Fax: (216) 444-9464
Sorafenib in patients with metastatic renal cell carcinoma refractory to either sunitinib or bevacizumab†
Article first published online: 30 AUG 2010
Copyright © 2010 American Cancer Society
Volume 116, Issue 23, pages 5383–5390, 1 December 2010
How to Cite
Garcia, J. A., Hutson, T. E., Elson, P., Cowey, C. L., Gilligan, T., Nemec, C., Dreicer, R., Bukowski, R. M. and Rini, B. I. (2010), Sorafenib in patients with metastatic renal cell carcinoma refractory to either sunitinib or bevacizumab. Cancer, 116: 5383–5390. doi: 10.1002/cncr.25327
Presented in part at the ASCO Genitourinary Cancer Symposium, San Francisco, California, February 14-16, 2008 (Abstract 346), and American Society of Clinical Oncology Annual Meeting, Chicago, Illinois, May 30-June 3, 2008 (Abstract 5123).
- Issue published online: 23 NOV 2010
- Article first published online: 30 AUG 2010
- Manuscript Accepted: 13 JAN 2010
- Manuscript Revised: 11 JAN 2010
- Manuscript Received: 5 NOV 2009
- metastatic renal cell carcinoma refractory;
Bevacizumab and sunitinib are standard initial therapy in metastatic renal cell carcinoma (mRCC). Despite common use, the safety and activity of sorafenib in bevacizumab- or sunitinib-refractory mRCC have not been prospectively investigated.
Metastatic RCC patients with Response Evaluation Criteria in Solid Tumors (RECIST)-defined disease progression (PD) after treatment with either bevacizumab or sunitinib received twice daily 400 mg of sorafenib in a multicenter, prospective phase 2 study. Dose escalation was permitted in the absence of significant toxicity. The primary endpoint was tumor burden reduction rate, defined as the proportion of patients with ≥5% reduction in the sum of RECIST-defined target lesions without other PD. Secondary endpoints included progression-free survival (PFS), duration of response, overall survival, and safety. A 2-stage accrual design was used to test the alternative hypothesis that the tumor burden reduction rate was >20% versus <5%.
Forty-eight patients were enrolled. The tumor burden reduction rate was 30% (95% confidence interval [CI], 17%-45%). One unconfirmed objective partial response was observed. The median PFS was 4.4 months (95% CI, 3.6-5.9). There was no association of PFS and tumor shrinkage with response to prior therapy. Most treatment-related adverse events were of mild-to-moderate intensity, and included fatigue, hypertension, diarrhea, and palmoplantar erythrodysesthesia (PPE). Patients previously treated with bevacizumab tended to develop more PPE (P = .03) and mucositis (P = .06), whereas sunitinib-treated patients tended to develop more skin rash (P = .06).
Administration of sorafenib is safe and feasible in patients with mRCC refractory to either bevacizumab or sunitinib. Modest clinical activity was observed supporting current practice patterns of sequential vascular endothelial growth factor-targeted therapy in mRCC. Cancer 2010. © 2010 American Cancer Society.
Recent advances in the management of metastatic renal cell carcinoma (mRCC), have established the vascular endothelial growth factor (VEGF)-targeted agents sunitinib (Sutent, Pfizer, New York, NY), pazopanib (Votrient, GlaxoSmithKline, Philadelphia, Pa), and bevacizumab (Avastin, Genentech, South San Francisco, Calif) plus interferon as standard initial treatment options.1-4 For mRCC patients who develop progressive disease (PD) or drug intolerance after front-line therapy with either sunitinib or sorafenib, everolimus (Affinitor, Novartis Pharmaceuticals, East Hannover, NJ), an oral mammalian target of rapamycin (mTOR) inhibitor, is currently an accepted standard of care in this setting.5 Other retrospective series have also demonstrated antitumor activity when VEGF-targeted agents are used in a sequential manner.6-8 These data, albeit limited by their sample size, retrospective nature, and inherent bias, serve as a proof-of-concept that sequential VEGF-targeted therapy is a feasible and safe approach that can also lead to meaningful antitumor responses in mRCC patients exposed to prior VEGF-targeted therapy.
Sorafenib (Nexavar, Bayer Pharmaceuticals, West Haven, Conn and Onyx Pharmaceuticals, Emeryville, Calif) is an orally bioavailable bi-aryl urea Raf kinase capable of inhibiting VEGF receptor (VEGFR) 1, 2, and 3 and platelet-derived growth factor receptor B,9 with demonstrated clinical activity in advanced RCC, most notably in cytokine-refractory patients.10-12 As initial treatment with cytokines has become significantly less common in mRCC, and in light of data failing to demonstrate a progression-free survival (PFS) benefit of sorafenib over interferon in the front-line setting,12 sorafenib is often used in the second-line setting after prior VEGF-targeted treatment. This practice algorithm is common, although it has not been prospectively investigated. With the availability of multiple new and active compounds in RCC, prospective investigation of sequential therapy is crucial to define the antitumor activity in patients with advanced disease. Thus, a prospective phase 2 study of sorafenib in patients with bevacizumab- or sunitinib-refractory mRCC was undertaken.
MATERIALS AND METHODS
Eligibility criteria included histologically documented mRCC with predominant clear cell histology, prior nephrectomy, Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 1, measurable disease, and evidence of PD after treatment with either sunitinib or bevacizumab as defined by Response Evaluation Criteria in Solid Tumors (RECIST) criteria.13 Patients treated with sunitinib must have received at least 1 cycle (4 weeks) of treatment and have RECIST-defined objective progression during or within 4 months of completing sunitinib treatment. Similarly, patients treated with bevacizumab must have received at least 2 doses of bevacizumab and have RECIST-defined objective progression during or within 4 months of completing bevacizumab treatment. Adequate renal, hepatic, and bone marrow function was required as defined by a serum creatinine ≤2.0 × institutional upper limit of normal (ULN), total bilirubin ≤1.5 × ULN, aspartate aminotransferase/alanine aminotransferase ≤2.5 × ULN , leukocytes ≥3000/μL, absolute neutrophil count ≥1500/ μL, and platelets ≥75,000/μL. Patients with previously treated central nervous system (CNS) metastases were eligible, provided they had no evidence of CNS disease progression on imaging performed at least 2 weeks after definitive treatment of their CNS disease. There were no restrictions regarding the number of previous systemic treatments received. Exclusion criteria included prior systemic treatment with sorafenib and significant cardiovascular disease, including congestive heart failure (New York Heart Association class III or IV), active angina pectoris requiring nitrate therapy, uncontrolled arrhythmias, or a recent cardiovascular event (defined as any of the following within the previous 6 months: transient ischemic attack/cerebrovascular accident, myocardial infarction, vascular surgery).
The Cleveland Clinic Institutional Review Board reviewed and approved this study in accordance with the Declaration of Helsinki and the International Conference on Harmonization Good Clinical Practice guidelines. All patients provided written informed consent before registration.
Baseline evaluations included medical history and physical examination, laboratory parameters (hematology and blood chemistry), and tumor imaging (computed tomography [CT] scan or magnetic resonance imaging [MRI] of the chest, abdomen, and pelvis; bone scan; and brain CT or MRI scan). In addition, patients self-measured blood pressure daily throughout the study and notified their physician if systolic blood pressure was >150 mm Hg or diastolic blood pressure was >90 mm Hg.
Patients received sorafenib 400 mg orally, twice daily, continuously in repeated 4-week cycles. Treatment was continued until RECIST-defined disease progression, unacceptable toxicity, or withdrawal of consent. Sorafenib dose reduction to 400 mg daily and then to 400 mg every other day was allowed on an individual basis depending on tolerability. Sorafenib dose escalation to either 600 or 800 mg twice daily was also permitted within the study. Dose escalations were implemented at any time after the 8-week disease reassessment at the discretion of the treating physician. Patients experiencing any grade 4 toxicity were not allowed to dose escalate.
The primary endpoint was the percentage of tumor burden reduction rate defined as the proportion of patients with ≥5% reduction in the sum of the RECIST-defined target lesions without other progression defined by RECIST. This endpoint was chosen in light of the low objective response rate of sorafenib, the unknown PFS of sorafenib in this setting, and the understanding that tumor burden reduction not meeting criteria for a RECIST-defined objective response can be used to gauge the antitumor activity of targeted therapy in mRCC.
Tumor assessments were performed according to RECIST criteria at baseline, then every 8 weeks or earlier if tumor progression was clinically suspected. Secondary endpoints included PFS, duration of response, overall survival (OS), and safety. Adverse events (AEs) were reported and graded according to Common Terminology Criteria for Adverse Events version 3.0.
The primary endpoint of this study was the efficacy of sorafenib in patients with mRCC who have progressed after treatment with bevacizumab or sunitinib. The efficacy of sorafenib was defined as the proportion of patients who experience at least a 5% reduction in tumor burden as measured by RECIST criteria (ie, tumor burden reduction rate). Secondary endpoints included safety, duration of response, time to progression, and OS. We anticipated that the effect of sorafenib could be different in these 2 populations; therefore, 2 separate but parallel studies using a 2-stage accrual design with a maximum goal of 26 patients were implemented to test the alternative hypothesis within each prior treatment group that the tumor burden reduction rate was >20% versus <5% (α = .13, β = .10).
Time to progression and survival were calculated from the date of registration on-study to the dates of documented evidence of progression and death, respectively. Patients who were progression-free (alive) at the time of analysis were censored. The method of Kaplan and Meier was used to summarize these secondary endpoints.
Fisher exact test, the Wilcoxon rank sum test, and the log-rank test were used to compare characteristics and outcome between patients initially treated with bevacizumab and those treated with sunitinib. The Wilcoxon signed rank test was used to compare the change in tumor burden seen with sorafenib on the current trial to that seen with patients' initial anti-VEGF therapy. All statistical tests of significance were 2-tailed; and all analyses were performed using SAS version 8.0 (SAS Institute, Cary, NC).
|Factor||Prior Bevacizumab, n = 20||Prior Sunitinib, n = 27||Overall, N = 47|
|Men||13 (65%)||21 (78%)||34 (72%)|
|Women||7 (35%)||6 (22%)||13 (28%)|
|Time from diagnosis, mo|
|0||11 (55%)||13 (48%)||24 (51%)|
|1||9 (45%)||13 (48%)||22 (47%)|
|2||0||1 (4%)||1 (2%)|
|CCF risk groupa|
|Favorable||11 (55%)||10 (40%)||21 (47%)|
|Intermediate||3 (15%)||6 (24%)||9 (20%)|
|Unfavorable||6 (30%)||9 (36%)||15 (33%)|
|Prior nephrectomy||19 (95%)||25 (93%)||44 (94%)|
|Prior radiotherapy||2 (10%)||4 (15%)||6 (13%)|
|Prior therapy (other than bevacizumab/sunitinib)|
|Other antiangiogenicb||1 (5%)||5 (19%)||6 (13%)|
|IL-2 and/or interferon||9 (45%)||10 (37%)||19 (40%)|
|Otherc||2 (10%)||3 (11%)||5 (11%)|
|Any other therapy||11 (55%)||12 (44%)||23 (49%)|
|Lung||16 (84%)||15 (56%)||31 (67%)|
|Lymph nodes||13 (68%)||15 (56%)||28 (61%)|
|Liver||5 (26%)||10 (37%)||15 (33%)|
|Kidney/Nx bed||3 (16%)||8 (30%)||11 (24%)|
|Adrenal(s)||2 (11%)||6 (22%)||8 (17%)|
|Pancreas||2 (11%)||5 (19%)||7 (15%)|
|Bone||1 (5%)||3 (11%)||4 (9%)|
|Othere||5 (26%)||6 (22%)||11 (24%)|
|No. of metastatic sitesd|
|1||4 (21%)||6 (22%)||10 (22%)|
|2||7 (37%)||5 (19%)||12 (26%)|
|≥3||8 (42%)||16 (60%)||24 (52%)|
|No. cycles of prior bevacizumab/sunitinib|
|No. patients with at least some tumor shrinkage on prior bevacizumab/sunitiniba||11 (61%)||19 (70%)||30 (67%)|
|Response to prior bevacizumab/sunitinibf|
|PR||3 (19%)||5 (24%)||8 (22%)|
|SD||11 (69%)||16 (76%)||27 (73%)|
|PD||1 (6%)||0||1 (3%)|
|Not evaluable||1 (6%)||0||1 (3%)|
Between March 2006 and April 2008, 49 patients with mRCC were enrolled (Table 1). Two patients withdrew consent before initiating treatment, and thus were excluded from all analyses. Twenty patients had received prior treatment with bevacizumab, and 26 patients had received prior sunitinib. One additional patient had received both agents (bevacizumab followed by sunitinib) and was enrolled on the sunitinib cohort, as this was the patient's most recent treatment. The accrual of the bevacizumab-refractory cohort was stopped at 20 patients because of slow accrual. Overall, 72% of patients were men, median age was 64 years, and most patients (51%) were ECOG performance status 0. The majority of patients (94%) had undergone a prior nephrectomy. By using the Cleveland Clinic criteria for previously treated RCC patients,14 47% of patients were considered to have a good-risk profile, 20% intermediate-risk, and 33% poor-risk disease. Lungs and then lymph nodes were the most prevalent sites of metastases, and 52% of patients had ≥3 metastatic sites. There were no statistically significant differences in any of these characteristics between patients initially treated with bevacizumab and those treated with sunitinib.
Forty-nine percent (23 of 47) of patients received prior treatment with therapies other than bevacizumab and sunitinib (median, 1; range, 1-3). Forty percent of patients (9 of 20 bevacizumab treated and 10 of 27 sunitinib treated) had received prior cytokine therapy before receiving subsequent anti-VEGF therapy. Similarly, 13% of patients (1 of 20 bevacizumab treated and 5 of 27 sunitinib treated) had received other classes of antiangiogenic agents different from sunitinib and/or bevacizumab. These agents included volociximab (n = 2), pazopanib (n = 2), thalidomide (n = 1), and MLN518 (n = 1). The median (range) duration of prior VEGF-targeted therapy before entering this trial was 8.7 (0.4-32.3) months; for bevacizumab it was 7.7 (0.5-21.6) months, and for sunitinib it was 13.1 (0.4-32.3) months. In addition, the median (range) time from discontinuation of prior therapy to trial was 4.3 (2.0-63.7) weeks; for bevacizumab it was 4.0 (2.0-22.1) weeks, and for sunitinib it was 4.3 (2.1-63.7) weeks. Previous tumor response to bevacizumab or sunitinib was not available in 21% of patients (4 of 20 bevacizumab and 6 of 27 sunitinib). Among the patients with previous response data, 22% of patients (8 of 37) had achieved a RECIST-defined PR during their prior treatment (3 of 16 bevacizumab and 5 of 21 sunitinib), and 73% (27 of 37) had stable disease (SD) (11 of 16 bevacizumab and 16 of 21 sunitinib). The median (range) percentage change in tumor volume with previous VEGF-targeted therapy was −21% (−71% to −52%); for bevacizumab it was −17% (−71% to −52%), and for sunitinib it was −30% (−58% to −27%).
To date, all but 1 patient have discontinued treatment. Patients received a total of 253 cycles (median, 3; range, 1-26). Thirty-five (76%) patients discontinued because of progressive disease or clinical deterioration, 7 (15%) patients were taken off treatment for AEs, 2 (4%) patients withdrew consent, and 2 (4%) patients died while on-study. All patients initiated treatment with sorafenib at the standard dose of 400 mg orally twice daily (total 800 mg/d). In the absence of toxicities and at the discretion of the treating physician, patients were allowed to increase sorafenib to either 600 mg orally twice daily (total of 1200 mg/d) or 800 mg orally twice daily (total of 1600 mg/d).
Overall, 62% (28 of 45) of patients required dose modifications. Half (14 of 28) were dose reductions. Among these, 5 (18%) patients had subsequent dose re-escalation to the original sorafenib dose, and 9 (32%) had dose escalations to either 600 (n = 7) or 800 mg twice daily (n = 2). All patients who underwent a dose escalation required subsequent dose reductions secondary to grade 2 and 3 AEs within the first 2 cycles of dose escalation. (See Adverse Events section below.)
Tumor burden reduction ≥5% was observed in 14 (30%) patients. No RECIST-defined confirmed objective responses were observed. As described in Table 2, 1 (2%) patient achieved an unconfirmed PR, 20 (43%) patients had a best response of SD, 19 (40%) patients had PD, including 7 patients who developed PD before their first restaging scan, and 7 (15%) patients were not evaluable or had missing data. As seen in Figure 1, the observed tumor burden reduction for those patients achieving RECIST-defined SD ranged from 0% to 31%. Among the 14 patients who achieved a ≥5% tumor burden reduction, 8 patients had received prior sunitinib, and 6 patients had received prior bevacizumab.
|Confirmed objective responsea||0|
|Unconfirmed partial response||1 (2%)|
|Stable diseasea||20 (43%)|
|Tumor burden reduction|
|Progressive diseasea,b||19 (40%)|
There was no indication that tumor burden reduction with sorafenib was impacted by prior anti-VEGF therapy. Forty-four percent of the patients previously treated with bevacizumab had at least some tumor burden reduction with sorafenib (median, 10%; range, 1%-20%), compared with 54% of patients previously treated with sunitinib (median, 15%; range, 1%-31%) (P = .75 and P = .97, respectively).
Although 43% of our patients experienced at least some tumor burden reduction while on sorafenib treatment, the relative change in tumor burden seen was superior to the change observed with the initial anti-VEGF therapy in only 25% of patients (P < .001) (Fig. 2).
The estimated median PFS for the entire cohort was 4.4 months (95% confidence interval [CI], 3.6-5.9) (Fig. 3). Similar to tumor burden reduction, there was no indication that PFS was impacted by the type of prior anti-VEGF therapy. Estimated median PFS for patients treated with bevacizumab was 3.7 months (95% CI, 2.9-8.9), compared with 4.4 months for patients treated with sunitinib (95% CI, 2.0-7.8) (P = .69). In addition, no evidence of an improved tumor burden reduction rate or PFS was observed in patients who underwent dose escalation in this study.
Twenty-four (51%) patients had died, and the estimated median OS was 16.0 months (95% CI, 7.6-32.3).
The most commonly reported treatment-related all grade AEs were fatigue, diarrhea, anorexia, nausea, mucositis, and palmoplantar erythrodysesthesia (PPE) (Table 3). Most of these events were of mild or moderate intensity (grade 1 or 2). The most commonly reported treatment-related grade 3 AEs were PPE (31%), fatigue (18%), diarrhea (9%), and hypertension (9%). The most common laboratory abnormalities observed included anemia, observed in 13% of patients, hypothyroidism was observed in 9%, and 1 patient each reported of grade 1 to 2 hyperkalemia, hyperglycemia, hypophosphatemia, and lymphopenia.
|Adverse Event||Overall, No. (%)||Grade 1, No. (%)||Grade 2, No. (%)||Grade 3, No. (%)|
|Hypertension||16 (36)||3 (7)||9 (20)||4 (9)|
|Anorexia||18 (39)||6 (13)||10 (22)||2 (4)|
|Diarrhea||27 (60)||14 (31)||9 (20)||4 (9)|
|Mucositis||14 (31)||9 (20)||4 (9)||1 (2)|
|Nausea/vomiting||14 (31)||6 (13)||7 (16)||1 (2)|
|Fatigue||26 (58)||8 (18)||10 (22)||8 (18)|
|Skin rash||14 (31)||9 (20)||2 (4)||3 (7)|
|PPE||31 (79)||9 (20)||8 (18)||14 (31)|
|Anemia||6 (13)||—||4 (9)||2 (4)|
|Thrombocytopeniaa||2 (8)||2 (8)||—||—|
Although not statistically significant, patients who experienced grade ≥3 toxicity with prior bevacizumab or sunitinib were more likely to experience grade 3 toxicity with subsequent sorafenib (P = .23). Notably, there was significantly more PPE (P = .03) and mucositis (P = .03) in patients who had received prior bevacizumab, and more skin rash (P = .06) in patients who had received prior sunitinib. There were 2 deaths in the study; both patients received 400 mg orally twice daily, and their deaths were thought to be secondary to PD.
This study demonstrates that sorafenib has modest clinical activity in sunitinib- or bevacizumab-refractory advanced RCC patients, and supports current clinical practice patterns of sequential VEGF inhibition. Despite of the lack of RECIST-defined objective responses observed, our study did meet its primary endpoint, with a significant number of patients achieving ≥5% of tumor burden reduction, indicating persistent RCC tumor reliance on VEGF signaling after exposure to either VEGF-binding or other VEGFR inhibitors.
Administration of second-line sorafenib after either sunitinib or bevacizumab was relatively safe, and the toxicity profile observed in this study is similar to that previously reported in patients with cytokine-refractory mRCC,11 but notable for an increase in the number of patients who discontinued therapy secondary to AEs (15%) and the incidence of any grade of hypertension (36%). Di Lorenzo et al15 recently reported a similar prospective experience in 52 sunitinib-refractory patients. The overall response rate of second-line sorafenib in their series was 9.6%, and the most common AEs of second-line sorafenib were similar to those observed in our study. In our study, a significant correlation between previous bevacizumab treatment with an increased incidence of mucositis and PPE on subsequent sorafenib was seen. Likewise, a correlation between previous treatment with sunitinib and an increased incidence of skin rash was noted. Despite the small sample size, these observations raise the concern that prior bevacizumab or sunitinib exposure may influence the AE profile of sorafenib in this clinical setting.
These data add to the growing prospective data regarding the clinical effect of sequential therapy in mRCC. Two contemporary, prospective, multi-institutional studies have demonstrated antitumor activity when sunitinib or axitinib (AG-013736; Pfizer) are given sequentially to patients with either bevacizumab- or sorafenib-refractory advanced RCC.16, 17 In addition, a phase 3 trial demonstrated an improvement in PFS for everolimus (Affinitor, Novartis Pharmaceuticals) in patients with sunitinib- and/or sorafenib-refractory advanced RCC, albeit compared with an inactive control (placebo).5 The objective response rates in these studies ranged from 1% to 23%, with PFS ranging from approximately 4 months to 7 months. Although the biologic mechanism responsible for the resistance to initial VEGF-targeted therapy remains unknown, it is clear that sequential therapy with either a VEGF or an mTOR inhibitor can favorably impact clinical outcome in mRCC. Furthermore, acknowledging the limitations of comparison across small trials, the most robust antitumor effect appears to be associated with the potency of VEGFR inhibition. Existing data supporting sequential therapy in mRCC also suggest that response or lack thereof to prior VEGF-targeted therapy does not appear to predict subsequent response.
Additional phase 3 trials that include multiple correlative studies are underway to define the objective response rate and PFS of sequential approaches with novel compounds. Such data may gradually help to define an optimal sequence with further insight into specific sequences for individual patients.
This study had several limitations. First, this was a single-agent, nonrandomized study without a control arm, and as such has limitations in determining the relative benefit of sorafenib as compared with other treatment strategies in this setting. Similarly, although our study used RECIST criteria to define previous failure to either sunitinib or bevacizumab, there are inherent limitations to RECIST and radiographic assessment with targeted agents in RCC, and thus it is likely patients were variably refractory to their prior VEGF-targeted therapy.
In conclusion, sorafenib has modest antitumor activity in patients with bevacizumab- or sunitinib-refractory advanced RCC. Although mechanisms of resistance to VEGF therapy remain undefined, this prospective study supports current patterns of clinical practice and supports the hypothesis of continued reliance of RCC tumors on VEGF signaling despite previous VEGF-targeted therapy.
CONFLICT OF INTEREST DISCLOSURES
Research Grant support provided by Bayer Pharmaceuticals, West Haven, Connecticut, and Onyx Pharmaceuticals, Emeryville, California.
- 4A randomized, double-blind phase III study of pazopanib in treatment-naive and cytokine-pretreated patients with advanced renal cell carcinoma (RCC) [abstract]. J Clin Oncol. 2009; 27( 15 suppl). Abstract 5021., , , et al.
- 7Sequential use of sorafenib and sunitinib in renal cancer: retrospective analysis in 90 patients [abstract]. J Clin Oncol. 2007; 25( 18 suppl). Abstract 5038., , , et al.
- 8Evaluation of VEGF targeted therapy efficacy in mRCC after sorafenib failure or intolerance [abstract]. J Clin Oncol. 2007; 25( 18 suppl). Abstract 15517., , , et al.
- 9BAY 43-9006 exhibits broad spectrum anti-tumor activity and targets raf/MEK/ERK pathway and receptor tyrosine kinases involved in tumor progression and angiogenesis [abstract]. Proc Am Assoc Cancer Res. 2003; 44: A78., , , et al.
- 13New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst. 2000; 92: 205-216., , , et al.