*2.2. Participants*

Patients over 18 years su ffering from DED of any underlying etiology were eligible for inclusion. The patients had to be under stable, unchanged dry eye therapy for at least two months (in case of concomitant cyclosporine therapy three month) by the time of inclusion. Patients were excluded if they participated in any other clinical trial, su ffered from eye diseases other than dry eyes, had ocular surgery less than three months prior to study inclusion, were using punctual plugs, or had masquerading conditions as identified by Karpecki [38]. Masquerading conditions are conjunctivochalasis, recurrent corneal erosions, epithelial basement membrane dystrophy, mucin fishing syndrome, floppy eyelid syndrome, giant papillary conjunctivitis, Salzmann's nodular degeneration, and ocular rosacea.

As inclusion criteria for severe dry eye, the primary criteria according to Baudouin et al. were chosen [39]. The dry eye symptoms were assessed using the ocular surface disease index (OSDI) questionnaire [40]. For inclusion, patients had to have an OSDI score of 33 or more [39]. As dry eye sign, corneal fluorescein staining (CFS) was chosen [41]. For inclusion, patients had to have at least one eye with CFS Oxford grade 3 or more, but no confluent CFS.

Based on the OSDI score and visually assessed CFS grade, the study centers decided on preliminary enrollment. CFS images were transferred to the reading center for quantitative electronic evaluation of corneal fluorescein staining (RC1), as described in Appendix D. The reading center was masked for the assigned treatment to minimize bias. If the submitted images met the criteria for automated assessment, the staining was not confluent, and at least one eye of the patient met the CFS inclusion criteria, the reading center confirmed the definite enrollment of the patient. Otherwise the patient was excluded as "screen fail".

The eyes with the higher staining score were defined as study eyes. However, the fellow eye was retrospectively redefined as a study eye if the masked reading center determined that the images from the fellow eye had significantly better contrast than those of the designated eye, given that the follow eye fulfilled all inclusion criteria at the baseline visit.

#### *2.3. <sup>E</sup>*ffi*cacy Assessment*

TFOS recommends to formally assess dry eye disease by symptoms using a questionnaire such as the OSDI in combination with at least one test for homeostasis markers (signs) such as tear break-up time (TBUT), tear osmolarity, or ocular surface staining [42]. Accordingly, the OSDI questionnaire was used to assess dry eye symptoms throughout the HYLAN M Study. Due to the variable and controversial association between dry eye signs and symptoms, the following standardized battery of tests for dry eye signs was applied in the HYLAN M study. Best corrected visual acuity (BCVA) was used as an indicator of visual stability. CFS, TBUT, and the Schirmer test without topical anesthesia (Schirmer I) were used to assess the lubricating properties, stability, and quantity of the tear film [43–47]. Tear film osmolarity in the lower tear meniscus was additionally measured using the TearLab osmolarity system (TearLab Corporation, San Diego, CA, USA). Specific training of the correct test performance and daily calibration of the test instrument was performed at every study center. Intraocular pressure (IOP) was determined by Goldman applanation tonometry or Icare® tonometry at the baseline and week 8 visits as a safety parameter to rule out uncontrolled glaucoma or ocular hypertension.

Additionally, most but not all study centers additionally performed lissamine green staining of the lid rim to assess lid wiper epitheliopathy (LWE) as the Korb score and the mucocutaneous junction (Marx line) as the Yamaguchi score [48–50]. Moreover, four out of 11 study centers assessed the subbasal nerve plexus by confocal laser scanning microscopy and provided images to a second masked reading center (RC2) for evaluation [51].

Table 1 provides the testing schedule of the HYLAN M study. For more details on diagnostic test methods and assessment of results by reading centers, see Appendices A and D.

#### *2.4. Statistical Analysis*

The primary endpoint for the demonstration of superiority of Comfort Shield eye drops (verum group) over other ocular lubricants currently used by the patients (control group) was the difference between CFS at week 8 and at baseline visits as quantitatively determined by RC1 (see Appendix D).

The key secondary endpoint was the difference between OSDI scores at week 8 and baseline [40]. To further analyze the improvement of symptoms, OSDI subscores for the questions related to pain OSDIpain (OSDI questions 1–3) and OSDI subscore for questions related to visual stability OSDIvision (OSDI questions 4–9) were analyzed according to the following formulas:

$$\text{OSDI}\_{\text{pain}} = \frac{\text{sensitivity to light} + \text{fealing gravity} + \text{pain score eye}}{\text{n}} \times 25 \tag{1}$$

$$\text{OSDI}\_{\text{visual}} = \frac{\text{burned vision} + \text{pcorr vision} + \text{raadding} + \text{driving at night} + \text{complete ATM} + \text{watch TV}}{\text{n}} \times 25 \tag{2}$$

*n* = number of questions answered (at most, 3 for the pain, and 6 for vision subscores, respectively)

Additional secondary endpoints were the differences between BCVA, TBUT, Schirmer I value, tear osmolarity, Korb score, and Yamaguchi score at week 8 and baseline, respectively.

The full analysis set (FAS) was defined as all patients who were not in "screen fail" status, have at least once used their eye drops, have data for the primary endpoint (CFS from reading center RC1) or the key secondary endpoint OSDI, and have had at least one follow-up visit. The per-protocol set (PPS) comprised all patients of the FAS without any major protocol deviation that could substantially affect the evaluation of the randomized treatment. One patient who had not taken lubricant eye drops before the time of inclusion and patients who did not show up for the week 8 follow-up visit were excluded from the PPS.

Considering the potential influence of climatic differences, two subgroups of the verum and control groups were defined: Desert = all patients from the two study centers in Riyadh, Saudi Arabia, and Europe = all other patients. For these subgroups, the results of CFS and OSDI were separately analyzed [52].
