Safety of BI 1358894 in patients with major depressive disorder: Results and learnings from a phase II randomized decentralized clinical trial Authors: Christopher Reist, Peide Li, Thuy Le Nguyen, Sigurd D. Süssmuth Published: 2024 ABSTRACT AbstractThe feasibility of conducting a fully remote, interventional, phase II decentralized clinical trial (DCT) was investigated in major depressive disorder (MDD). Key learnings were collated to improve future DCTs. A double‐blind, placebo‐controlled, parallel‐group, DCT enrolled adult MDD patients with inadequate response to first‐line antidepressant monotherapy (ongoing ≥8 weeks) and a Montgomery‐Åsberg Depression Rating Scale total score (MADRS) ≥22 at screening. Patients were randomized 1:1 to BI 1358894 125 mg or placebo daily for 6 weeks remotely. Safety parameters, primary end point (change from baseline in MADRS at Week 6), and patient experience were assessed. The DCT was considered feasible if the trial protocol could be successfully executed. Overall, DCT procedures were successfully executed per protocol. However, despite achieving a vast patient outreach, the trial was terminated early due to deficient enrollment. Of the 136 patients who consented for enrollment and underwent screening, 45 were randomized and 43 received treatment (BI 1358894, n = 20; placebo, n = 23); 97.7% of patients completed the trial. Patients had a mean (SD) age of 42.2 (13.1) years and most (83.7%) were female. Adverse events were reported by 86.0% of patients (BI 1358894, 90.0%; placebo, 82.6%). Most patients (88%) reported a positive experience with the DCT. Key learnings related to the impact of stringent eligibility criteria, recruitment optimization strategies, plus the benefits and limitations of digital technologies. A fully remote, interventional DCT was feasible in MDD, and was well perceived by trial participants. Learnings related to recruitment optimization and trial design should be considered for future interventional DCTs. Study Highlights WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC? To the best of our knowledge, no prior clinical trial in patients with major depressive disorder (MDD) has utilized a randomized interventional design in a fully remote, decentralized, clinical trial (DCT) setting, alongside robust medical screening. WHAT QUESTION DID THIS STUDY ADDRESS? This study investigated the feasibility of conducting a phase II safety and efficacy trial of BI 1358894 125 mg vs. placebo in patients with MDD using a fully remote randomized interventional DCT approach. WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE? The feasibility of conducting an interventional DCT was demonstrated in and well perceived by patients with MDD. Although efficacy was not evaluated due to the small sample size, this trial demonstrated that daily adjunctive BI 1358894 along with standard‐of‐care antidepressants was generally safe and well‐tolerated over 6 weeks. HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE? DCTs could extend the reach of trial investigators to underserved patient populations in real‐world settings. Further investment in DCTs could help to advance our knowledge and may usher in a new era of patient‐centered clinical trial research. Study Highlights WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC? WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC? WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC? To the best of our knowledge, no prior clinical trial in patients with major depressive disorder (MDD) has utilized a randomized interventional design in a fully remote, decentralized, clinical trial (DCT) setting, alongside robust medical screening. WHAT QUESTION DID THIS STUDY ADDRESS? WHAT QUESTION DID THIS STUDY ADDRESS? WHAT QUESTION DID THIS STUDY ADDRESS? This study investigated the feasibility of conducting a phase II safety and efficacy trial of BI 1358894 125 mg vs. placebo in patients with MDD using a fully remote randomized interventional DCT approach. WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE? WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE? WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE? The feasibility of conducting an interventional DCT was demonstrated in and well perceived by patients with MDD. Although efficacy was not evaluated due to the small sample size, this trial demonstrated that daily adjunctive BI 1358894 along with standard‐of‐care antidepressants was generally safe and well‐tolerated over 6 weeks. HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE? HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE? HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE? DCTs could extend the reach of trial investigators to underserved patient populations in real‐world settings. Further investment in DCTs could help to advance our knowledge and may usher in a new era of patient‐centered clinical trial research. INTRODUCTION Despite various treatment options for major depressive disorder (MDD), ~30% of depressive episodes do not respond sufficiently to any first‐line antidepressant. 1 This suggests there is an unmet need for new and effective treatments for MDD. 1 The dysregulation of the cortico‐limbic system of the brain is putatively implicated in the pathophysiology of MDD and other affective disorders. 2 , 3 , 4 In rodents, transient receptor potential canonical 4 (TRPC4) and 5 (TRPC5) ion channels are expressed in areas of the cortico‐limbic system, including the prefrontal cortex and amygdala, and modulate neuronal excitability. 5 , 6 Preclinical studies in mice have shown neuronal excitability in the amygdala can be reduced by HC‐070, a TRPC4/TRPC5 ion channel inhibitor. 6 Additionally, HC‐070 has demonstrated anxiolytic and antidepressant‐like effects in rodents; HC‐070 reduced anxiety in mice following cholecystokinin tetrapeptide (CCK‐4) administration, a compound that induces feelings of fear and anxiety, and reduced immobility in the tail suspension and forced swim tests, respectively. 6 BI 1358894, a novel TRPC4/5 inhibitor, has shown proof‐of‐principle in a functional magnetic resonance imaging study in humans; BI 1358894 attenuated hyperreactivity in several cortico‐limbic brain regions, including the amygdala bilaterally, when patients with MDD were asked to view negative emotional faces and scenes. 7 Another proof‐of‐principle phase I study demonstrated that BI 1358894 reduced psychological and physiological responses to CCK‐4–induced anxiety, panic symptoms, and biomarkers of stress compared with placebo in healthy males. 8 Moreover, phase I studies have demonstrated that BI 1358894 is generally safe and well‐tolerated at doses up to 200 mg, with higher medication exposure in the fed vs. fasted state. 9 Across these phase I studies, headache, dizziness, and fatigue were the most commonly reported adverse events (AEs). 9 Together, these findings warrant the continued development of BI 1358894 for the treatment of depression. The coronavirus disease 2019 (COVID‐19) pandemic revealed flaws in clinical trials and created an opportunity to re‐envision trial design 10 , 11 using a decentralized clinical trial (DCT) approach to increase accessibility for patients and investigators. A DCT is a patient‐centric clinical trial that utilizes digital innovations such as telemedicine and smartphone applications to enable trial conduct at patients' homes. 12 Although DCTs are developing rapidly, there is a dearth of directly comparable information regarding the associated recruitment, patient retention, and medication adherence. 13 This trial investigated the feasibility of conducting a fully remote, interventional, DCT in a phase II trial in MDD. This DCT also aimed to assess the safety and efficacy of adjunctive BI 1358894 125 mg vs. placebo in patients with MDD and inadequate response to first‐line antidepressants. As using a fully remote DCT model is a novel approach in MDD, the key learnings were also collated to improve future DCTs. METHODS Trial design This phase II randomized, double‐blind, placebo‐controlled, parallel‐group DCT (ClinicalTrials.gov identifier: NCT04423757) was conducted remotely at patients' homes through a collaboration between Boehringer Ingelheim (BI) and Science 37 (CA, United States [US]), as a single (meta)site in the US. This trial included 10 visits, which occurred on Day ‐28 (Visit 1; screening), Day ‐7 (Visit 1A), Day 1 (Visit 2; randomization), followed by weekly visits starting from Week 1 (Visit 3) to Week 6 (Visit 8; end of treatment [EoT]), and after a 28‐day follow‐up period (Visit 9; end of study [EoS]). Visits 1, 2, 4, 6, 8, and 9 included home visits by a mobile nurse, while other visits were conducted remotely via videoconference (Visits 1A and 3) or by telephone (Visits 5 and 7) by trained trial staff (Figure 1). All study visits were scheduled via the Science 37 platform or telephonically. Study design. EoT, end of treatment; EoS, end of study; V, Visit. The trial was conducted in accordance with the principles of the Declaration of Helsinki, 14 the International Council for Harmonization of Good Clinical Practice, 15 applicable regulatory requirements, and BI standard operating procedures. The Institutional Review Board service provider, Advarra, 16 was responsible for reviewing study materials, including the trial protocol and patient consent forms. Recruitment Patients were recruited using a multiplatform approach through social media, online search engines, digital patient communities, online medical information providers, the Science 37 patient database, advocacy groups, and referrals from healthcare providers. Interested patients were directed to the Science 37 trial‐specific website to answer pre‐qualifying questions. Pre‐qualified patients completed a telephonic pre‐screening questionnaire with trial staff and provided their contact information. For the pre‐screening questionnaire, please refer to the Materials S1. Patients meeting pre‐screening eligibility provided authorization via a signed medical release to Science 37 to obtain their medical records from their respective healthcare provider(s). If deemed eligible after medical record review, the potential participant could sign an electronic informed consent form that was registered in the Interactive Response Technology (IRT) system. During enrollment, local urgent care or emergency department contact information was documented in the Science 37 platform to manage any emergent health issues. This was augmented by Local Laboratory Patient Service Centers in case any laboratory evaluations were required outside of trial visits. Patients Patients were 18–65 years of age, with an established MDD diagnosis at screening, a current depressive episode of ≥8 weeks with at least moderate disease severity (Montgomery‐Åsberg Depression Rating Scale [MADRS] total score ≥ 22), and ongoing antidepressant monotherapy with either a selective serotonin reuptake inhibitor (SSRI) or a serotonin–norepinephrine reuptake inhibitor (SNRI) for ≥8 weeks. Patients with a diagnosis of any other mental disorder at the time of screening or a mental disorder that was the primary focus of treatment in the last 6 months or patients with any personality disorder likely to impact trial participation as per investigator's judgment were excluded. Suicidal behavior in the past year or during screening or suicidal ideation of type 4/5 (Columbia‐Suicide Severity Rating Scale [C‐SSRS]) in the 3 months before and during screening also resulted in exclusion. Patients could have no more than three unsuccessful monotherapy treatments with an approved antidepressant for the current ongoing major depressive episode. The use of bupropion was not allowed within 7 days prior to randomization and during the trial. The full eligibility criteria are included in the Materials S1. Randomization and blinding Eligible patients were randomized 1:1 to receive BI 1358894 125 mg or placebo once daily for 6 weeks followed by a 28‐day follow‐up period. Randomization was conducted using IRT; assignment of trial medications via the IRT system could occur up to 7 days (Visit 1A) before randomization (Visit 2) to allow time for the medication kit assignment in the IRT system, as well as shipment and delivery of trial medications to patients' homes. Patient assignment to each treatment was determined by a computer‐generated random sequence and access to the randomization code was controlled and documented. The kits corresponding to the assigned medication numbers were given to the patient. To ensure blinding for patients and trial staff, the packaging was identical for both treatments. Patients were instructed to take the BI 1358894 tablets or matching placebo tablets orally with water and in a consistent way, that is, with or without food every morning at approximately the same time. Medication adherence and compliance were regularly monitored using the pill‐counting method by mobile nurses during home visits. A medication adherence monitoring platform was also used to confirm medication ingestion using facial recognition technology via a smartphone application (AiCure, LLC, USA). Patients' faces were anonymized by AiCure before the data were analyzed. The smartphone application also issued reminders to patients to take their trial medication and provided a communication system to allow real‐time intervention in case of medication interruptions. End points and assessments Safety Safety was assessed based on the incidences of on‐treatment AEs, medication‐related AEs, serious AEs (SAEs), and AEs of special interest (AESIs), which were reported using the Medical Dictionary for Regulatory Activities (MedDRA) version 25.0. The only predefined AESI was hepatic injury (as is standard for pre‐marketing clinical evaluation), which was defined as an elevation of aspartate aminotransferase (AST) and/or alanine aminotransferase (ALT) ≥3‐fold upper limit of normal (ULN) combined with an elevation of total bilirubin ≥2‐fold ULN measured in the same blood draw sample, or aminotransferase (ALT and/or AST) elevations ≥10‐fold ULN. Other safety assessments included physical examination, laboratory tests, vital signs, electrocardiogram (ECG), and evaluation of C‐SSRS scores (suicidality). Safety assessments were performed throughout the trial to ensure safety monitoring in a remote setting. Participants were instructed to report AEs to the investigator, who provided appropriate clinical management for AEs and SAEs. Efficacy The primary efficacy end point was the change from baseline in MADRS total score at Week 6. The MADRS comprises 10 items to assess the core symptoms of depression. MADRS items are rated from 0 (no abnormality) to 6 (severe), with total scores ranging from 0 (normal with absence of symptoms) to 60 (severe depression). 17 The secondary end points included Symptoms of Major Depressive Disorder Scale (SMDDS) total score, 18 Clinical Global Impression Severity Scale (CGI‐S) scores, 19 Patient Global Impression Severity Scale (PGI‐S) score, and Patient Global Impression of Change Scale (PGI‐C) score. 20 Apart from the PGI‐C score, which was assessed once at Visit 8, all other secondary end points were assessed at Visits 2, 3, 4, 6, and 8. Patient experience The overall patient experience of the trial was assessed using a patient feedback questionnaire issued via a QR code at EoS or via an email sent on the day of, or the day after, EoS. All the assessments in this trial were similar to those included in a traditional trial (ClinicalTrials.gov identifier: NCT04521478) but were completed at patients' homes. Feasibility The DCT was considered feasible if it was possible to execute the trial per protocol. Procedures Communications between the investigator, trial staff, and patients were conducted remotely either via the Science 37 platform using a Health‐Insurance‐Portability and Accountability‐Act compliant Zoom videoconference or via a telephone call. At Visit 1A, participants were trained on the AiCure app and the Science 37 platform. Mobile nurses visited patients' homes to complete the designated trial procedures (collecting vitals and laboratory samples, assisting with trial devices, ECGs, telemedicine physical exams, and administering the trial medication) and entered readings into the Science 37 platform. Safety blood samples were collected and processed by a mobile nurse using a portable centrifuge when necessary. Blood samples were then stored and shipped with dry ice to the central laboratory via overnight courier. The sample preparation, storage, and shipping parameters were specified by central laboratory to ensure quality. ECGs were collected at the patients' homes by a mobile nurse via Spaulding Electrocardiograph machine and uploaded to the Science 37 platform, which was used by the Science 37 investigators to clinically assess the tracing in real time. Further details of trial equipment and supplies carried by mobile nurses or shipped to patients' homes are included in the Materials S1. All clinician‐administered assessments including the Structured Clinical Interview for DSM‐5 (SCID‐5) and MADRS were conducted remotely via telemedicine by Science 37 investigators. The assessment procedures were standardized; provisioned smartphones were arranged such that the Science 37 investigator could clearly observe the patient, and patients were advised to use a private, distraction‐free setting. Audio recordings of these diagnostic and key end point assessments were recorded by Science 37 investigators and centrally reviewed by a third‐party vendor to identify errors in administration and scoring. No other confirmatory assessments were conducted. The third‐party reviewer confirmed eligibility based on diagnosis and depression severity. All patient‐reported instruments were completed by the patient on the provisioned smartphones. Statistical analysis and sample size A two‐sided alpha of 0.10 was used for the sample size calculation using a treatment allocation ratio of 1:1 for the placebo and BI 1358894 125 mg treatment groups. Within a projected 12‐month recruitment period, it was expected that the DCT would achieve a sample size of 164 patients. Based on the sample size assumptions, a power of 77% could be achieved if 82 patients per treatment were recruited into the trial. Safety analyses were conducted on the treated set and comprised all randomized patients who had received ≥1 dose of the trial medication. Efficacy analyses were performed on the full analysis set (FAS), which comprised all randomized patients who received ≥1 dose of trial medication during the trial and had 1 baseline and post‐baseline observation recorded. Treatment compliance for a visit was based on non‐missing tablet counts and calculated using a formula described in the Materials S1. Missing data were not imputed and remained missing for all analyses including a summary of laboratory assessments. All data were analyzed descriptively. RESULTS Patient recruitment Enrollment to the trial started in July 2020; the initial recruitment was slow as some patients were hesitant to participate during the COVID‐19 pandemic, possibly to avoid trial staff entering their homes to conduct assessments. Hence, it was decided to expand the number of recruitment states from 13 to 35 by the end of the study (Figure S1). The recruitment delays necessitated implementing optimization strategies, such as an increase in social media platforms used for digital recruitment (Facebook [primary], Twitter, Snapchat, Google ads, Reddit, digital patient communities, and online medical information providers), the addition of more provider networks for referrals of patients with MDD and establishing the Bring Your Own Investigator (BYOI) initiative. The BYOI initiative was a collaborative effort between BI and Science 37 to seek potential physicians or physician networks with populations of patients with MDD (with or without an existing clinical research infrastructure) who could refer patients to the DCT. The physicians received necessary training on how to use the Science 37 platform and function as trial sub‐investigators. As such, the DCT attracted people from all over the US with 80,836 patients signing up for the trial (Figure S2). Although the DCT design created a wide interest in the trial, this did not translate to accelerated recruitment during the 22‐month recruitment period. Of those patients who signed up for the trial, 37,939 were contact‐eligible (and typically contacted by trial staff within 48 hours), 13,816 qualified for pre‐screening, and 136 consented for enrollment and underwent screening (Figure 2). Of the patients who consented, 45 were randomized (BI 1358894 125 mg: n = 21; placebo: n = 24). The main reasons for screening failures were that patients did not meet the MDD diagnosis criteria or the MADRS qualifying score. Because of the deficient enrollment, it was decided to terminate the trial early. Patient disposition. AE, adverse event; MADRS, Montgomery‐Åsberg Depression Rating Scale; MDD, major depressive disorder; SNRI, serotonin‐norepinephrine reuptake inhibitor; SSRI, selective serotonin reuptake inhibitor. Patient disposition and demographics Of the 45 randomized patients, 2 did not receive the assigned trial medication (1 in each group). The patient randomized to BI 1358894 125 mg group withdrew consent, and the one allocated to placebo arm was lost to follow‐up post‐randomization. Of the 43 patients included in the FAS, 42 (97.7%) completed the trial and 41 (95.3%) completed the trial medication. Of those who received treatment, two patients in the BI 1358894 125 mg group discontinued treatment due to AEs, although one of them completed the trial. The disposition of enrolled patients is presented in Figure 2. Age distribution was similar across groups; the mean (SD) age of patients was 42.2 (13.1) years, and the majority (83.7%) were female (Table 1). The mean (SD) MADRS total score at baseline was 26.55 (6.6) and 28.43 (7.1) in the BI 1358894 125 mg and placebo groups, respectively. Treatment adherence according to pill‐counting was 98.5%, while visually confirmed medication adherence using AiCure, based on the highest confidence level (“green”), indicated trial medication intake occurred in 52.0% of patients. TABLE 1Baseline demographics and assessments.Baseline demographic characteristicBI 1358894 125 mg, n = 20Placebo, n = 23Total, N = 43Female, n (%)18 (90.0)18 (78.3)36 (83.7)Age, mean (SD)41.1 (13.4)43.2 (13.0)42.2 (13.1)Race, n (%)Black or African American1 (5.0)2 (8.7)3 (7.0)Asian01 (4.3)1 (2.3)White18 (90.0)20 (87.0)38 (88.4)Multiple1 (5.0)01 (2.3)Hispanic/Latino ethnicity (Yes), n (%)3 (15.0)2 (8.7)5 (11.6)BMI (kg/m2) (n, %)≤30 kg/m2 7 (35.0)9 (39.1)16 (37.2)>30 kg/m2 13 (65.0)14 (60.9)27 (62.8)Baseline assessmentsTime since diagnosis of MDD (years), mean (SD)14.6 (9.2)16.3 (12.6)15.5 (11.1)Substance use a (n, %)Alcohol6 (30.0)8 (34.8)14 (32.6)Caffeine19 (95.0)21 (91.3)40 (93.0)Cannabis5 (25.0)2 (8.7)7 (16.3)Hallucinogens000Inhalants000Opioids000Sedatives, hypnotics, and anxiolytics2 (10.0)2 (8.7)4 (9.3)Stimulants1 (5.0)01 (2.3)Tobacco2 (10.0)7 (30.4)9 (20.9)Psychotherapy use (Yes), n (%)4 (20.0)4 (17.4)8 (18.6)Baseline C‐SSRS suicidal ideation, n (%)3 (15.0)1 (4.3)4 (9.3)Lifetime C‐SSRS suicidal ideation, n (%)19 (95.0)20 (87.0)39 (90.7)Lifetime C‐SSRS suicidal behavior, n (%)4 (20.0)16 (69.6)20 (46.5)Abbreviations: C‐SSRS, Columbia‐Suicide Severity Rating Scale; MDD, major depressive disorder; SD, standard deviation. a Patients may have taken more than one substance so the percentages may not add up to 100%. Baseline demographics and assessments. Abbreviations: C‐SSRS, Columbia‐Suicide Severity Rating Scale; MDD, major depressive disorder; SD, standard deviation. Patients may have taken more than one substance so the percentages may not add up to 100%. Safety AEs were reported by 86.0% of patients. On‐treatment AEs were slightly higher with BI 1358894 125 mg (90.0%) versus placebo (82.6%; Table 2). Most AEs were mild to moderate in intensity. One severe AE of headache was reported in the BI 1358894 125 mg group and resolved at Day 7 after initiation of treatment. More patients treated with BI 1358894 125 mg (55.0%) vs. placebo (30.4%) had investigator‐defined medication‐related AEs. The most common medication‐related AEs were headache, fatigue, and dizziness (Table S1). Two medication‐related AEs of palpitation (moderate) and feeling abnormal (mild) led to the discontinuation of BI 1358894 125 mg in two patients and were resolved after treatment was discontinued. Furthermore, one of these patients continued with the study procedures and completed the study. There were no AESIs and no AEs leading to death. Serious AEs (depression and major depression) were reported in two patients in the placebo group and were assessed as not related to BI 1358894 treatment (Table 2). TABLE 2Adverse event overall summary (Treated set).Number of patients, N (%)BI 1358894 125 mg, n = 20Placebo, n = 23Total, N = 43Patients with any AE, n (%)18 (90.0)19 (82.6)37 (86.0)Patients with severe AEs, n (%)1 (5.0)0 (0)1 (2.3)Patients with investigator‐defined medication‐related AEs, n (%)11 (55.0)7 (30.4)18 (41.9)Patients with AEs leading to discontinuation of trial medication, n (%)2 (10.0)02 (4.7)Patients with AEs of special interest, n (%)000Patients with serious AEs, n (%)02 (8.7)2 (4.7)Results in death000Is life‐threatening000Persistent or significant disability/incapacity000Requires or prolongs hospitalization000Congenital anomaly or birth defect000Other medically important serious event02 (8.7)2 (4.7) Note: A patient may have serious AE(s) with multiple seriousness criteria. Percentages are calculated using total number of patients per treatment as the denominator. MedDRA version used for reporting: 25.0.Abbreviation: AE, adverse event. Adverse event overall summary (Treated set). Note: A patient may have serious AE(s) with multiple seriousness criteria. Percentages are calculated using total number of patients per treatment as the denominator. MedDRA version used for reporting: 25.0. Abbreviation: AE, adverse event. There were no abnormalities in laboratory parameters associated with on‐treatment SAEs, medication‐related AEs, or AEs leading to discontinuation of trial medication. Abnormalities in liver enzymes were reported with hepatic non‐serious AEs in two patients receiving BI 1358894 125 mg. There were no clinically meaningful differences between groups in the changes in vital signs over time. Systolic blood pressure was increased in one patient with placebo during the treatment period; this was mild in intensity and resolved in 29 days. There were no clinically meaningful differences between groups in the ECG parameters; no abnormalities were associated with on‐treatment SAEs or medication‐related AEs. There were no reports of self‐injurious or suicidal behavior. Efficacy Due to the trial being terminated early and the subsequent small patient numbers, there was insufficient statistical power to compare the efficacy end points between the treatment groups. The unadjusted mean change (SD) from baseline in the MADRS total score to Week 6 was ‐10.44 (7.2) with BI 1358894 125 mg and ‐13.35 (11.0) with placebo. Descriptive results for the secondary end points are presented in the Materials S1. Patient experience Of the 43 patients who participated in the feedback survey, 23 (54%) patients completed all survey questions. Overall, 86% of patients had not participated in a clinical trial before this DCT. The most common reason for previous non‐participation was a lack of awareness about clinical trial opportunities (60%), followed by concerns about missing time at work or school (8%), and concerns that participation could make their condition worse (8%; Figure 3a). The top four reasons patients chose to participate in this DCT included the convenience and flexibility of having study visits at their homes (34%), to help advance science and the treatment of MDD (26%), monetary compensation for participation (19%), and the opportunity to obtain treatment that would otherwise be inaccessible (11%; Figure 3b). Overall, 65% of patients indicated that they would not have participated in this trial if it had involved traveling to a doctor's office for all visits (Figure 3c). Most patients who completed the survey reported satisfaction with their experience of the trial (88%; Figure 3d); this overall level of satisfaction aligned with responses to individual questions in which most were positive (Table 3). Key questions and responses from the patient experience survey at EoS. The overall patient experience of the trial was assessed using a patient experience survey issued via a QR code at EoS or via an email sent on the day of, or the day after, EoS. The data have been rounded off to the nearest whole number. Some questions were multiple choice where patients could select more than one answer. Therefore, the number of responses may differ from the number of patients who completed the survey. DCT, decentralized clinical trial; EoS, end of study; QR, quick response. TABLE 3Additional questions and responses from the patient feedback survey at EoS.No.QuestionResponsesStrongly agreeAgreeNeutralDisagreeStrongly disagree1Reviewing and signing the informed consent electronically was easy65%27%–4%4%2I was well trained on the use of technology needed for the trial (such as phone, apps, video chats, etc.)62%23%12%–4%3Scheduling study visits was easy and did not take too long62%19%15%–4%4The site staff (mobile nurse, study coordinator, study doctor) were well prepared for each study visit73%19%4%–4%5The time taken to participate in each study visit was what I expected39%31%23%4%4%6The site staff (mobile nurse, study coordinator, study doctor) treated me with kindness, dignity, and respect88%4%4%–4%7I was satisfied with the technology provided to communicate with the site staff (mobile nurse, study coordinator, study doctor)81%8%4%4%4%8I was satisfied with the speed and adequacy of the site staff responding to my needs73%12%12%–4%9I was comfortable overall with the experience of having a mobile nurse come to my home for study visits81%12%4%–4%10I would have preferred to see a doctor face‐to‐face for SOME study visits–12%31%35%23%11I would have preferred to see a doctor face‐to‐face at ALL study visits––19%31%50%12I was comfortable talking to site staff on video chat about my illness and how I felt62%35%4%––13I was satisfied with answering the questionnaires about my illness and how I felt54%42%4%––14I was comfortable with having my interview with the site staff recorded for review by researchers69%27%4%–– Note: The overall patient experience of the trial was assessed using a patient experience survey issued via a QR code at EoS or via an email sent on the day of, or the day after, EoS. The data have been rounded off to the nearest whole number. Some questions were multiple choice where patients could select more than one answer. Therefore, the number of responses may differ from the number of patients who completed the survey.Abbreviation: EoS, end of study; QR, quick response. Additional questions and responses from the patient feedback survey at EoS. Note: The overall patient experience of the trial was assessed using a patient experience survey issued via a QR code at EoS or via an email sent on the day of, or the day after, EoS. The data have been rounded off to the nearest whole number. Some questions were multiple choice where patients could select more than one answer. Therefore, the number of responses may differ from the number of patients who completed the survey. Abbreviation: EoS, end of study; QR, quick response. Feasibility The DCT design was considered feasible in this first interventional phase II trial in patients with MDD, because all trial procedures were executed successfully as per the protocol. DISCUSSION This interventional DCT was performed to assess the safety and efficacy of adjunctive BI 1358894 125 mg vs. placebo in addition to SSRIs or SNRIs, in patients with MDD. Overall, BI 1358894 125 mg treatment was well‐tolerated, and the safety profile was consistent with the previous phase I trials of BI 1358894 125 mg 9 , 21 with no new safety signals reported. On‐treatment AEs were slightly higher with BI 1358894 125 mg vs. placebo; the most common AEs were headache, fatigue, and dizziness. Efficacy evaluation was not performed because of insufficient power to assess differences in efficacy between the treatment groups due to the early trial termination and associated low sample size. Although the trial was terminated early due to deficient enrollment over 22 months, it was feasible to conduct this fully remote interventional DCT in patients with MDD as all the trial procedures were successfully executed per protocol at the patients' homes. Recruitment methods are not a unique aspect of the DCT model other than the fact that any campaign respondent can participate regardless of location. This study had very stringent inclusion criteria, which made recruitment challenging for both site‐based local strategies as well as national campaigns conducted by Science 37 and third‐party recruitment vendors. Furthermore, the DCT approach was well perceived by trial participants (88%). Most patients were satisfied with the technology provided to communicate with the trial staff and were comfortable with having a mobile nurse visit their homes for study visits. Therefore, the overall positive patient feedback coupled with the high trial retention (97.7%) suggests that the DCT model may be valuable for engaging a patient population with MDD. This DCT had significant strengths. The Science 37 platform effectively collected trial information (in real‐time) and supported communication between trial staff and patients, and it was easy for patients to cancel or reschedule appointments using the platform. The AiCure application was convenient for patients to use, provided daily medication reminders and allowed remote monitoring of medication adherence and compliance. This DCT had a vast geographical reach, with 80,836 patients expressing interest by signing up for the trial on its website. Furthermore, the DCT had a high completion rate with only one randomized patient withdrawing from the trial; the patient who withdrew discontinued the trial treatment. The convenience of the DCT model allowed easy collection of the clinician‐ and patient‐reported outcomes through the provisioned smartphones, thus minimizing missing data. Lastly, the trial investigators and sub‐investigators successfully managed the complex logistics of this trial, including the shipping and coordination of the trial treatment and bio‐samples intended for laboratory evaluations to and from patients' homes. Despite these benefits, several limitations were also observed in this DCT. Firstly, recruitment was much slower than anticipated, particularly during the COVID‐19 pandemic, and the trial was unable to enroll the planned number of patients required to assess efficacy. Most screening failures were due to the stringent eligibility criteria, including duration of current depressive episode, past antidepressant medication failure, concomitant psychotropic drug use, and use of hypnotics. Despite the use of a pre‐screening questionnaire, not all relevant health information could be gathered, necessitating the assessment of a trained research professional to complete the screening process. Therefore, these screening failures were largely unavoidable in a first‐in‐patients decentralized phase II trial. For example, the use of bupropion, a commonly used antidepressant for treating MDD, was not allowed due to the unknown potential pharmacokinetic interaction between bupropion and BI 1358894. Furthermore, patients were required to have serum confirmation of antidepressant therapy; this blood test was outsourced by the central lab and had a long turnaround time resulting in the loss of otherwise qualified patients. Patients who received ≥4 unsuccessful antidepressant monotherapies, or those receiving low‐dose antidepressants for facilitating sleep in addition to their antidepressant were excluded. This resulted in many interested patients being ineligible, especially since patients with MDD frequently receive polytherapy. Thus, despite many patients expressing interest and signing up for this trial, this interest did not translate to facilitated recruitment into this phase II trial, although several recruitment optimizations were implemented. Contrasting results were reported in a pilot trial assessing the feasibility of home‐based trials using a wearable device for adolescent migraine data collection; broadening of inclusion criteria and increased social media advertisements helped to improve enrollment. 22 Interestingly, two fully remote studies, BRIGHTEN (V1 and V2) were able to successfully recruit a diverse patient group with depression (including the underserved Hispanic/Latino communities) using a combination of traditional, social networking, and search engine‐based methods. 23 , 24 , 25 However, the BRIGHTEN studies screened patients primarily based on smartphone or iPad ownership, 24 , 25 whereas the present DCT employed stringent medical inclusion and exclusion criteria. Although these criteria were required, given the context of an interventional trial with the administration of trial medication, this limited the conversion of recruited to enrolled patients. Another limitation was the hindered data collection resulting from the malfunction of the adherence app due to a technical fault or a lack of Wi‐Fi signal or data services to upload the collected data. Additionally, there were missing data when laboratory assessments or values could not be collected as some blood samples were either unstable or insufficient in quantity. Lastly, the placebo response in this trial could have been influenced by factors such as the decentralized setting and the context of the COVID‐19 pandemic, which may have influenced patient interactions and baseline symptom reporting. As a general observation, the discordance between medication adherence estimates using pill counting and the adherence app is noteworthy. While the AiCure rate reflects a conservative approach including only fully confirmed intakes and excluding suspected intakes, potentially underestimating true adherence, pill counting does the opposite. Given these limitations, several lessons were learned from planning and executing this DCT in patients with MDD that would benefit the development of future DCTs. Firstly, it is vital to match the DCT model to the stage of development; a DCT approach alone may not be sufficient to overcome the challenges caused by the stringent eligibility criteria that are necessary in early‐stage drug development, that is, a first‐in‐patients, interventional phase II trial. Secondly, digital technology is not always reliable and relies on adequate Wi‐Fi signal to upload data. Thirdly, given the overall positive feedback from patients, DCTs or hybrid trials (combining elements from both DCTs and traditional trials) should be considered in the future to ensure the inclusion of a broad and diverse patient population. Indeed, a recent non‐interventional DCT in patients with MDD took a decentralized approach and its investigators favored hybrid trials, to leverage the advantages of traditional and digital methods. 26 Fourthly, a comprehensive recruitment plan, especially one that includes early implementation of BYOI, expansion of social media channels, and continuous strategy optimization may also accelerate recruitment in a DCT. See Table 4 for an overview of lessons learned and mitigation/optimization strategies from the implementation of a DCT in MDD. TABLE 4Lessons learned and mitigation/optimization strategies from the implementation of a decentralized clinical trial (DCT) in major depressive disorder.IssueLessonMitigation/optimizationRecruitment challengesDespite a large outreach effort, recruitment remained deficient (only 45 of 13,816 pre‐screened patients were enrolled)Develop a more comprehensive and targeted recruitment plan early, including optimizing eligibility criteria and expanding recruitment channelsStringent eligibility criteriaMany patients were excluded due to strict inclusion/exclusion criteria required for early‐stage drug development, particularly around antidepressant useConsider modifying eligibility criteria to be more inclusive in future trials; match the DCT model to the stage of development, potentially leveraging hybrid modelsTechnology‐related issuesData collection was hindered by third‐party application malfunctions and inadequate Wi‐Fi in some casesEnsure robust testing and alternative data upload solutions, such as offline data collection with later synchronizationGeographical reachThe DCT design enabled a broad geographical outreach across 35 states, creating interest in participationMaintain and expand the geographical flexibility in future DCTs, while optimizing patient engagement through localized strategiesPatient retention and satisfactionHigh patient retention (97.7%) and satisfaction (89.0%) demonstrated the feasibility of remote trial participationLeverage positive patient feedback to enhance future DCT models, including hybrid approaches to balance patient convenience and trial needs Lessons learned and mitigation/optimization strategies from the implementation of a decentralized clinical trial (DCT) in major depressive disorder. The DCT model with its vast geographical reach may be particularly suitable for trials designed for specific difficult‐to‐recruit target populations, such as racial and ethnic minority groups, or those with mental disorders or rare diseases. However, potential participants would likely benefit from a detailed explanation of the DCT concept before enrollment, specifically for interventional trials involving a pharmacotherapy. Further investment in new and innovative trials leveraging decentralized elements, and in understanding the patient's experience of DCTs could shift the focus of clinical trial research towards meeting individual patient needs. CONCLUSIONS To the best of our knowledge, this first interventional DCT was feasible in patients with MDD, and it was well perceived by the participating patients. Data indicate that daily BI 1358894 125 mg treatment over 6 weeks was generally well‐tolerated and safe. Efficacy evaluation was not performed due to the limited number of enrolled patients. The learnings from this DCT will inform the development of future interventional DCTs or hybrid trials and may allow for the inclusion of broad and diverse patient populations. AUTHOR CONTRIBUTIONS T.L.N. designed the research; T.L.N., S.D.S., P.L., and C.R. performed the research; T.L.N., S.D.S., P.L., and C.R. analyzed the data; T.L.N., S.D.S., P.L., and C.R. wrote the manuscript. FUNDING INFORMATION This study was funded by Boehringer Ingelheim International GmbH (BI study number: 1402–0014; ClinicalTrials.gov identifier: NCT04423757). CONFLICT OF INTEREST STATEMENT C.R. is an employee of Science 37. P.L., T.L.N. are employees of Boehringer Ingelheim Pharmaceuticals, Inc. S.D.S. is an employee of Boehringer Ingelheim International GmbH and member of the Faculty of Medicine of the University Hospital of Ulm, Germany. Supporting information Data S1: