**3. Results**

#### *3.1. Participant Flow*

Figure 1 summarizes the participant flow. In total, 148 patients were pre-screened for severe dry eye disease. Out of these, eight patients were excluded at the initial interview and not randomized, and 121 patients have at least once taken their eye drops (= safety set SS). In total, 140 patients were preliminarily included by the study centers and randomized (= randomized set RS), out of these 75 in the Comfort Shield group and 65 in the control group. Out of the 140 randomized patients, 47 were classified as "screen fails" by the reading center or did not use their eye drops. Of the remaining 93 patients, 49 had been randomized to the Comfort Shield group and 44 had been randomized to the control group (=full analysis set FAS). Out of these, five patients of the Comfort Shield group and four patients of the control group did not show up for the week 8 visit. Therefore, the per-protocol set (PPS) comprises 44 patients in the Comfort Shield group and 40 patients in the control group (see Figure 1).

**Figure 1.** Participant flowchart.

#### *3.2. Demographic Data*

The full analysis set (FAS) of the HYLAN M study comprises 93 patients. Out of these, 84 belonged to the per-protocol set (PPS), 44 belonged to the Comfort Shield group, and 40 were in the control group. The study was performed in di fferent climate zones, and patients from di fferent ethnicities were enrolled. Table 2 provides an overview of the socio-demographic data of the study.



Abbreviations: sd = standard deviation; iqr = interquartile range; min = minimum; max = maximum [53].

An overview of the medical history for the PPS set is provided in Appendix E, Table A1.

#### *3.3. <sup>E</sup>*ffi*cacy Results*

The results presented below refer to the 84 per-protocol patients in the study (PPS population in Figure 1).

#### 3.3.1. Corneal Fluorescein Staining

The di fference in corneal fluorescein staining (CFS) between baseline and week 8 determined by the masked reading center RC1 was the primary endpoint of the HYLAN M Study. The test method, electronic assessment, and calculation are described in Appendix D. Figure 2 and Table 3 describe the test results for the Comfort Shield group and the control group.

The changes from baseline to week 4 and to week 8 are documented in Table 4.

**Figure 2.** Mean (±SD) of corneal fluorescein staining (CFS) (central reading value, transformed into grade exact value) by group according time—PPS (*n* = 84). Open circles = Comfort Shield group, filled squares = control group.

**Table 3.** CFS: Value at baseline and at each post-baseline visit—PPS (*n* = 84)—Descriptive analysis, by group.


**Table 4.** CFS: Value at baseline and change from baseline to each post-baseline visit—PPS (*n* = 84)—Descriptive analysis, by group.


There was no significant (*p*-value < 0.05) difference between the two groups for the primary endpoint CFS, as documented in Table 5.

**Table 5.** CFS: multivariate analysis on change from baseline to week 8—mixed-effects model for repeated measures—PPS population (*n* = 84). *n* used = 79.


(1) Estimate (E) and associated 95% two-sided confidence interval (CI) of the difference between treatment group adjusted means: mixed-effects model for repeated measures (MMRM) with the fixed, categorical effects of treatment, visit, and treatment-by-visit interaction, the random categorical effect of center, as well as the continuous, fixed covariates of baseline and baseline-by-visit interaction. A positive estimate of the difference between treatment group adjusted means is in favour of the Comfort Shield, a negative ones in disfavor of the Comfort Shield. (2) two-sided *p*-value associated with the test of treatment effect.

#### 3.3.2. Ocular Surface Disease Index

The key secondary endpoint of the HYLAN M study was the difference in ocular surface disease index (OSDI) between baseline and week 8 assessed by a questionnaire to be filled by the patients at the beginning of each visit. The Comfort Shield group had experienced at the end of the study (week 8 visit) significantly more relief from dry eye symptoms than the control group as documented in Figure 3a and Tables 6–8 (*p*-value 0.001).

**Figure 3.** (**a**) Mean (±SD) OSDI total score; (**b**) mean (±SD) OSDI pain subscore; (**c**) mean (±SD) OSDI vision subscore; (**d**) mean (±SD) best corrected visual acuity (BCVA). Open circles = Comfort Shield group, filled squares = control group.


**Table 6.** Ocular surface disease index (OSDI): value at baseline and at each post-baseline visit—PPS (*n* = 84)—descriptive analysis, by group.

**Table 7.** OSDI: value at baseline and change from baseline to each post-baseline visit—PPS (*n* = 84)—Descriptive analysis, by group.


**Table 8.** OSDI: Multivariate analysis on change from baseline to week 8—mixed-effects model for repeated measures—PPS population (*n* = 84)—*n* used = 84.


(1) Estimate (E) and associated 95% two-sided confidence interval (CI) of the difference between treatment group adjusted means: MMRM with the fixed, categorical effects of treatment, visit, and treatment-by-visit interaction, the random categorical effect of center, as well as the continuous, fixed covariates of baseline and baseline-by-visit interaction. A positive estimate of the difference between treatment group adjusted means is in favor of the Comfort Shield, a negative one in disfavor of the Comfort Shield. (2) two-sided *p*-value associated with the test of treatment effect.

The subscores for pain and visual stability-related symptoms were calculated and analyzed as described in the section on statistical analysis. The results are provided in Figure 3b,c and in Tables 9 and 10. Both subscores OSDIpain and OSDIvision improved significantly in the Comfort Shield group as compared to the control group (*p*-values 0.002 and 0.003, respectively).

**Table 9.** Pain OSDI subscore: multivariate analysis on change from baseline to week 8—mixed-effects model for repeated measures—PPS population (*n* = 84)—*n* used = 84.


(1) Estimate (E) and associated 95% two-sided confidence interval (CI) of the difference between treatment group adjusted means: MMRM with the fixed, categorical effects of treatment, visit, and treatment-by-visit interaction, the random categorical effect of center, as well as the continuous, fixed covariates of baseline and baseline-by-visit interaction. A positive estimate of the difference between treatment group adjusted means is in favour of the Comfort Shield, a negative one in disfavour of the Comfort Shield. (2) two-sided *p*-value associated with the test of treatment effect.

**Table 10.** Vision OSDI subscore: Multivariate analysis on change from baseline to week 8—mixed-effects model for repeated measures—PPS population (*n* = 84)–*n* used = 84.


(1) Estimate (E) and associated 95% two-sided confidence interval (CI) of the difference between treatment group adjusted means: MMRM with the fixed, categorical effects of treatment, visit, and treatment-by-visit interaction, the random categorical effect of center, as well as the continuous, fixed covariates of baseline and baseline-by-visit interaction. A positive estimate of the difference between treatment group adjusted means is in favour of the Comfort Shield, a negative one in disfavor of the Comfort Shield. (2) two-sided *p*-value associated with the test of treatment effect.

#### 3.3.3. Best Corrected Visual Acuity

The BCVA slightly improved after eight weeks of Comfort Shield treatment as compared to the control group (*p*-value 0.033). Details are provided in Figure 3d and Table 11.



#### 3.3.4. Other Secondary Endpoints

The secondary endpoints TBUT, Schirmer I, lid wiper epitheliopathy Korb score, Yamaguchi score, and tear film osmolarity are summarized in Table 12. No significant differences between the Comfort Shield group and control group were observed (all *p*-values > 0.05).

**Table 12.** Values at baseline and change from baseline to week 8 for tear film break-up time (TBUT), Schirmer I, lid wiper epitheliopathy (LWE) Korb score, Yamaguchi score, and tear osmolarity—PPS (*n* = 84).


\* See Appendix F for details of statistical handling of measurement values below the detection limit of the TearLab test instrument.

#### 3.3.5. Observation of the Subbasal Nerve Plexus by Confocal Microscopy

Confocal laser scanning microscopy was performed on 16 patients (eight patients each in the Comfort Shield group and in the control group) at four out of 11 study centers. Images of the subbasal nerve plexus were taken at baseline and week 8 and assessed at RC2. There was a significant increase of total nerve fiber length in the Comfort Shield group (51% growth; *p*-value 0.030), whereas in the control group, the total subbasal corneal nerve fiber length did not significantly change from baseline to week 8. Detailed results will be subject to a separate publication.

#### 3.3.6. Dropping Frequency

The patients were instructed to use their lubricant eye drops whenever ocular discomfort occured. They recorded the dropping daily. The average dropping frequency was not significantly different in the two treatment arms. By the time of inclusion into the study, the patients reported using 7.6 (minimum: 2; maximum: 36) artificial tear drops or autologous serum eye drops per day in the control group, and 8.2 (minimum: 3; maximum: 20) in the Comfort Shield Group. During week 8 of the study, the average dropping frequency was 6.5 (minimum: 1; maximum: 24.6) per day in the control group and 7.1 (minimum: 2; maximum: 23.8) in the Comfort Shield group.

#### 3.3.7. Influence of Climate on CFS and OSDI

In order to investigate whether or not climate has a significant impact on the study results, the primary endpoint CFS and the key secondary endpoint OSDI were analyzed separately for the nine study centers located in Europe and the two study centers in Riyadh in the desert region of Saudi Arabia. There were no significant differences between these two subgroups. The complete results are presented in Appendix G, Tables 6–8 and A5.

#### *3.4. Safety Results*

The assessment of safety results refers to the safety set (SS), i.e., all patient that had at least once received eye drops (*n* = 121).

The average intraocular pressure in both study arms at baseline and week 8 was 14 mmHg. All values were between 8 and 23 mmHg. There were no patients suspect of uncontrolled glaucoma or ocular hypertension.

Of the Comfort Shield group, one patient discontinued the participation in the study after one week because the dry eye symptoms had worsened. Two patients reported during the week 4 visit about blurred vision for 10 min after the instillation of Comfort Shield eye drops, but they wanted to continue to participate in the study. One patient reported during the week 4 visit about persistent redness but wanted to continue to participate in the study. One patient reported during the week 4 visit about burning sensation, but they wanted to continue to participate in the study. One patient reported during the week 4 visit an episode of three days of red, painful, itching left eye, but they wanted to continue to participate in the study.

Of the control group, three patients had experienced not device-related adverse events between the week 4 and week 8 visits. One patient had nausea for two days, one patient had a mild viral conjunctivitis, and one patient had to be admitted to the hospital with cervical pain and was treated for six days with analgesics.

## **4. Discussion**

The population of this study shows the typical predominance of age and female gender in dry eye disease, as the majority of patients was older than 45 years (90.4%, respectively, 65 years (33.3%) and most patients were female (82.1%) [54]. There was no significant di fference between the verum and the control group.

The design of clinical trials on dry eye disease needs to consider symptoms, namely ocular discomfort and visual disturbance, as well as signs, such as tear film instability, damage of the ocular surface, increased tear osmolarity, and inflammation of the ocular surface [55]. International regulatory agencies rely on ocular surface staining as a primary endpoint for new drug approvals [56]. The ODISSEY European Consensus Group recommended CFS as the primary sign for severity of DED [39]. For this reason, the HYLAN M study used CFS as the primary endpoint of the study and standardized the test method as well as the objective assessment of staining as far as reasonably possible. Other dry eye signs, such as tear osmolarity, TBUT, Schirmer I, lid wiper epitheliopathy, and position of the mucocutaneous junction at the lid rim were chosen as secondary endpoints, having in mind the well-known poor correlation between symptoms and signs in DED. As corneal nerve damage has in recent years been recognized as an important pathomechanism in severe ocular surface disease, the assessment of the subbasal nerve plexus using confocal laser scanning microscopy had been included as an additional optional test in the study design. OSDI was chosen as the key secondary endpoint for the assessment of dry eye symptoms as it is widely used and easy to interpret.

The HYLAN M study did not find a statistically significant di fference between the Comfort Shield group and the control group at the week 8 visit for the primary endpoint CFS or any of the following secondary endpoints: TBUT, Schirmer I, lid wiper epitheliopathy, and tear osmolarity. This emphasizes that in patients with severe dry eyes, a change from the therapy with individualized lubricant eye drops to HMWHA eye drops does not result in a worsening of dry eye signs.

The primary endpoint of the HYLAM M study, CFS, did not show any significant di fference between the two study arms. The value of CFS as an absolute number to judge the improvement or deterioration of the corneal surface condition has been questioned. As it is known that the di fference in grading between di fferent investigators may limit the sensitivity of detectable changes in CFS over time, the CFS test method within the HYLAN M study had been highly standardized, and a reading center performing electronic assessment of CFS images has been involved (see Appendix D). The well-known, possibly even physiological variation of CFS between measurements has been supported by the present study. Moreover, it is known that CFS is sensitive to e ffects of quenching and pooling, which may a ffect the repeatability and accuracy of measurement. A post hoc analysis of the control group demonstrated that there is a significant fluctuation in CFS over time even in the best treated patients under stable optimum treatment (see Appendix H). This variation emphasizes the di fficulty of judging the ocular surface condition from surface staining intensity. Similar fluctuations such as the one experienced for CFS are known for other dry eye signs [17,52,57–63]. Originally enthusiastically welcomed as a highly reliable parameter for DED and used as a major decision maker with respect to the severity of DED, the diagnostic significance of tear osmolarity determined in the lower tear meniscus has recently been questioned [64]. The average osmolarity of 298 mOsm/L found in the HYLAN M study for patients su ffering from severe DED is not hyperosmotic, as expected for severe DED.

OSDI was the key secondary endpoint of the HYLAN M study. The OSDI questionnaire is one of the most commonly used tests to assess dry eye symptoms [42]. The OSDI score was assessed at the baseline visit, after four weeks, and after eight weeks. Whereas, in the control group, the OSDI score slightly improved in the first four weeks, which was presumably due to better compliance of the patients with their treatment regimen, but it did not further improve beyond the four-week study participation. Contrarily, the OSDI score significantly improved under Comfort Shield treatment in the first four weeks and continued to improve in the second four-week period (see Figure 3a). Such improvement applies also to the subscore for discomfort and pain, as well as for the subscore for visual instability (see Figure 3b,c). After eight weeks treatment, the di fference between the Comfort Shield group and the control group were for the total OSDI score 13.5 (*p*-value 0.001), for the pain subscore 14.5 (*p*-value 0.002), and for the vision subscore 14.0 (*p*-value 0.003). This unexpectedly grea<sup>t</sup> improvement in dry eye symptoms under the treatment with HMWHA eye drops deserves further investigation. The Asia Dry Eye Society recently concluded that subjective severity (symptoms) could be used as a marker for therapeutic e fficacy in dry eye treatment [9].

The improvement of the symptoms of visual stability in the HYLAN M study was reflected by a minor but significant improvement of BCVA. Whereas, BCVA determines the best visual acuity within a certain period of time, functional visual acuity continuously determines visual acuity and, therefore, better reflects the subjective stability of vision [65–67]. Therefore, in future clinical studies on dry eye disease, functional visual acuity rather than BCVA might be used as an endpoint.

As an optional test within the HYLAN M study, the subbasal nerve plexus was analyzed in a subgroup of 16 patients. There was a significant increase of total nerve fiber length in the Comfort Shield group as compared to the control group after eight weeks of treatment. This observation correlates well with the significant improvement of pain symptoms. The fact that at the same time there was no significant change of other dry eye signs suggests that the observed therapeutic effect cannot be attributed to a physical effect of the eye drops such as hydration or lubrication, but it is likely to result from a pharmacological effect downregulating ocular inflammation and supporting corneal nerve recovery.
