Safety analyses from the phase 3 ASCENT trial of sacituzumab govitecan in metastatic triple-negative breast cancer

Sacituzumab govitecan (SG) is an anti-Trop-2 antibody-drug conjugate with an SN-38 payload. In the ASCENT study, patients with metastatic triple-negative breast cancer (mTNBC) relapsed/refractory to ≥2 prior chemotherapy regimens (≥1 in the metastatic setting), received SG or single-agent treatment of physician’s choice (eribulin, vinorelbine, capecitabine, or gemcitabine). This ASCENT safety analysis includes the impact of age and UGT1A1 polymorphisms, which hinder SN-38 detoxification. SG demonstrated a manageable safety profile in patients with mTNBC, including those ≥65 years; neutropenia/diarrhea are key adverse events (AE). Patients with UGT1A1 *28/*28 genotype versus those with 1/*28 and *1/*1 genotypes had higher rates of grade ≥3 SG-related neutropenia (59% vs 47% and 53%), febrile neutropenia (18% vs 5% and 3%), anemia (15% vs 6% and 4%), and diarrhea (15% vs 9% and 10%), respectively. Individuals with UGT1A1 *28/*28 genotype should be monitored closely; active monitoring and routine AE management allow optimal therapeutic exposure of SG.


INTRODUCTION
Triple-negative breast cancer (TNBC) comprises approximately 15% of all breast cancers and is associated with a poor prognosis [1][2][3] .Due to absent hormone and HER2 receptors, targeted strategies used in other types of breast cancer are not effective for metastatic TNBC (mTNBC), and current treatment guidelines recommend single-agent chemotherapy for patients without known biomarkers associated with targeted therapy 2,4 .Patients with TNBC have poorer outcomes compared with those with other breast cancer subtypes, including higher and earlier risk of relapse and shorter survival 5,6 .Among patients with previously treated mTNBC receiving single-agent chemotherapy, median progression-free survival (PFS) is only approximately 2-3 months, with median overall survival (OS) of approximately 8-15 months [7][8][9][10] .Common adverse events (AEs) with singleagent chemotherapy include hematologic AEs such as neutropenia, as well as diarrhea, nausea, and alopecia 8,11 .Although TNBC is more common in younger compared with older breast cancer patients, a significant number of TNBCs arise in older patients due to an aging population 12,13 .Approximately 20% of patients diagnosed with TNBC are ≥65 years old 14,15 .Furthermore, older patients are generally less fit for chemotherapy due to a greater rate of comorbidities, increased use of concomitant medications, and the potential for greater impact on health-related quality of life 16,17 .Due to the limited survival outcomes with single-agent chemotherapy regimens and the potential for increased AEs in older patients, there is an unmet medical need for novel targeted agents for all patients with pretreated mTNBC.
Trophoblast cell surface antigen-2 (Trop-2) is a calcium signal transducer that has been linked to poor outcomes in multiple cancer types 18 .Overexpression of membrane-localized Trop-2 has been associated with poor prognosis and increased tumor growth in breast cancer 19,20 .Trop-2 is the target of sacituzumab govitecan (SG), an antibody-drug conjugate (ADC) in which the payload of SN-38, a topoisomerase-1 inhibitor and the active metabolite of irinotecan, is conjugated with an anti-Trop-2 humanized monoclonal antibody via a proprietary hydrolyzable linker 21 .This allows for cytotoxic SN-38 liberation in the acidic tumor microenvironment without prerequisite internalization and subsequent enzymatic cleavage 22 , enabling "bystander effect" tumor cell killing 23 .Upon binding, its drug-antibody ratio of 7.6:1 allows the release of a high localized SN-38 concentration from SG without adversely impacting binding or pharmacokinetics 21,24 .
Neutropenia and diarrhea are toxicities associated with irinotecan, attributable to its active metabolite, SN-38, which is the cytotoxic payload of SG 21,27 .The enzyme uridine diphosphate glucuronosyltransferase 1A1 (UGT1A1) plays a key role in detoxifying SN-38 by glucuronidation 24 , producing a metabolite that is then eliminated from the body primarily by biliary excretion 28 .UGT1A1 activity is reduced in the UGT1A1 *28/*28 phenotype, which is found in approximately 10% of North American patients; this polymorphism leads to a higher risk of neutropenia and diarrhea with irinotecan therapy [27][28][29] .For this reason, FDA guidelines call for a reduction in the irinotecan starting dose in patients with colorectal cancer with known UGT1A1 *28/*28 status 27,28,30 , and European Society for Medical Oncology guidelines recommend UGT1A1 genotyping in patients with metastatic colorectal cancer for whom an irinotecan dose >180 mg/m 2 is planned and there is a suspicion of UGT1A1 deficiency, as reflected by low conjugated bilirubin 31 .
As part of the continuing safety evaluation of SG, we conducted a detailed and updated safety analysis of AEs of interest, an analysis of time to onset and duration of key TRAEs, and examined TRAEs in older patients and by UGT1A1 variant status.

Patient disposition and baseline characteristics
Of 529 patients randomized in ASCENT, 9 (3%) and 38 (15%) in the SG and TPC arms, respectively, did not receive treatment (Supplementary Fig. 1).The safety population consisted of 482 patients (of whom 53 had brain metastases) who received at least one dose of study treatment, including 258 in the SG arm and 224 in the TPC arm.Among patients in the safety population who were treated with TPC, 123 (55%), 28 (13%), 32 (14%), and 41 (18%) received eribulin, capecitabine, gemcitabine, and vinorelbine, respectively.At data cutoff (March 11, 2020), 17 and 0 patients remained on treatment in the SG and TPC arms, respectively.In both arms, the primary reason for discontinuation (SG vs TPC) was disease progression (86% vs 82%).
Patient baseline characteristics were well-balanced between arms (Table 1).The median age was 54 (range 27-82), the median number of prior anticancer regimens (any setting) was 4 (range 2-17), and all patients had previously received a taxane.There were 49 (19%) and 48 (21%) patients aged ≥65 years in the SG and TPC arms, respectively; of these, 8 and 13 patients were aged ≥75 years.
in the SG vs TPC arms had treatment-related grade ≥3 neuropathy.No patients experienced treatment-related interstitial lung disease of any grade with TPC.In the SG arm, there was 1 event (0.4%) of grade 3 pneumonitis complicated by prior radiation and progressive disease.This patient was a 52-year-old woman who experienced the event 14 days after her last dose of SG; it resolved 7 weeks after onset without sequalae 25 .SG-related grade ≥3 rash (<1%), all-grade ocular disorders (<5%), and hyperglycemia (<2%) were also infrequent.Rates of discontinuation due to TRAEs were low in both the SG (2%) and TPC (3%) arms.No treatment-related deaths were reported in the SG arm, and 1 death (neutropenic sepsis related to eribulin) was reported in the TPC arm.There were no discontinuations due to SG-related neutropenia, febrile neutropenia, or diarrhea.
Neutropenia.Median time to onset of the first event of treatment-related grade ≥3 neutropenia was 21 days and 14 days for the SG and TPC arms, and median duration of an individual episode of grade ≥3 neutropenia was 6.0 and 6.5 days, respectively (Fig. 1).Treatment-related neutropenia of all grades was more frequent in the SG arm compared with the TPC arm (any grade: 63% vs 43%; grade ≥3: 51% vs 33%) (Fig. 2A).In the first treatment cycle with SG vs TPC, 43% vs 33% and 28% vs 23% of patients had any grade and grade ≥3 neutropenia, respectively; 4% vs 0.4% of patients in the SG vs TPC arms had febrile neutropenia, all grade ≥3.Myeloid growth factor use was more common in the SG vs TPC arm (total use, 49% vs 23%; for secondary prophylaxis, 29% vs 10%; as treatment, 30% vs 17%; Fig. 2B), was most common in the first 3 cycles of treatment, and treatment was less common in later cycles (<10% of SG-treated patients; Fig. 2C).Dose interruptions due to treatment-related neutropenia or febrile neutropenia occurred in 46% and 21% of patients in the SG and TPC arms, respectively, whereas dose reductions for this reason occurred in 11% and 19% of patients, respectively.Treatment-related neutropenia and febrile neutropenia led to discontinuations in 0% vs 1% and 0% vs <1% of patients in the SG vs TPC arms, respectively.Diarrhea.Median times to onset of the first event of treatmentrelated grade ≥3 diarrhea were 19 days and 26.5 days for the SG and TPC arms, and median durations of an individual episode of grade ≥3 diarrhea were 5 days and 1 day, respectively (Fig. 1).
Treatment-related diarrhea of all grades was more frequent in the SG arm (any grade: 59% vs 12%; grade ≥3: 10% vs 0.4%) (Fig. 2A).Grades 1 and 2 diarrheas occurred in 30% vs 7% and 19% vs 5% of SG-vs TPC-treated patients.There were no events of grade 4 or 5 diarrhea.Concomitant medicine was used for diarrhea management in 55% vs 10% of SG-vs TPC-treated patients.In the SG vs TPC arms, 55% vs 8% received loperamide and 10% vs 2% received atropine, respectively.Dose interruptions due to treatment-related diarrhea occurred in 5% of patients in the SG arm and no patients in the TPC arm, whereas dose reductions for diarrhea occurred in 5% and <1% of patients, respectively.No treatment discontinuations due to treatment-related diarrhea occurred in either arm.
Nausea and vomiting.In the SG vs TPC arms, median time to onset of the first event of any grade nausea and vomiting was 8 vs 8 days and 24.5 vs 17 days after treatment initiation, respectively (Fig. 1); however, the frequencies of grade ≥3 nausea and vomiting were low (<3%).Premedication or concomitant medication for nausea and vomiting, such as 5-HT3 antagonists (e.g., ondansetron) and receptor antagonists (e.g., prochlorperazine), was used in 86% and 63% of patients in the SG and TPC arms, respectively.
Other adverse events.Among other TRAEs, alopecia (46% vs 16%) was more common in the SG arm.Any grade (45% vs 30%) but not grade ≥3 fatigue (3% vs 5%) was also more common in the SG arm.

Efficacy by dose reductions and interruptions
Efficacy endpoints were assessed in the primary study analysis population of brain metastases-negative patients who received at least one dose of SG (10 mg/kg) or TPC.Dose interruptions occurred in 61% and 33% of patients in the SG and TPC arms, respectively, and dose reductions occurred in 22% and 26%, respectively.In the SG arm, efficacy outcomes for patients with dose reductions or interruptions were similar to those of patients without dose reductions or interruptions.SG treatment was associated with improved ORR, clinical benefit rate (CBR), and PFS compared with TPC in patients with and without dose interruptions, and with and without dose reductions (Table 3).Median PFS was 8.3 vs 2.9 months and 4.6 vs 1.5 months in SG-vs TPC-treated patients with and without dose reductions, respectively; it was 5.7 vs 2.7 months and 4.2 vs 1.6 months in SG-vs TPC-treated patients with and without dose interruptions, respectively.
Table 2. Treatment-related adverse events of all grades reported in >20% and of grade 3 or 4 reported in >5% of patients by age.Combined preferred terms of "anemia", "hemoglobin decreased", and "red blood cell count decreased." Combined preferred terms of "leukopenia" and "white blood cell count decreased."

DISCUSSION
With results from the confirmatory phase 3 ASCENT study 25 , SG became the first ADC to demonstrate a significant survival improvement relative to standard single-agent chemotherapy in patients with pretreated mTNBC.Key AEs associated with SG included neutropenia and gastrointestinal toxicity, AEs expected with an SN-38 payload.The present analysis further elucidates the SG safety profile and AE management in patients with mTNBC, with special relevance to older individuals and those with UGT1A1 *28 homozygous genotype, due to the potential role of UGT1A1 genotype as a predictor of toxicity.The safety profile of SG is consistent with previous reports 23,32,33 .In the overall study population (median age, 54 years; median of 4 prior anticancer regimens), TRAEs were primarily hematologic (neutropenia; grade ≥3, 51%) and gastrointestinal (diarrhea; grade ≥3, 10%).Use of supportive care medications was effective in managing these AEs, including myeloid growth factors and antipropulsive agents, and discontinuation due to toxicity was rare.Interestingly, in a comparison of this study population with the small subgroup of patients in the ASCENT trial who received SG in the second-line (2 L) metastatic setting (i.e., those who received only 1 line of therapy in the metastatic setting following recurrence ≤12 months after [neo] adjuvant chemotherapy; median of 3 prior anticancer regimens in any treatment setting, including the [neo]adjuvant setting), rates of key TRAEs were generally similar 34 .However, rates of SG-related all-grade alopecia were lower in the overall study population compared with that of the 2 L subgroup (Table 2; 46% vs 70% 34 ).Because older age is a risk factor for chemotherapy toxicity, impacting treatment decisions and increasing the risk for AEs 35,36 , it is essential to understand the potential safety risks with SG in older patients.Irrespective of age, toxicities associated with SG were manageable, and the safety profile of SG in patients aged ≥65 and ≥75 years was generally similar to that of patients aged <65 years and the overall study population.
Of note, the observed delayed appearance of nausea and vomiting (median of 8 and 24.5 days after initiation of SG treatment, respectively) should be considered when managing treatment-associated symptoms.Patients should be provided with take-home medications for preventing and treating delayed nausea and vomiting, with clear instructions 26 .Antiemetics should be employed as clinically indicated, and atropine can be used to manage excessive cholinergic response 26 .Other AEs in the SG arm included alopecia (46%), which may affect quality of life 37 .Although therapeutic scalp cooling is effective for many chemotherapeutic agents, its efficacy with ADCs is unknown 38,39 .SG response and PFS advantage relative to TPC was maintained in patients who started at 10 mg/kg and then required SG dose reductions/interruptions to manage AEs; efficacy outcomes for patients with dose reduction or interruptions in the SG arm were similar to those for the overall study population 25 .In the ASCENT protocol, dose reduction/interruption recommendations for the TPC arm were based on local prescribing information or institutional standard practice, with no specific recommendations for myeloid growth factor use.A higher frequency of patients in the SG vs TPC arms received myeloid growth factors (49% vs 23%) and dose interruptions (61% vs 33%), which may have been due to heterogeneity in local guidelines for the different agents in the TPC arm; the frequency of patients with dose reductions (22% vs 26%, respectively) was similar.However, differences in the frequency of patients with dose reductions/interruptions should not affect the efficacy analysis for these groups, as there was a clinical benefit for SG in both patients with and without dose reductions/interruptions.These results support the recommended initial dosing of SG at 10 mg/kg followed by as-needed dose reductions and interruptions, over initiating treatment at a lower dose, which has the potential to compromise efficacy benefit 24 .
Table 3. Efficacy outcomes for patients with dose reductions or interruptions (brain metastasis-negative population).

Outcome
Median PFS (BICR), mo (95% CI) 5.6 (4.3-6.3)1.7 ( Interestingly, median PFS was numerically higher in patients who received SG or TPC and had dose reductions/interruptions vs those who did not have dose reductions/interruptions.However, lack of pharmacokinetics, the small numbers of patients in these subgroup analyses, and the potential for confounding effects associated with dose reductions/interruptions limit further interpretation of these data.Although hematologic toxicities are concerns for all the treatments administered in ASCENT, the safety profile of SG is distinct from those of eribulin, vinorelbine, capecitabine, and gemcitabine.Importantly, there were no instances of SG-related grade ≥3 neuropathy and low frequencies of SG-related interstitial lung disease (1 case of grade 3 pneumonitis, <1%), grade ≥3 rash (<1%), all-grade ocular disorders (<5%), and hyperglycemia (<2%); no SG-related deaths occurred.In comparison, peripheral neuropathy is a safety concern for eribulin, with 8% of patients who received single-agent eribulin in a metastatic breast cancer trial experiencing grade ≥3 events; peripheral neuropathy was also the most common adverse reaction resulting in discontinuation of eribulin (5%) in this trial 40 .Peripheral neuropathy is also a concern for single-agent vinorelbine, with clinical trial data showing 1% of patients experiencing grade ≥3 events 41 .Pulmonary toxicity, including fatal pneumonitis, has also been reported with vinorelbine 41 .Pulmonary toxicity and associated respiratory failure are also a major safety concern for gemcitabine 42 .In patients who receive single-agent capecitabine for metastatic breast cancer, a major safety concern is skin toxicity, with 57% and 11% experiencing capecitabine-related all-grade and grade 3 handand-foot syndrome, respectively, during clinical trials 43 .
Although not yet approved for TNBC, other ADCs being evaluated for breast cancer have distinct safety profiles.As with SG, common AEs with other ADCs include hematologic and gastrointestinal toxicities 44,45 .AEs associated with ADCs under investigation but not with SG include interstitial lung disease, neuropathy, ocular disorders, rash, and hyperglycemia 44,45 .These safety profile differences may be due in part to the payload of each ADC; with an SN-38 payload, SG has demonstrated a lower rate of diarrhea than that of SN-38's prodrug, irinotecan 22 .
Our examination of potential UGT1A1 polymorphism effects on SG toxicity was prompted by the enzyme's known role in mediating the toxicity of SN-38, the cytotoxic species delivered to cells by SG.Clearance of SN-38, a lipophilic molecule, largely depends on its glucuronidation by UGT1A1, by which it is made water soluble for subsequent excretion 27 .In irinotecan-treated patients with colorectal cancer who have a UGT1A1 polymorphism that limits SN-38 glucuronidation, increased SN-38 exposure results in higher rates of neutropenia and late-onset diarrhea 29,46 , the dose-limiting toxicities associated with irinotecan 47 .One such report found that patients receiving the combination of capecitabine and irinotecan with the *28 homozygous genotype were 14.2 times more likely to experience febrile neutropenia than those with the wild-type genotype 48 .
In line with other reports, the *28 homozygous genotype was found in a relatively small percentage of ASCENT patients (13%).As in previous irinotecan studies, rates of hematologic toxicities and diarrhea in ASCENT were modestly higher for patients with the UGT1A1 *28 homozygous genotype than for those with the wild-type genotype.Specifically, rates of grade ≥3 neutropenia, febrile neutropenia, anemia, and diarrhea were higher in the *28 homozygous group compared with the heterozygous and wildtype groups.Despite this, the discontinuation rate for SG due to TRAEs for patients with the *28 homozygous genotype was low (6%), and no patients of any UGT1A1 genotype discontinued SG due to SG-related neutropenia or diarrhea, suggesting favorable tolerability when using the current AE management strategies of active monitoring and early intervention (e.g., dose reductions and supportive medication use).These results imply that the currently recommended SG starting dosage (10 mg/kg) is appropriate regardless of patient UGT1A1 genotype.Although this detailed safety analysis of ASCENT provides evidence of a manageable safety profile in patients with UGT1A1 polymorphisms, the relatively low frequency of the UGT1A1 *28 homozygous mutation Table 4. Key treatment-related adverse events of all grades in >20% and of grade ≥3 in >5% of patients treated with sacituzumab govitecan significantly impacted by UGT1A1 genotype.limits the ability to discern differences or draw firm conclusions.Although prescreening for UGT1A1 genotype is not required, individuals with known UGT1A1 homozygous *28/*28 genotype should be monitored closely for neutropenia and diarrhea, regardless of age.In summary, the results of this expanded ASCENT safety analysis show that SG has a manageable safety profile, irrespective of age, with an AE profile consistent with prior reports.These data confirm prior observations 24 , suggesting that UGT1A1 variant status should not significantly alter recommendations for SG treatment or AE management in this patient population.Initial dosing at 10 mg/kg is recommended, with dose reductions as needed for toxicity, and these data suggest that dose reductions to manage toxicity do not appear to impact efficacy.Active monitoring, prophylaxis, and early intervention with routine AE management strategies like dose reductions allow for optimal therapeutic exposure of SG for patients with pretreated mTNBC.

Study design and participants
Detailed methods for ASCENT (NCT02574455; registered October 12, 2015), a phase 3 randomized, open-label multicenter study, have been reported 25 .ASCENT enrolled patients with mTNBC (per American Society of Clinical Oncology/College of American Pathologists criteria 49 who relapsed after or were refractory to ≥2 previous standard chemotherapy regimens (no upper limit) for unresectable, locally advanced or metastatic disease.Patients whose disease recurred ≤12 months after completing (neo)adjuvant therapy were considered as having had one prior line of therapy.Screening for brain metastasis was not mandatory, and patients with brain metastases stable for ≥4 weeks before treatment were eligible.Patients were randomized 1:1 to SG 10 mg/kg intravenously on Days 1 and 8 of each 21-day cycle, or TPC (eribulin; vinorelbine; capecitabine; or gemcitabine) and received treatment until disease progression or unacceptable toxicity.Severe neutropenia and non-neutropenic toxicities were managed with dose delays or reductions and medications, as shown (Supplementary Fig. 2).Prophylactic growth factors at study start were not allowed, and antidiarrhea prophylaxis was not recommended.
Crossover to the SG arm was not allowed upon disease progression in the TPC arm.Patients were stratified by number of prior therapies (2-3 vs >3), geographic region (North America vs Europe), and the presence or absence of known brain metastasis (Yes or No).The primary endpoint was PFS in patients without known brain metastases by independent, centralized, and blinded review per Response Evaluation Criteria in Solid Tumors (RECIST), version 1.1.Secondary endpoints included safety.Exploratory endpoints included safety by UGT1A1 variant status.
The study was approved by each institutional review board or ethics committee prior to initiation and was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice Guidelines per the International Conference on Harmonization.A full list of institutions which granted ethical approval can be found in the Supplement.All patients provided written informed consent.

Assessments
Response was evaluated as reported previously 50 .AEs were assessed at each study visit.Complete blood counts (CBC) and serum chemistries were performed at baseline and at last study visit for all treatments, and prior to infusion of SG or TPC agents administered intravenously (eribulin, gemcitabine, and vinorelbine), and at the start of each treatment cycle and as required per local standard of care for TPC agents administered orally (capecitabine).In the event of grade ≥3 hematologic toxicity, CBCs were obtained more frequently at physician's discretion until toxicity recovered to baseline or grade 1. Descriptive statistics were used to characterize AEs.AEs were assessed per the National Cancer Institute Common Terminology Criteria for AEs, version 4.03, and coded per the Medical Dictionary for Regulatory Activities, version 22.1.At baseline, a single whole blood sample was collected from all patients receiving SG for determination of UGT1A1 genotype for retrospective assessment of safety.UGT1A1 genotype was determined by Sanger sequencing, performed centrally.The relationship between the incidence of AEs of special interest to SG, including hematologic events and diarrhea, and UGT1A1 genotype were of particular focus.

Statistical analysis
The safety population consisted of all patients who received ≥1 dose of study drug.Relative dose intensity was calculated by dividing the cumulative dosage received (in mg/kg) by the product of the assigned dose (in mg/kg) based on the number of doses the patient was scheduled to receive during the patient's treatment period.Time to onset of first AEs was defined as time from first dosing date to the start date of the AE.Duration of an individual episode of AE was calculated as the AE end-date minus the AE start-date.Median of the individual episodes of AE was used for calculation at the population level for patients with multiple episodes.PFS was defined as time from randomization until objective tumor progression or death or censored at last radiographic assessment for patients without progression or death.PFS was analyzed by the Kaplan-Meier method; 95% CIs were determined according to the method of Brookmeyer and Crowley.

DATA AVAILABILITY
Gilead Sciences shares anonymized individual patient data upon request or as required by law or regulation with qualified external researchers based on submitted curriculum vitae and reflecting non conflict of interest.The request proposal must also include a statistician.Approval of such requests is at Gilead Science's discretion and is dependent on the nature of the request, the merit of the research proposed, the availability of the data, and the intended use of the data.Data requests should be sent to datarequest@gilead.com.

Fig. 1
Fig.1Time course of treatment-related adverse events of special interest.a Time to onset of first event of treatment-related AESI and (b) duration of an individual episode of treatment-related AESI of any grade and grade ≥3 in the Safety Population.Box and whisker plots, with upper and lower boundaries of each box plot representing the 25th and 75th percentiles and the horizontal lines within the box representing median values.Whiskers extend to the last observation if it was not an outlier (defined as greater than Q3 + 1.5 × IQR or less than Q1-1.5 × IQR) or to the minimum/maximum values if an outlier was not identified.Outliers are indicated by an asterisk.AESI adverse event of special interest, IQR interquartile range, Q1 first quartile, Q3 third quartile, SG sacituzumab govitecan, TPC treatment of physician's choice.

Fig. 2
Fig. 2 Incidence of neutropenia and diarrhea, and G-CSF use. a Incidence of neutropenia and diarrhea by grade, b use of G-CSF in the safety population, and c use of G-CSF by cycle in the safety population.Percent of patients using G-CSF by treatment cycle is based on the total number of patients treated in cycle 1 (SG, n = 258; TPC, n = 224).G-CSF granulocyte colony-stimulating factor, SG sacituzumab govitecan, TPC treatment of physician's choice.

Table 1 .
Patient demographics and disease characteristics at baseline (Safety Population).