV?

As mentioned above, IAC was not easily accessible to our patients. We now acknowledge in the manuscript that the main treatment modality was systemic chemotherapy.

We removed bone marrow suppression as an indication for PPV because it did not appear independently from other indications for PPV.Systemic chemotherapy: Added line 122

Page 9, Line 132: What was the nature of the "salvage chemotherapy"? Does it include targeted chemotherapy? What was the regimen of "salvage chemotherapy"?

The indication of chemotherapy was curative. The regimen of systemic chemotherapy was uniform across the three centers, which we now include in the method section.Added lines 139-142

Page 10, line 147: "Intravitreal melphalan (20 microg/0.05 ml) was given after PPV if extensive tumor was removed". When was this injection performed? What kind of precaution regarding the melphalan dose was taken in presence of silicone oil? What kind of precaution was taken in lentectomized eyes with silicone? What kind of additional measures were taken in eyes with aqueous seeding and no silicone? What was the outcome of eyes with anterior segment invasion?

We now clarified if there were diffuse vitreous seeds in the eye, we typically recommended 2-4 cycles of monthly prophylactic intravitreal melphalan, with the first dose given 1-month post-PPV.

Unfortunately, there is currently no available data on the effectiveness of melphalan in an eye with silicone oil. We still chose to offer melphalan injections to these eyes. To minimize silicone oil leaking, we applied cryotherapy after the injection.

In lentectomised eyes, we create an inferior peripheral iridotomy. For anterior segment seeding, a lensectomy was performed and tumor aspirated.

We have not collected data on which eyes had anterior segment invasion for this cohort. We have recently started collecting data on PPV in retinoblastoma eyes with hypopyon and anterior segment invasion. We hope to discuss about this topic in future manuscript.

Intravitreal injections:Added lines 180-182

Lentectomised eye with silicone oil: Added lines 175-176

Anterior tumor:Added lines 167-168Page 11, line 157: What were the criteria to exclude "death with MRI evidence of tumor spread from the non-PPV eye"? How many patients have been excluded?

Of 225 patients, there was 12 death in total. We did not exclude any death cases; the 12 death were classified as related to PPV eye or not related to PPV eye.

There were 4 deaths that were not related to the PPV eye. Among which 1 died of sepsis secondary to neutropenia. The 3 other patients had MRI of the head and showed tumor in the orbit of the non-PPV eye and not in the orbit of the PPV eye. Detail about those three patients is as followed:

Patient #1 was diagnosed with bilateral D/D retinoblastoma. Patient was initially treated with six cycles of systemic chemotherapy. The left eye was successfully treated while the right eye had residual active tumor that was treated with PPV. The right eye developed recurrence 11 months following PPV and was enucleated. Five months after the enucleation, there were no evidence of active tumor in the left eye, right orbit, and head MRI ruled out any orbital tumor in either eyes. However, 3 months later, the left eye developed extensive tumor recurrence and MRI showed tumor extension into the left orbit, left shoulder and right knee. No tumor in the right orbit. The patient died 3 months later.

Patient #2 was diagnosed with bilateral C/D retinoblastoma. The patient initially received 7 cycles of systemic chemotherapy, there was residual active tumors in both eyes. The left eye had tumor

invading the macula, retinal detachment and was enucleated. The right eye received 5 additional cycles of systemic chemotherapy to attempt eye salvage. However, 3 months after last cycle of systemic chemotherapy, the right eye developed massive vitreous seedings. The right eye then received 2 additional cycles intravitreal chemotherapy. However, there was still refractory seeds. PPV was performed. Sadly, 1 month following PPV or 7 months after last systemic chemotherapy, the left orbit developed tumor recurrence and patient died shortly later from brain metastasis.

Patient #3 was diagnosed with bilateral C/E retinoblastoma. The patient was initiated on systemic chemotherapy. However, after only 1 cycle of treatment, the patient was lost to follow-up. The patient returns 1 year later. It turned out; the child was receiving traditional Chinese herbal medicine for 1 year. Upon examination, the left eye developed proptosis and MRI identified tumor extension approaching the optic chiasm from the left optic nerve. The tumor in the right eye increased in size however, the macula was not invaded. No enhancement of the right optic nerve on MRI. The parents declined any systemic chemotherapy, radiation or any curative therapy; however, they wished to ensure the child sees before he dies. We explained to the parents that PPV would not save the child’s life because of the extraocular disease; however, parents insisted. With approval from ethics board, we eventually offered PPV to this child to preserve vision under the exceptional circumstance. At the time, we estimated the child would have anywhere from a few months to 1 year of life. The child was then lost to follow-up following PPV. We later phoned the child’s local hospital and found out the child died 3 months following PPV from brain metastasis.

Revised lines 243-245Page11, line 159: The fact that 21 patients were lost of follow-up is ignored in this PPV-ERS estimate. In consequence it cannot be stated that the survival estimate is conservative! Please change accordingly.

We want to clarify that the 21 patients who were lost of follow-up were censored at last follow-up. Kaplan Meier method estimates the probability of survival at any given time point, taking into account of patients who did not experience the event of interest. In this study, patients who are lost to follow-up or patients who survived until the end of the study (June 16, 2019) were censored at last follow-up. The survival of the censored patient is considered ONLY as long as he/she is followed. Kaplan Meier method takes into account both censored and uncensored observations in estimating survival.

We originally used the phrase “conservative survival estimate” because for 3 deaths in which the site of tumor extension was undetermined, we attribute them to the PPV eye as oppose to the non-PPV eye. However, we agree to remove the sentence.

We now included a survival estimate assuming all 21 lost patients as died.Removed original line 159.

Assume all lost patients as diedAdded lines 233-235Page 11, line 163 "IIRC staging" should read IIRC grouping across the whole manuscript (see page 11 line 171…etc)

Agreed. Revised per reviewer’s suggestion.Revised across the manuscript

Page 11, line 164: "same staging" should read same grouping. How many patients had PPV in both eyes with the same grouping?

14 patients had bilateral Group D eyes received PPV in both eyes. We added this information to the result section.Added lines 206-207In general this section is omitting important data, namely:- Number of eyes with silicone oil tamponade- Number of eyes with concomitant lentectomy- Number of eyes with anterior segment seeding- Number of histo-pathologic exams of tumoral material removed during PPV (tumorectomized eyes) and proportion of samples with viable tumor- Number of secondary enucleations post PPV and proportion of enucleated eyes with residual active tumor (retinal, subretinal, vitreal, aqueous) and histopathologic risk factorsPlease make sure that all these above-mentioned points are addressed in your response.

We now included information on number of eyes with silicone oil tamponade, lensectomy, histopathologic staging of enucleated eyes.

Unfortunately, we have not recorded information on anterior segment seeding at time of PPV. As mentioned above, analysis of aspirated tumor was not performed.

We have information on the histopathology risk features; however, we think detailed discussion of histopathology is outside the scope of this manuscript. For this cohort, we have not obtained the proper pathology sections to assess whether there were any tumor seeding in the sclerotomy sites. We hope to write address the pathology following PPV in detail in a focused paper in the future. Silicone tamponade:Added lines 227-228

Lensectomy: Added lines 224

Histopathologic staging: Added lines 298-300

Histopathologic sclerotomies sites not performed:Added lines 339-341Page 11, line 171/Table 1: during the 22 months of the study period, 262 eligible patients were treated by PPV. How many new retinoblastoma cases were seen in these 3 centers during the same period?

We hope to clarify that the 262 patients were managed at 29 hospitals across China by our retinoblastoma team. Only 3 centers have the equipment and approval to perform PPV for retinoblastoma. So, we referred patients from across China to these 3 centers for PPV and continue their care at another center closer to home. J.Z sees approximately 300 new patients each year across the 29 centers. However, the patients described in this manuscript may not be new patients. Although they were treated with PPV in 2013-2014, they may have been diagnosed and started on chemotherapy earlier.

Page 11, line 173/174: systemic neoadjuvant chemotherapy was given to 234 patients (95.5%) and neoadjuvant intra-arterial to 40 (16.3%), i.e. a total of 274 neoadjuvant chemo! Please explain. Also among the 245 eyes of the study, 210 (86%) were group D or E, but only 16 received intra-vitreal chemo and apparently only 68 (?) benefited from IAC. Please explain.

234 patients received systemic chemotherapy and 40 received IAC. The two groups are not mutually exclusive as there were patients who received both. We regrouped patients into systemic chemotherapy only, IAC only and both. We hope it is now clearer.

With regard to IAC and intravitreal chemotherapy, we hope we have now clarified our approach with intravitreal chemotherapy and the accessibility issue with IAC. Revised lines 207-210

Page 11, line 177: please specify which radio-isotope?

One patient was treated with I125.Revised line 212Page 12, line 178/179: the reported median time between diagnosis and PPV of 3 months for residual active disease strongly suggests that PPV was scheduled after only 3 courses of chemotherapy. This timing usually coincides with the administration of the first consolidation focal treatments. In other words it looks like PPV was given instead of focal therapies which are not mentioned in the methods (with the exception of intra-vitreal injections)!

In this cohort of 245 eyes, primary chemotherapy was intended to be curative. PPV was exclusively used as a second line therapy after either residual active or recurrent tumor. Chemotherapy was not used as a neoadjuvant therapy in the sense of shrinking the tumor prior to PPV. We have started doing this in more recent years as we become more experienced, but this was not the case for the 2013-2014 cohort. We hope this is now clear with the new figure 1 and clarification on the indications for PPV.

We have replaced the word “neoadjuvant” with “pre-PPV” wherever it appears in the manuscript.

We recognize the potential for confusion, we now clarified that we evaluated tumor response at each EUA to decide subsequent treatment.

Finally, we typically give focal therapy at every other cycle of chemotherapy. PPV did not replace laser consolidation. This is now added to the manuscript.See new figure 1

PPV was not scheduled:Added lines132-133

Laser schedule:Added lines 141-142

Page 12, line 181: 21 patients were lost of follow-up, i.e. close to 10% of the entire cohort. It is unbelievable that a citizen can disappear in China. In order to elucidate the number of death in this group, we urge the authors to perform a simple search in the registry office of all deceased children belonging to this list of 21 kids.

We have attempted to contact the parents of these 21 children on multiple occasion and was unable to reach them. All we have are the contact information they provided to the hospital. There is no government agency in China to keep track of lost to follow-up patients. Unfortunately, in low- and middle-income countries lost to follow-up in pediatric cancer is not uncommon.

Interestingly, in a study by Vasquez et al., “Factors Associated with Abandonment of Therapy Diagnosed with Solid Tumors in Peru”. They found retinoblastoma treatment abandonment is more common than any other pediatric solid tumors.

Page 12, line 186/187: 62 eyes with vitreous or "retinal" seeding" (sic). What is "retinal seeding" referring to? I presume it is corresponding to subretinal seeding?

By retinal seeding, we meant vitreous seedings that have accumulated on the retina surface. Unfortunately, we have not consistently document subretinal seeding for this cohort.

We apologize for the confusion, we now replaced retinal seeding with epiretinal seeding to avoid confusion. The distinction between epiretinal and vitreous seeding were made because of management differences with regard to PPV. Epiretinal seedings may be treated with laser or endoresection. Vitreous seedings were aspirated.

Page 12, line 187: Page 12, line 188: 58 patients (23.7%) received post-PPV adjuvant chemo. Please comment on the indications to adjuvant chemo - was the regimen the same as neoadjuvant? why such a big variation in the number of courses (1-5)?

We mentioned in this discussion that a limitation of PPV is that it only treats intraocular tumor but does not offer any systemic chemoprotection for microscopic metastasis.

We further included in the discussion that anterior segment is a common site of recurrence because of limited access with PPV, optic nerve invasion cannot be managed by PPV and the potential for residual choroidal seedings which leads to tumor recurrence. In the 2013-2014, adjuvant chemotherapy was only suggested to parents as an option as we did not have any evidence of survival benefit. However, after we reviewed the data in 2017, we saw a survival benefit with adjuvant chemotherapy, as seen in this data. We now routinely offered 2 cycles of adjuvant systemic chemotherapy following PPV.

Page 12, line 214: Does tumor location (retinal vs subretinal vs vitreal) impact eye retention rates and recurrence free eye salvage? Please provide missing data.Page 13, line 215: 72 eyes relapsed post PPV. Please specify the location of relapse (retinal, vitreal, subretinal, aqueous).

We do not have data on tumor location at time of PPV.

We included data on the location of tumor recurrences post-PPV.Recurrence locations: Added lines 258-265Page 13, line 218: 19 eyes had a second PPV. Can you specify the indications and results in this sub-group.

We now included detail about management and outcome of post-PPV recurrences. These 19 eyes had additional PPV to treat tumor recurrenceRecurrences:Added lines 266-288Page 14, line 232: 92 eyes had complete retinectomy. What is the reason of subsequent enucleation in 14 of them? Also what is the ocular outcome of the 78 conserved eyes with complete retinectomy (phthisis bulbi...etc)?

12/14 eyes with complete retinectomy was enucleated because of tumor recurrence. We included locations of tumor recurrence for eyes with complete and partial retinectomy.

One eye was enucleated because of phthisis bulbi and another one was prophylactically removed at another center, because the physician convinced the parents that PPV is dangerous. In these two eyes, there was no tumor on histopathologic examination.

We included more details about ocular outcome following PPV.Recurrence locations: Added lines 258-265

Enucleation of eyes with complete retinectomy:Added lines 346-350, 266-288

Ocular outcomes:Added lines 294-298Page 14, line 264: The risk of sub-Tenonian or sub-conjunctival contamination remains and cannot be nullified, due to the unavoidable reflux at the time of cannula removal and cannot be trivialize. Please amend the statement accordingly.

We revised the wording. We think it is fair to say cannula reduces the risk of subconjunctival and scleral seeding.

Furthermore, we now included more detail in the procedure section regarding safety precautions during PPV. We irrigate the subconjunctival and sub-Tenonian space with melphalan and deliver a dose of subconjunctival melphalan injection at the end of the procedure.Cannula:Revised lines 336-338

Subconjunctival chemoprotection:Added lines 169-171, 177-178.

Page 16, line 280: "enucleation is commonly regarded as the most conservative treatment for intraocular retinoblastoma." The adjective "conservative" usually designed eye preserving management of intraocular retinoblastoma. Please modify accordingly.

Per reviewer’s recommendation, we have changed the word conservative to safest. Revised line 370Page 16, line 282: The survival rate reported does not take into account the patients lost of follow-up. The authors should also give a presumed survival rate considering the worth scenario, i.e. all 21 patients lost of follow-up actually died secondary to PPV. Ignoring this possibility is introducing an enormous bias leading to the false conclusion that 5-year PPV-ERS is comparable to primary enucleation survival. Please comment.

Per reviewer’s suggestion. We included the worst-case scenario survival by assuming all 21 lost patients as died for reader’s reference. With that said, we think it is not appropriate to compare PPV worst-case scenario with reported overall survival of other studies of enucleation. In our paper on enucleation, which we compared to, children who were lost to follow-up were also censored at last follow-up.

Unfortunately, most hospitals seeing retinoblastoma patients in China are located in major urban centers. As such, follow-up is costly both in time and money for many families who live in rural areas. We do not think all children who were lost to follow-up are died. In fact, some may choose to not follow-up because there is no recurrence. On multiple occasions, we see patients who we thought were lost to follow-up return to care after their community physician identified recurrence on eye examination.

Although not the focus of this paper, one of the biggest benefitsof PPV is that many parents who would otherwise give up treatment were willing to try PPV as a last resort. In our setting, we routinely see parents refuse enucleation after eye salvage therapy has been exhausted. Although some tumor may recur post-PPV, these children undoubtedly have much better chance of survival than those who abandon treatment with active tumor.

Worst-case scenario survival:Added lines 234-236.

Lost to follow-up discussion: Added lines 402-409Page 15, line 262: the quoted reference is not supporting reduction of tumor seeding risk.. The sentence should be modified accordingly.We have removed the citation. We think it is still reasonable to say the use of cannula, which acts as a physical barrier, reduces the risk of subconjunctival and scleral wall seeding. We did not say it eliminate the risk of seeding.

We further included more detail in the method section outlining the safety precautions we took during PPV.Cannula:Revised lines 336-338

PPV procedure:Revised lines 154-182Page 18, line 317: "…not observed a single case of tumor extending out of the sclerotomy sites" this is too strong a statement since it is specified page 15, line 264/265 that "Histopathologic evaluation of sclerotomy ports was not performed in this cohort."

We agree with the reviewer and deleted this comment.

Deleted original line 317.