**3. Results**

## *3.1. Study Participants*

Twenty-three patients were recruited between April 2018 and March 2019, of whom, 18 (78%) were female. The mean age at study entry was 54.6 ± 10.8 years (range 35–74 years).

Ten participants were randomized to the propranolol group, and 13 to the placebo group. Three participants from the placebo group withdrew from the study, two because of low compliance and one after the development of acute otitis media. None of the participants in the propranolol group withdrew.

Twenty participants completed the double-blind phase, 10 participants in each study group. Their baseline characteristics are presented in Table 2. Di fferences in the baseline ESS, and in demographic and clinical parameters between the groups were not statistically significant.


**Table 2.** Baseline characteristics.

\* Total number of units of IV iron (Ferinject 500 mg)/IV packed cells (PC) in the 2 months prior to screening. ESS, epistaxis severity score; ACVRL1, activin receptor-like 1; ND, not done; QOL, quality of life; IV, intravenous; PC, packed cells; BP, blood pressure; HR, heart rate.

Compliance was high for both groups (propranolol 90.2 ± 10.3% vs. placebo 94.2 ± 8.1%, *p* = 0.371).

#### *3.2. Primary Outcome*

During the first phase of the study, the ESS showed a significant improvement in the propranolol group (from a mean of 6.50 ± 1.84 to 4.47 ± 1.75, *p* = 0.004) and no change in the placebo group (5.68 ± 1.8 to 5.33 ± 2.1, *p* = 0.133, Table 3). The change in ESS was −2.03 ± 1.7 in the propranolol group as compared with −0.35 ± 0.68 in the placebo group, *p* = 0.009 (Figure 2, Table 4).

**Table 3.** Outcome measures during the double-blind phase of the study. (Placebo group and propranolol group are analyzed separately).


ESS, epistaxis severity score; DB, double-blind; Hb, hemoglobin; PC, packed cells; IV, intravenous; QOL, quality of life. \* In the 2 prior months and \*\* dose, Ferinject 500 mg.

**Figure 2.** Demonstrating a statistically significant decrease in epistaxis severity score in the propranolol group in the randomized, double-blind phase of the trial.



ESS, epistaxis severity score; Hb, hemoglobin; PC, packed cells; IV, intravenous; QOL, quality of life. \* Dose, Ferinject 500 mg.

Box and whisker plots of the decreases in ESS in the placebo and propranolol groups during the double-blind phase of the study. The black horizontal lines represent the median values of decrease and the range. The dots represent the mean and the boxes represent the interquartile range of decrease (25% to 75%). ESS change of 0.71 was suggested to be the MID (minimal important difference) [16].

#### *3.3. Secondary Outcomes*

Significant improvement in Hb level was observed in the propranolol group but not in the placebo group (Table 3). PC transfusion requirement decreased significantly in the propranolol group. Changes were not observed in these parameters in the placebo group. The change in intravenous iron units required was not statistically significant in both groups. QOL improved significantly in both groups.

We also compared the change in the secondary outcomes during the study period between the placebo and the propranolol groups (Table 4). The propranolol group had a significant decrease in PC requirement as compared with the placebo group. There was no significant change in the other secondary outcome measures.

#### *3.4. Nasal Endoscopy*

Four patients in the placebo group improved (40%), of whom three were grade I at study entry. Improvement was observed in seven patients (70%) in the propranolol group, of whom five were grade II/III at study entry. One patient in the propranolol group worsened. The difference in improvement between the groups at the end of the placebo-controlled phase was not significant (*p* = 0.18). Among patients with higher grading (II and III) of nasal involvement, improvement was observed in 71.4% of the propranolol group as compared with 16.7% of the placebo group (*p* = 0.078).

#### *3.5. Side E*ff*ects*

Hypotension or bradycardia were not observed among participants in the placebo and propranolol groups. Mean BP measures were similar between the groups (systolic BP 108.5 ± 8.1 mmHg vs. 111.3 ± 5.95 mmHg, *p* = 0.402 and diastolic BP 60.2 ± 7.6 mmHg vs. 67.5 ± 12.6 mmHg, *p* = 0.137), as was the mean heart rate (73.6 ± 9.7 beats per minute vs. 74.4 ± 13.45, *p* = 0.876).

The most important side effect observed was a burning sensation in the nasal mucosa and the pharynx (Table 5). Five patients in the propranolol group and two in the placebo group had mild and transient burning sensations that resolved during treatment. Four patients in the propranolol and none in the placebo group complained of a substantial burning sensation. Overall, 9 patients in the propranolol group and 2 in the placebo group had some burning sensation (*p* = 0.005). All the patients had a thorough otolaryngologic examination and no visible mucosal injuries were observed. Rhinorrhea was observed in three patients in the propranolol group and in none in the placebo group (*p* = 0.06). Nasal dryness was observed in one participant in the placebo group.


**Table 5.** Side effects of the double-blind period.

One patient in the propranolol group had acute otitis media. One patient in the placebo group withdrew the study because of otitis media which led to her hospitalization. The patients both recovered. No other side effects were observed.

#### *3.6. Results of the Open-Label Phase of the Study*

Fifteen participants completed the open-label phase, seven in the propranolol group and eight in the placebo group. Two patients in the placebo group withdrew from the study at this phase, as they were awaiting a Young's procedure prior to the study and decided to proceed with this process. Three patients in the propranolol group withdrew from the open-label label phase because of a burning sensation; two of them also had rhinorrhea.

During the open-label phase, the former placebo group, now receiving propranolol gel, showed significant improvement in the ESS score, with a mean drop of 1.99 ± 1.41 from 4.86 ± 1.85 to 2.87 ± 1.6 (*p* = 0.005). During this phase, the propranolol group maintained the decrease in ESS (open-label period 4.6 ± 2.05 to 4.18 ± 0.99, *p* = 0.51) and completed a mean drop of 2.76 ± 1.52 in the 16 weeks of treatment (*p* = 0.001). In addition, the entire cohort showed significant improvement in epistaxis control, frequency and severity, and in daily duration of epistaxis (Table 6).


**Table 6.** Diary outcome measures in the open label phase, diary comparison.

There was no significant improvement in hemoglobin levels (11.42 ± 2.06 to 11.86 ± 1.79, *p* = 0.317) and QOL (41.38 ± 7.38 to 42.92 ± 8.12, *p* = 0.17).

Compliance for the entire cohort during the open-label phase was a mean 78.74±25.2%. Six patients (40%) reported a burning sensation. One participant reported nasal dryness. There was no sign of mucosal damage in any of the participants on rhinology examination, and all chose to continue treatment. No significant change in BP or HR, nor any other side e ffects, were observed between the beginning and the end of the open label period (systolic BP 110.4 ± 5.9 to 114.6 ± 13.8 mmHg, *p* = 0.27; diastolic BP 63.5 ± 8.4 mmHg to 64.4 ± 9.1 mmHg, *p* = 0.54; HR 72.6 ± 11.8 to 71.6 ± 13 per minute, *p* = 0.55, all respectively).

## **4. Discussion**

In a randomized double-blind placebo-controlled study, we demonstrated a significant improvement in epistaxis control in persons with HHT treated with topical propranolol gel for moderate-severe epistaxis. A considerable proportion of persons with HHT have mutations in the endoglin (ENG) and ACVLR1 genes, which encode components of the transforming growth factor (TGF)-β receptor [1]. Recently, a Knudsonian two-hit mechanism was suggested for telangiectases formation [17]. Nonetheless, increased blood and tissue levels of VEGF have been found in persons with HHT [18], suggesting a role for this protein in the abnormal angiogenesis process [6]. The non-selective beta blocker propranolol is used routinely to treat infantile hemangiomas [10,11,19]. The possible mechanisms of action for propranolol include both vasoconstrictive and antiangiogenic effects. The latter are related to VEGF expression, which is controlled by adrenergic stimulation. Propranolol's antiadrenergic properties directly reduce VEGF-stimulated angiogenesis [11]. In addition, propranolol causes apoptosis of endothelial cells and reduces VEGF tissue expression [11]. Infantile hemangioma resembles HHT in several aspects, including dysregulated angiogenesis and high levels of tissue VEGF [20,21]. These similarities sugges<sup>t</sup> a role for propranolol in the treatment of HHT. Furthermore, propranolol has been demonstrated to decrease in vitro migration and angiogenesis of HHT endothelial cells [22].

Oral and topical propranolol for epistaxis in HHT has been reported in several uncontrolled studies. In a retrospective study, followed by a prospective case series, 21 persons with HHT were treated with oral propranolol, with daily doses of 80–160 mg. In the retrospective arm, a significant decrease in ESS was observed, with improvement in both epistaxis duration and frequency. In the prospective arm, 11 patients were treated with propranolol for three months, which led to a significant decrease in epistaxis duration and in the number of episodes per month. Side e ffects caused one patient to discontinue propranolol due to hypotension. Other side e ffects included asthenia, nightmares, and erectile dysfunction [23]. Our group published a case series of six HHT patients treated with nasal propranolol gel. All the patients reported improvement, with a mean decrease of ESS from 6.43 ± 2.11 to 3.47 ± 1.75 during 12 weeks of treatment (*p* = 0.007). Mean hemoglobin levels increased from 8.42 ± 3.06 g/dL to 10.98 ± 1.78 g/dL (*p* = 0.01), and the number of blood transfusions over a 24-week period decreased (*p* = 0.01). One patient reported a burning sensation during the first week of treatment, which resolved during the second week. No other local or systemic e ffects were observed [12]. A recent observational report demonstrated the e fficacy of combined treatment of sclerotherapy and propranolol on epistaxis control in persons with HHT [24]. Anecdotal use of topical timolol has also been reported [25,26].

The targeting of VEGF for treatment of epistaxis in HHT has been attempted with the antibody bevacizumab [27–29]. Systemic administration of bevacizumab was found to control nosebleeds and other HHT symptoms [27,28,30,31]. However, systemic bevacizumab caused undesired side effects such as fever, headache, rash, and chills, and notably, epistaxis occurred [32]. Local nasal mucosal injections or spray of bevacizumab have been shown to have a beneficial effect in several case studies and uncontrolled studies, but failed to show an advantage in randomized trials [29]. Recent systematic reviews summarizing the published studies sugges<sup>t</sup> that intranasal bevacizumab treatment does not have a significant effect on epistaxis in persons with HHT [7,8]. Other therapeutic options based on the role of PI3K and mTOR role in telangiectases formation have been recently suggested [33,34]. Severe epistaxis may cause gushing bleeds several times a day, resulting in iron deficiency anemia and transfusion dependency [1]. Response is sometimes inadequate to standard measures of nasal packing, laser coagulation, and arterial embolization. Surgical procedures including septal dermoplasty and Young's nasal closure may be effective; however, these are partly irreversible and can result in significant morbidity [27].

This randomized placebo-controlled study corroborates previous non-randomized uncontrolled findings of a significant improvement in the ESS score in persons with HHT treated with propranolol. Propranolol was superior to the placebo for improving epistaxis, measured by a significant drop in the ESS. The improvement in epistaxis control with propranolol gel treatment was also demonstrated by the significant increase in HB levels, and decreases in PC transfusions and IV iron requirements in the treatment group but not in the placebo group. These results are strengthened by the significant improvement in epistaxis control in the participants in the placebo group when they were switched to propranolol treatment in the open-label phase. It should be noted that the propranolol group maintained the decrease in ESS during the open label phase, and overall, demonstrated a significant improvement during the 16 weeks of treatment.

Although a one-month run-in period was an option in the protocol, since patients recruited had moderate to severe epistaxis, they were reluctant to have a run-in period. As there was no run-in period, a diary was not available prior to medication administration to measure changes in epistaxis severity, duration, and frequency in the double-blind phase; however, these outcome measures improved in both groups in the open-label phase.

Nasal endoscopy findings in our study groups were defined during follow up visits as no change, improved, or worsened. The rationale for these definitions was based on our experience that improvement was mostly associated with flattening and paling of the lesions and much less often with a change in grading. Improvement in nasal endoscopy findings was observed in 70% of the propranolol and 40% of the placebo group. The improvement in the placebo arm may have been associated with the rigorous regimen of daily nasal lubrication. Daily intranasal application of propranolol gel was associated with improvement in an additional 30% of patients as compared with the placebo group. Although these results were not statistically significant, a larger cohort may show endoscopic improvement in patients treated with topical propranolol. QOL improved in both groups. The improvement in the placebo group could be related to several factors. First, a major component of the SF-12 questionnaire relates to the emotional state of the patient and its influence on their general health. The close follow up and care, the hope for improvement when joining the open-label phase, as well as the moistening effect of the gel, might have contributed to improvement in the patients' well-being and in QOL. A questionnaire that is more specific to the effects of epistaxis might have shown a statistically significant difference between the groups.

We did not observe any systemic side effects with propranolol treatment. A major side effect was a burning sensation, which affected most of the participants, and led to the withdrawal of three participants from the open-label study. However, the majority of participants reported disappearance or reduction in this sensation with continued treatment. This side effect was less pronounced in the open-label phase. While 40% of the patients reported a burning sensation, all chose to continue treatment, and reported improvement in this side effect. Furthermore, no visible damage was observed

in any of the participants in otolaryngologic examination. We believe that propranolol is a mucosal irritant. Since this treatment might be given for longer periods of time, this side e ffect, as well as other systemic side e ffects and dose adjustment, should be addressed in future studies.

One patient in the propranolol group and one who withdrew the study from the placebo group developed otitis media. Whether or not methylcellulose gel is the cause is unknown.

Our study is limited due to the relatively small number of patients. Thus, we were unable to reach conclusions regarding some secondary outcomes, as the need for iron infusion. Moreover, important parameters that may a ffect bleeding such as clinical variants of HHT, type of mutations, as was previously shown by Lesca et al. [35], family relations of the participants, age, and gender were not evaluated. Notably, we had female predominance and, although recent publication did not find a di fference in epistaxis frequency between genders [36], a match for gender should be considered in a larger study.

Our novel findings justify and urge a further study with a larger number of patients that could refer to these matters. Further studies should also include a run-in period with an epistaxis diary. The lack of this period in our study precluded assessing the improvement based on these diaries in the placebo-controlled phase.

In summary, our novel findings sugges<sup>t</sup> that topical propranolol gel is safe and e ffective for epistaxis treatment in patients with HHT-related epistaxis. Larger studies are required to assess the e fficacy and safety of systemic and topical propranolol treatment for epistaxis and other HHT related phenomena.

**Author Contributions:** Conceptualization, M.M.-Z., N.G., E.B. (Elchanan Bruckheimer), and D.P.; methodology, M.M.-Z., N.G., E.B. (Elchanan Bruckheimer), Y.G., E.S., and D.P.; software, Y.G., M.W.; validation, M.M.-Z., N.G., E.B. (Elchanan Bruckheimer), M.W., and E.S.; formal analysis, M.M.-Z., Y.G., N.G., E.S., E.B. (Elchanan Bruckheimer), and E.B. (Einat Birk); writing—original draft preparation, M.M.-Z. and N.G.; writing—review and editing, E.S., E.B. (Elchanan Bruckheimer), D.P., E.B. (Einat Birk), and Y.G.; funding acquisition, M.M.-Z. and N.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors thank the Levi Family for their donation to this study in the memory of their son Oriki.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
