**5. Conclusions**

In an international multicenter randomized clinical study on patients su ffering from severe dry eye disease, 0.15% high molecular weight hyaluronan (HMWHA) eye drops have been compared with lubricant eye drops individually selected as optimum therapy. HMWHA eye drops have shown superior potential to significantly ameliorate symptoms including discomfort and pain, as well as visual instability without a ffecting dry eye signs.

**Author Contributions:** See Appendix A. Authors have confirmed that format for that Section.

**Funding:** The HYLAN M study received unrestricted funding from i.com medical GmbH, Munich, Germany.

**Acknowledgments:** The principle investigators wish to acknowledge the grea<sup>t</sup> support by their study teams, in particular Bernadette Matijak-Kronschachner and Andrea Heidinger in Graz, Semra Akkaya Turhan in Istanbul, Anoud Saati and Nouf Aljwaiser at PSMMC, Jose Vargas, Gharam Al Zahrani and Sarah Al Harbi at KKESH, Anita Koschmieder in Rostock, and Laurence Quérat and Christine Åkerstedt in Stockholm.

**Conflicts of Interest:** The study director Wolfgang G.K. Müller-Lierheim is also CEO of the company i.com medical GmbH, Munich, Germany. The remaining authors declare no conflict of interest.

#### **Appendix A. Study Centers, Administrative Structure of the Study, and Author Contributions**

Eleven study centers in eight countries with different climate zones and different ethnicities participated in the HYLAN M Study:


The HYLAN M Study was designed in accordance with the international standard ISO 14155:2012 by the sponsor CORONIS GmbH (Munich, Germany) in close cooperation with the PIs Christophe Baudouin, Gysbert van Setten, and Jutta Horwath-Winter, the representatives of the reading centers, Daniel Böhringer for RC1 and Oliver Stachs for RC2, the biostatistician Sébastian Marque (IQVIA, France), and the randomization center (C2R, Paris, France). CORONIS is operating under a quality managemen<sup>t</sup> system including clinical research, which was certified according to the international standard 13485:2016 and annually supervised by the European notified body mdc medical device certification GmbH (Stuttgart, Germany).

The representative of the sponsor, Wolfgang G.K. Müller-Lierheim, had overall responsibility as study director (SD). In obtaining local ethics committee approval, registering the study with national authorities, and monitoring the study, he was supported by


The medical scientific managemen<sup>t</sup> of the HYLAN M Study was under the overall responsibility of the coordinating investigator (CI) Gysbert-Botho van Setten, who was supported by Jutta Horwath-Winter with respect to the lissamine green staining procedure and diagnostic judgement, and Oliver Stachs regarding confocal microscopy.

The HYLAN M Study used detailed work instructions including the diagnostic procedures corneal fluorescein staining (see Appendix D), tear film break-up time, lissamine green staining (including assessment of lid wiper epitheliopathy and mucocutaneous junction), tear film osmolarity using the TearLab osmolarity system (TearLab Corporation, San Diego, CA, USA), and taking and evaluating confocal microscopy images of the subbasal nerve plexus using the Heidelberg Retina Tomograph HRT in combination with the Rostock Cornea Module RCM (Heidelberg Engineering GmbH, Heidelberg, Germany).

Paper case report forms (CRF) were used by the study centers throughout the study and were archived by each study center for a minimum of 10 years beyond the completion of the HYLAN M study. The contents of the case report forms were anonymously transferred to electronic case report forms (eCRF) using a web-based system allowing electronic transfer to the HYLAN M study database hosted at the University Eye Hospital in Freiburg, Germany. For monitoring purposes, the SD had online access to the HYLAN M database. Printouts of the eCRFs were used to assure the correct and complete data transfer from the paper CRFs to the eCRFs during regular monitoring visits of the study centers. At the closure visits of each study center, printouts of all eCRFs were signed by the responsible PI and provided to the sponsor who scanned and electronically archived the signed eCRFs. After the closing the last study center and monitoring the reading centers and the host of the study database, the database was locked on 26 March 2020, and exported for statistical analysis.

The company C2R (Paris, France) under the responsibility of the data managers Pascale Croix and Mélinda Ezzedine provided the electronic block randomization of patients of the HYLAN M study in two equally sized study arms A (control group) and B (Comfort Shield group = verum group). The investigators were kept unaware of block randomization and the block size of four. For the randomization, the module CSRandomization of the ENNOV CLINICAL ® software (version 7.5) was used. The clinical data managemen<sup>t</sup> system had been developed and validated to meet all regulatory requirements of data managemen<sup>t</sup> and create a non-modifiable audit trail. The study centers had password-protected secure CS Online internet interface access to the randomization tool. After entering the patient number, the system automatically created the result of randomization. Then, the study centers printed the patient number and result of randomization and filed this together with their CRFs to be controlled during monitoring visits.

The HYLAN M Study used two masked reading centers for the analysis of digital images provided by the study centers. The cornea reading center located at the Eye Center, University Hospital Freiburg, Germany, in the text referred to as RC1, analyzed the corneal fluorescein staining (see Appendix D). Oliver Stachs (Department of Ophthalmology, University Medical Center, Rostock, Germany), in the text referred to as RC2, analyzed the subbasal nerve fibers in confocal microscopy images, which were taken as an additional optional diagnostic test at the baseline and week 8 visits of the HYLAN M study (the results of this subgroup analysis will be published separately).

The statistical analysis plan (SAP) for the HYLAN M Study was developed by IQVIA, Bordeaux, France in cooperation with the SD. IQVIA performed the statistical analysis by using the software R version 3.5.3 (The R Foundation for Statistical Computing, Vienna, Austria) after database lock and provided the statistical analysis report (SAR) of the HYLAN M study. Upon request, the sponsor of the HYLAN M study will provide a copy of the SAR.

The statistical analysis of the results from the optional assessment of the subbasal nerve plexus by confocal laser scanning microscopy was performed by RC2.

The manuscript of this report was prepared by Gysbert-Botho van Setten, Jutta Horwath-Winter, Daniel Böhringer, Oliver Stachs, and Wolfgang G.K. Müller-Lierheim.

#### **Appendix B. Ethics Committee Approval, Compliance with the Declaration of Helsinki, and Registration of the Study**

The HYLAN M study was approved by ethics committees in all eight countries where study centers were located:


All subjects gave their informed consent for inclusion before they participated in the HYLAN M Study. The study was conducted in accordance with the Declaration of Helsinki.

The HYLAN M study was registered on the database of the European Commission for medical devices EUDAMED under the registration number CIV-16-06-015964. Moreover, the study was registered with national competent authorities in Austria, France, Germany, Italy, Sweden, and Turkey.

#### **Appendix C. Investigational Device**

The control group of the HYLAN M study continued to use the lubricant eye drops which the individual participant had been using by the time of inclusion. In the verum group (Comfort Shield group), the lubricant eye drops which the individual participant had been using by the time of inclusion were replaced by Comfort Shield ® eye drops (i.com medical GmbH, Munich, Germany). Comfort Shield ® eye drops were available to the participants in two dosage forms with identical preservative-free composition, as Comfort Shield ® SD in boxes containing 15 monodoses each, and as Comfort Shield ® MDS in 10 mL bottles. Comfort Shield ® contains 0.15% high molecular weight hyaluronan (Hylan A; intrinsic viscosity 2.9 m<sup>3</sup>/kg) dissolved in isotonic saline solution with 1.20 mmol/L phosphate bu ffer.

Comfort Shield ® eye drops are approved in Europe as Class IIb medical device and were used within the HYLAN M study in accordance with the labeling.

#### **Appendix D. Corneal Fluorescein Staining and Electronic Analysis Method**

Sterile, preservative-free 0.5% sodium fluorescein solution in 0.4 mL monodose containers (FLUORESCEINE FAURE 0.5 PER CENT ophthalmic solution in unit dose, SERB SAS, Paris, France) and Eppendorf Reference 2 pipettes with 10 μL fixed volume (Eppendorf AG, Hamburg, Germany) were provided by the sponsor to each study center.

Then, 10 μL fluorescein solution was released in the cul-de-sac of the patient's eye without touching the ocular surface. The patient blinked five times to evenly distribute the fluorescein on the ocular surface. After 20 to 120 s, the corneal and conjunctival fluorescein staining was judged visually at 16× magnification using the slitlamp with "cobalt" blue light illumination and a yellow barrier filter in the observation beam and graded using the Oxford score [41]. Digital images are taken. The images are electronically transferred to the reading center RC1 with the file names containing the patient number, visit, and eye (OD = right eye or OS = left eye).

During the qualification of the study centers, each study center submitted images of corneas with fluorescein staining together with their visual Oxford grading of staining to ensure and verify that RC1 will correctly assess the staining grade. Then, the settings of slitlamp, illumination, filters, and camera were fixed and kept constant throughout the HYLAN M study.

RC1 analyzed the staining in two steps. Figure A1a shows a typical image that is uploaded to the reading center. First, the region of interest comprising the cornea to be analyzed is segmented manually. In the second step, the CSF-positive lesions are electronically segmented using a threshold-based image processing algorithm. Figure A1b shows the image after electronic processing. The final readout is the percentage of total green pixels divided by the total pixels of the cornea area. The parameters for the automated segmentation were carefully set up for each study center individually as part of the center certification procedure.

**Figure A1.** (**a**) CFS image as uploaded by the trial sites to the reading center; (**b**) CFS image after manually segmenting the cornea and application of a threshold-based segmentation of the CFS-positive lesions. The final readout is the percentage of total green pixels divided by the total pixels of the cornea area.

RC1 had empirically found that 2.00% staining correlates well with Oxford grade 3, and therefore, this was used as the inclusion criteria to be met at least by one eye of the patient. Patients with confluent staining were excluded because the staining cannot be electronically quantified.

The Oxford grading is based on 0.5 log units difference between grade 1 and grade 5 [41]. Therefore, for the transformation percentage of fluorescein staining into "continuous Oxford grades", the following formula was used:

If staining < 0.2% → Grade = 0 If staining [0.2–20%] → Grade = 2.398 + 2 × log10(% staining) If staining > 20% → Grade = 5.

#### **Appendix E. Medical History**

**Table A1.** Tabular overview of the medical history according to the treatment arm—PPS population (*n* = 84).



**Table A1.** Tabular overview of the medical history according to the treatment arm—PPS population (*n* = 84).

Six patients of the Comfort Shield group and two patients of the control group had received autologous serum eye drops by the time of inclusion into the study. In the Comfort Shield group, the autologous serum therapy was substituted by Comfort Shield eye drops.

Fifteen (34.1%) of 44 patients in the Comfort Shield group and 15 (37.5%) of 40 patients in the control group received cyclosporine eye drops and continued their application throughout the study.

Twenty-five (56.8%) of 44 patients in the Comfort Shield group and 23 (57.5%) of 40 patients in the control group were using hyaluronan containing artificial tears by the time of inclusion into the HYLAN M study.

#### **Appendix F. Tear Osmolarity Test Results**

**Figure A2.** Mean (±SD) of osmolarity by treatment arm according time—PPS (*n* = 84). Open circles = Comfort Shield group, filled squares = control group.


**Table A2.** Osmolarity: value in mOsm/L at baseline and at week 8—PPS (*n* = 84)—Descriptive analysis, by treatment arm.

\* values below detection limit (<275 mOsm/L) were not included in the statistical analysis.

**Table A3.** Osmolarity: Value in mOsm/L at baseline and at week 8—PPS (*n* = 84)–Descriptive analysis, by treatment arm.


\* values below detection limit (<275 mOsm/L) were not included in the statistical analysis.

**Figure A3.** Mean (±SD) change from baseline to week 8 (week 8–baseline) of osmolarity by group—PPS (*n* = 84). Open circle = Comfort Shield group, filled square = control group.


**Table A4.** Osmolarity: value in mOsm/L at baseline and change from baseline to week 8–PPS (*n* = 84)–Descriptive analysis, by treatment arm.

\* No missing value but at least one value below detection limit (<275 mOsm/L). \*\* At least one missing value at baseline or Week 8.

#### **Appendix G. Influence of Climate on CFS and OSDI**

The primary endpoint CFS and the key secondary endpoint were separately assessed for two subgroups: "Europe" (patients from all European study centers, including Istanbul) and "Desert" (patients from the two study centers in Riyadh, Saudi Arabia). The results are summarized in the following tables.

**Table A5.** CFS: value at baseline and at each post-baseline visit—PPS (*n* = 84)—descriptive analysis by subgroup.



**Table A6.** CFS: value at baseline and change from baseline to each post-baseline visit—PPS (*n* = 84)—descriptive analysis by subgroup.

**Table A7.** OSDI: value at baseline and at each post-baseline visit—PPS (*n* = 84)—descriptive analysis by subgroup.



**Table A8.** OSDI: Value at baseline and change from baseline to each post-baseline visit—PPS (*n* = 84)—descriptive analysis by subgroup.

#### **Appendix H. Fluctuation of Corneal Fluorescein Staining in the Control Group**

The HYLAN M study included patients with severe dry eye, whose optimum therapy had not changed within the two months (in the case of concomitant cyclosporine therapy, three months) prior to inclusion. Therefore, it was assumed that the corneal fluorescein staining in the control group, where the therapy remained unchanged, will remain fairly constant over the eight-week study period. To prove this assumption, we post hoc analyzed the changes in corneal fluorescein staining of the study eyes of the control group from the baseline to the week 4 visit and from the week 4 to the week 8 visit. We found that from the baseline visit to the week 4 visit, the CFS Oxford grade improved on average by 0.69 with a standard deviation of 1.05, and from the week 4 visit to the week 8 visit, it worsened on the average by 0.23 with a standard deviation of 1.12. An improvement from the baseline to the week 4 visit might be attributable to the fact that the patients participated in a study with controlled dropping frequency and, therefore, adhered more strictly to the prescribed therapy. However, this argumen<sup>t</sup> should not apply to the differences observed between the week 8 visit and the week 4 visit.

We interpret our findings in the sense that CFS as an endpoint is subject to significant fluctuation in patients suffering from severe dry eye disease.
