Article Text
PDF
Abstract
Objectives To report the 10-year outcome of lupus nephritis (LN) treated with mycophenolate mofetil (MMF) or tacrolimus (TAC) induction in a randomised controlled trial.
Methods Patients with active LN were treated with MMF or TAC combined with high-dose prednisolone. Responders were switched to azathioprine (AZA) at month 6. Clinical outcomes at 10 years (renal flares, renal function decline and mortality) were assessed. Factors affecting prognosis were studied by Cox regression. Urine protein-to-creatinine ratio (uPCr) and estimated glomerular filtration rate (eGFR) at different time points were evaluated for their prediction of a poor prognosis by receiver operating characteristic (ROC) analysis.
Results 150 patients were studied (age 35.5±12.8 years). Complete renal response rate was similar between MMF (59%) and TAC-treated patients (62%; p=0.71). AZA maintenance was given to 79% patients. After 118.2±42 months, proteinuric and nephritic renal flares occurred in 34% and 37% of the MMF, and 53% and 30% of the TAC groups of patients, respectively (p=0.49). The cumulative incidence of a composite outcome of ↓eGFR ≥30%, chronic kidney disease stage 4/5 or death at 10 years was 33% in both groups (p=0.90). Factors independently associated with a poor renal prognosis were first-time LN (HR 0.12 (0.031 to 0.39); p=0.01), eGFR (HR 0.98 (0.96 to 0.99); p=0.008) and no response at month 6 (HR 5.18 (1.40 to 19.1); p=0.01). ROC analysis revealed an uPCr >0.75 and eGFR of <80 mL/min at month 18 best predicted a poor renal prognosis.
Conclusions Long-term data confirmed non-inferiority of TAC to MMF as induction therapy of LN. An uPCr≤0.75 and eGFR of ≥80 mL/min at month 18 best predicted a favourable 10-year outcome and may be suitable targets for induction/consolidation therapy.
Trial registration number NCT00371319.
- systemic lupus erythematosus
- lupus nephritis
- outcomes research
- treatment
Statistics from Altmetric.com
Key messages
What is already known about this subject?
Previous randomised controlled trials (RCTs) of lupus nephritis (LN) have shown that the difference in renal function preservation of various treatment regimens would only be apparent beyond 5 years. Long-term data on the efficacy of the calcineurin inhibitors as induction therapy of LN are scant.
What does this study add?
This study reports the 10-year outcomes of a large cohort of patients with LN treated with either mycophenolate mofetil (MMF) or tacrolimus induction, followed by azathioprine maintenance. There were no significant differences in renal flares or renal function decline between the two regimens. Receiver operating characteristic analysis showed a urine protein-to-creatinine ratio (uPCr) ≤0.75 and estimateglomerular filtration rate (eGFR) ≥80 mL/min at month 18 best predicted a favourable outcome. Maintenance therapy of <62.5 months’ duration best predicted a renal flare.
How might this impact on clinical practice?
Tacrolimus may be considered as an alternative to MMF as induction therapy of LN. Treatment of LN should aim at uPCr ≤0.75 and eGFR ≥80ml/min by month 18, with repeat renal biopsy if necessary. Maintenance therapy should best be continued for 5 years or more, particularly in high-risk patients.
Introduction
Renal involvement is one of the most frequent and serious complications of systemic lupus erythematosus (SLE), particularly in certain ethnic groups such as the African-Americans, Hispanic-Americans and the Asians.1–3 Lupus nephritis (LN) is associated with significant morbidity and mortality because of the risk of progression to end-stage renal disease (ESRD) and treatment-related adverse effects.4 5 Up to one-third of patients with severe LN develop ESRD6–8 at 10 years, and mortality rate was increased by eightfold compared with the general population.5 Thus, the goals of LN therapy are to control kidney inflammation timely (induction/consolidation) and reduce the risk of renal flares (maintenance).
Conventional induction therapy of severe LN includes high-dose glucocorticoids in combination with cyclophosphamide (CYC; National Institute of Health (NIH) high-dose or Euro-Lupus low-dose regimen) or mycophenolate mofetil (MMF).9–11 MMF emerges to become first-line induction therapy of LN because of the lack of ovarian and bladder toxicity.12 However, MMF is not superior to CYC in short-term efficacy of LN.13 14 The calcineurin inhibitors (CNIs), such as ciclosporin A (CSA) and tacrolimus (TAC), have been used and approved in some Asian countries for induction therapy of LN.15 Small randomised controlled trials (RCTs) have shown similar efficacy of TAC with intravenous pulse CYC for inducing remission of LN.15 16 On the other hand, low-dose combination of MMF and TAC has been shown to be more effective than intravenous pulse CYC as induction therapy of LN.17 18
In a recent RCT, we compared the efficacy of MMF and TAC, combined with high-dose prednisolone, as induction therapy in patients with active LN.19 TAC was found to be non-inferior to MMF in terms of the complete renal response (CR) rate at 6 months. We hereby report the outcome of the same cohort of patients at 10 years. The incidence of renal flares, renal function decline (reduction in estimate glomerular filtration rate (eGFR)), chronic kidney disease (CKD) progression and mortality would be compared between the two treatment arms.
Patients and methods
Study population
The inclusion criteria of our RCT19 were: (1) age≥18 years; (2) fulfilment of ≥4 American College of Rheumatology (ACR) criteria for SLE20; (3) biopsy proven active LN (International Society of Nephrology/Renal Pathology Society (ISN/RPS) class III/IV/V) within 4 weeks of study entry; (4) serum creatinine (SCr)<200 μmol/L. Exclusion criteria were: (1) refusal for randomisation and preferred conventional regimens such as CYC; (2) SCr≥200 μmol/L; and (3) patients who planned for pregnancy within 12 months after randomisation.
All participants were prescribed oral prednisone (0.6 mg/kg/day for 6 weeks, then tapered by 5 mg/day every week to <10 mg/day) which was continued indefinitely as maintenance therapy. Patients were randomised by block in a 1:1 ratio to either (1) MMF (2 g/day initially, augmented to up to 3 g/day if clinical response was suboptimal at month 3), in two divided doses for 6 months; or (2) TAC for 6 months (initial dosage 0.1 mg/kg/day in two divided doses, reduced to 0.06 mg/kg/day if clinical response was satisfactory at month 3). CR or good partial renal responders (PR) were switched to azathioprine (AZA; 2 mg/kg/day) for maintenance therapy. AZA was continued indefinitely until patients experienced serious adverse events or disease flares in which reinduction and subsequent switching to other agents for maintenance was decided by attending physicians. Details of the study protocol can be referred to our previous publication19 and online supplementary table 1 shows that definitions of renal response at 6 months. Written consent was obtained from all patients.
Supplemental material
Long-term clinical outcomes
Patients participated in this RCT were followed longitudinally by the same group of physicians. We studied the clinical outcomes at 10 years of all the recruited patients on an intention-to-treat basis regardless of the maintenance regimens following induction or reinduction therapies. These outcomes included the incidence of renal flares (proteinuria, nephritic), non-renal flares, renal function decline (drop in eGFR by ≥30%), progression to CKD stage 4/5 (eGFR<30 mL/min) and mortality, which were compared between the MMF and TAC induction arms. Factors associated with a poor renal prognosis were studied by statistical analysis. Renal parameters, namely urine protein-to-creatinine ratio (uPCr) and eGFR, at different time points between 6 and 24 months were studied for their predictive value of renal prognosis at 10 years.
A proteinuric renal flare was defined as an increase in proteinuria to >2 g/day (or uPCr>2.0) after a CR, or doubling of proteinuria (or uPCr) after a PR with or without an increase in SCr (<30%). An increase or recurrence of active urinary sediments (red blood cell or cellular casts) with a concomitant increase in proteinuria (or uPCr) or increase in SCr (≥30%; after exclusion of other causes) would qualify a nephritic renal flare. An ex-renal lupus flare was defined as any non-renal clinical manifestations that warranted augmentation of therapies. Organ damage of SLE was assessed by the SLE International Collaborative Clinics/ACR damage index (SDI).21 Damage (after SLE diagnosis) must be present for 6 months before it is scored, irrespective of the cause, and must be irreversible.
Patient and public involvement
Patients were not involved in the design and conduct of this trial.
Statistical analyses
Unless otherwise stated, values in this study were expressed as mean±SD. Comparison between two groups was performed by the independent Student’s t-test for continuous variables and χ2 test for categorical variables (Fisher’s exact test was used when the frequency of any cell was <5). The cumulative incidence of renal flares, renal function deterioration or mortality over time was evaluated by Kaplan-Meier’s analysis. Difference between the MMF and TAC group was compared by the log rank test. For deceased patients or those who were lost to follow-up, data were censored at their last visits. Factors associated with renal flare or long-term renal prognosis were studied by Cox regression taking into account of all contributing covariates without selection. Covariates that were considered in the regression models included age, sex, first-time LN (vs relapsed LN), treatment arms (MMF vs TAC), histological class, activity and chronicity scores, nephrotic syndrome and hypertension at baseline, serological and urinary parameters at 6 months, renal response, maintenance therapy and renal flares. Receiver operating characteristic (ROC) curve analysis was used to explore the cut-offs of uPCr or eGFR at various time points that best predicted the long-term renal prognosis based on area under the curve (AUC) and the shape of the curves that determined the turning points that yielded the maximum sum of the sensitivity and specificity values.
Statistical significance was defined as a p value of <0.05, two tailed. All statistical analyses were performed by using the SPSS program (V.21.0) for Windows 10.
Results
Clinical characteristics of participants
A total of 150 patients with active LN were randomised to receive MMF (n=76) or TAC (n=74) between 2005 and 2012. There were 138 (92%) women and the distribution of the histological classes (ISN/RPS) was: III±V 36%; IVG/S±V 46%; pure V 19%. The mean age at study entry was 35.5±12.8 years and the mean histological activity and chronicity score were 8.2±3.4 and 2.6±1.6, respectively. Fifty-nine (39%) patients were hypertensive, 62 (41%) had active urinary casts, 112 (75%) had microscopic haematuria and 67% patients had an eGFR<90 mL/min. No significant differences in the clinical characteristics of the patients were observed between the two treatment arms. As reported, the rate of CR (primary end point) was 59% in the MMF and 62% in the TAC group at month 6 (difference 3% (−12% to 18%); p=0.71).19
Renal flares over time
Maintenance therapy with AZA was given to 59 (78%) MMF-treated (dose 82.5±24 mg/day) and 60 (81%) TAC-treated patients (dose 86.5±21 mg/day; p=0.32). AZA was discontinued in four patients because of intolerance (agranulocytosis (n=2), disseminated tuberculosis (n=1) and gastrointestinal upset (n=1)) and three patients because of pregnancy. Six of these patients were switched to low-dose cyclosporine A for maintenance. Patients who had no response at 6 months were reinduced with alternative regimens that included high-dose prednisolone with CYC (n=20), low-dose combination of MMF and TAC (n=5), cross-over to TAC (n=4) or MMF (n=2). After successful reinduction, the choice of maintenance regimen was at the discretion of the attending physicians.
No patients were lost to follow-up. After a mean of 118.2±42 months, proteinuric and nephritic renal flares occurred in 34% and 37% of patients treated initially with MMF, and 53% (group difference; p=0.02) and 30% (group difference; p=0.36), respectively, in those treated with TAC. There was a total of 77 renal flares in 43 (57%) patients treated with MMF (0.11/patient year) and 92 renal flares in 46 (62%) of patients treated with TAC (0.12/patient year; p=0.23). The retreatment regimens for these flares were summarised in online supplementary table 2. The daily dose of prednisolone used ranged from 0.4 to 0.8 mg/kg (for 4–8 weeks before tapering). Numerically, more patients in the MMF group were retreated with CYC-based regimens for renal flares.
The cumulative risk of having a renal flare of patients treated with MMF/AZA was 28% at 3 years, 42% at 5 years and 58% at 10 years, whereas the corresponding figures for patients treated with TAC/AZA was 32% at 3 years, 53% in 5 years and 66% in 10 years (log rank test; p=0.43) (figure 1). For those who achieved CR after induction therapy at month 6, the mean time to first renal flare was 70.4±47.1 months (median 62 months) in the MMF group and 65.2±50 months (median 47 months) in the TAC group (p=0.61). ROC analysis showed that a duration of maintenance therapy <62.5 months best predicted the first renal flare (AUC 0.80 (0.73–0.88); specificity 0.77; sensitivity 0.77; figure 2).
Cumulative incidence of renal flares (any) over time. AZA, azathioprine; MMF, mycophenolate mofetil; TAC, tacrolimus.
Receiver operating characteristic (ROC) analysis of the length of maintenance therapy and renal flares (AUC=area under the curve).
Cox regression analysis showed that first time LN (HR 0.41 (0.22 to 0.77); p=0.006) and persistent elevation of anti-dsDNA level at 6 months (HR 0.45 (0.25 to 0.83); p=0.01) were independently associated with renal flares over time (table 1). Male patients showed a trend of having more flares but this did not reach statistical significance (HR 4.67 (0.96 to 22.7); p=0.06).
The cumulative rates of renal flares in subgroup of patients with pure membranous LN (N=28) and those who did not require re-induction therapy at 6 months (N=119) are shown in online supplementary figures 1 and 3, respectively. There were no significant differences between the TAC and MMF groups.
Supplemental material
Supplemental material
Factors independently associated with renal flares (Cox regression)
Repeat renal biopsy
Renal biopsy was repeated in 40 patients mainly because of renal flares. Data from four patients could not be used because of an inadequate number of glomeruli. In the remaining patients (MMF n=18; TAC n=18), the time interval from the initial biopsy was 78.2±36 and 67.8±40 months, respectively, in the MMF and TAC group (p=0.40). Five patients in the MMF and four patients in the TAC group had further exposure to the CNIs for reinduction therapy of refractory renal disease or SLE flares before repeat renal biopsy. The increase in histological chronicity score was not significantly different between the two groups (1.29; 95% CI (−0.21 to 2.78) in MMF and 1.64; 95% CI (0.43 to 2.85) in TAC; p=0.69).
Non-renal flares and organ damage over time
In the MMF group of patients, non-renal lupus flares occurred at a rate of 0.083 per patient/year, which was not significantly different from that of the TAC group (0.091 per patient year; p=0.64). The incidence of mucocutaneous, arthritic, serosal, haematological and neuropsychiatric SLE flares was not significantly different between the MMF and TAC groups of patients (p values non-significant for all comparisons; figure 3). Treatment regimens of these flares were summarised in online supplementary table 2. The daily dose of prednisolone used ranged from 0.2 to 0.4 mg/kg (for 2–4 weeks) when there was no concomitant renal or major organ flare.
Rate of extrarenal flares according to treatment groups (per patient year). MMF, mycophenolate mofetil; TAC, tacrolimus.
The proportion of patients who accrued new organ damage over time was not significantly different between the MMF and TAC groups (45% vs 47%; p=0.75). The increase in SDI score from baseline was 1.0±1.4 in TAC and 0.79±1.1 in MMF groups of patients (p=0.31). The increase in SDI score was significantly higher in those who failed to remit at 18 months than the others (1.6±1.5 vs 0.7±1.1; p=0.009), which was mainly contributed by renal and neuropsychiatric scores (data not shown). Diabetes mellitus developed in 4 (5.4%) TAC and 2 (2.6%) MMF-treated patients (p=0.44). The proportion of patients who were treated with antihypertensive agents was non-significantly higher in the TAC than MMF group (51% vs 37%; p=0.07).
Renal function deterioration and prognostic factors
In patients treated with MMF, the proportions with eGFR decline by ≥30% compared with baseline, CKD stage 4/5 progression and mortality were 21%, 18% and 14%, respectively. These figures were not significantly different from those treated with TAC (corresponding figures 24%, 19% and 12%; p values of 0.63, 0.94 and 0.68, respectively, compared between groups). Figure 4 shows that cumulative incidence of a composite outcome of decline of eGFR by ≥30%, CKD stage 4/5 progression or death at 5 and 10 years was 24% and 33% in patients treated with MMF, and 17% and 33%, respectively, in those treated with TAC (p=0.90). Factors significantly associated with this composite outcome were first time LN (HR 0.12 (0.03 to 0.39); p=0.01), eGFR (HR 0.98 (0.96 to 0.99); p=0.008) and no renal response at month 6 (HR 5.18 (1.40 to 19.1); p=0.01; table 2).
Cumulative risk of a poor renal outcome (↓estimated glomerular filtration rate ≥30%, chronic kidney disease 4/5 or death). AZA, azathioprine; MMF, mycophenolate mofetil; TAC, tacrolimus.
Factors independently associated with a poor clinical outcome (≥30% drop in estimated glomerular filtration rate (eGFR), chronic kidney disease stage 4/5 development or death; Cox regression)
Exploratory ROC analysis demonstrated that an eGFR cut-off of 80 mL/min (AUC 0.70; sensitivity 0.64, specificity 0.66) and uPCr cut-off of 0.75 (AUC 0.73; sensitivity 0.69, specificity 0.74) at month 18 best predicted CKD stage 4/5 or decline of eGFR by ≥30% (online supplementary table 3; figure 5).
The long-term outcome of patients with pure membranous LN and those who did not require reinduction therapy at 6 months is shown in online supplementary figures 2 and 4, respectively. No significant differences were observed between the TAC and MMF groups.
Supplemental material
Supplemental material
Receiver operating characteristic (ROC) analysis of the values of (A) proteinuria and (B) estimated glomerular filtration rate (eGFR) in predicting renal prognosis (AUC=area under the curve). uPCr, urine protein-to-creatinine ratio.
Figure 6 shows the cumulative incidence of the same composite outcome according to the status of renal response at month 18 in all patients studied, regardless of the treatment arms. The incidence of a poor renal outcome at 10 years was significantly higher in those who did not respond to treatment at month 18 (NR) compared with other patients with PR or CR (HR 4.94 (2.72 to 8.97); p<0.001).
Cumulative risk of a poor renal outcome according to renal response at 18 months.
Discussion
Given the negative impact of renal disease on morbidity, mortality and quality of life in patients with SLE, there are unmet needs in the development of novel therapies of LN. The biological agents have been explored in the treatment of LN but most recent clinical trials ended up with unfavourable results. In a pivotal RCT (Lupus Nephritis Assessment with Rituximab (LUNAR)), rituximab, a chimeric monoclonal antibody that directs against anti-CD20, was not demonstrated to be superior to placebo when added to a regimen of high-dose corticosteroids and MMF in treating proliferative LN.22 Belimumab, a fully humanised monoclonal antibody directing against B-cell activation factor, is licensed for childhood and adult active SLE that fails standard of care.23 However, patients with severe neuropsychiatric manifestations and renal disease (proteinuria ≥6 g/day or SCr >2.5 mg/dL) were excluded in its pivotal trials.24 25 A post hoc analysis on patients with milder renal disease in the BLISS trials showed numerically more patients treated with belimumab had improvement of proteinuria and renal remission.26 As interpretation is confounded by the small sample size, belimumab is not currently indicated for severe LN. Preliminary results from the phase III placebo-controlled RCT (BLISS-LN) are promising and the indication of belimumab in LN will be revisited.23
The CNIs block T-cell activity through the inhibition of the calcium/calmodulin-dependent phosphatase calcineurin.15 CSA and TAC bind to cyclophilin and FKBP12, respectively, after cellular entry, suppress the activity of calcineurin and nuclear translocation of transcription factors such as nuclear factor of activated T cells (NF-AT) that are involved in IL-2 gene transcription.27 As a result, T-cell activation is impaired and the production of cytokines is attenuated.28 In addition, the CNIs exhibit antiproteinuric effects by stabilising the podocytes in the kidneys.29 30
Small RCTs have shown that CSA or TAC is as effective as CYC for induction therapy of LN.31–33 Combining the CNIs, such as TAC or voclosporin (a chemical analogue of CSA) with MMF, has been shown to enhance the renal response rate for induction therapy of LN.34 35 A low-dose regimen consisting of MMF and TAC has also been shown to ameliorate proteinuria in up to two-third of patients with refractory LN.36 Our RCT was the first which was powered to test the efficacy of TAC against MMF as induction therapy of LN.19 We demonstrated that TAC was non-inferior to MMF, when combined with prednisolone, in achieving CR after 6 months post therapy. The current work reports the 10-year outcome of the same cohort of patients and we showed that the sequential regimen TAC-AZA remained non-inferior to the MMF-AZA in terms of renal flares and renal function deterioration.
One concern of the use of CNIs in LN is a high rate of renal flare after cessation of the drugs. In the NIH RCT for membranous LN, although CSA was more effective than prednisolone alone at month 12, the cumulative risk of renal flare was 60% at 48 months after drug discontinuation.37 In the treatment protocol adopted by our RCT, instead of discontinuing immunosuppression, we switched all responders to AZA for maintenance. We could not show any difference in the overall rate of renal flares between the MMF and TAC groups, although the subtype proteinuric renal flare was more frequent with TAC induction. As most proteinuric flares did not occur shortly after TAC discontinuation, reversal of the podocyte stabilising or haemodynamic effects of the drug could not be a good explanation. The similar preservation of renal function between the two groups at 10 years suggests that retreatment of renal flares in both groups was equally effective.
The optimal duration of maintenance therapy is not certain. A longer period of maintenance should be considered in high-risk patients and those with residual renal activity.38 A study on 32 patients with proliferative LN who were discontinued immunosuppression showed that 17 patients developed renal flares over a median of 34 months.39 Those who remained in remission had received a longer period of immunosuppression. In our experience of patients with diffuse LN, the lack of maintenance therapy for ≥3 years was independently associated with an outcome of SCr doubling, ESRD or death (HR4.62; p=0.02).7 However, the cut-off of ‘3 years’ was arbitrary to facilitate statistical analysis in this study. In the present long-term cohort, maintenance therapy of <62.5 months best predicted renal flares by ROC analysis. This provides more scientific evidence on a suitable duration of maintenance therapy in LN.
Another drawback of long-term CNI use is nephrotoxicity. In our study, repeat renal biopsy in selected patients with renal flares did not show any difference in the accrual of histological chronicity scores over time between MMF and TAC treated patients. However, a solid conclusion could not be drawn for the small sample size and relatively short period of TAC induction treatment (6 months). Our results are similar to a previous RCT comparing CSA with AZA as maintenance treatment after successful CYC induction in that the increase in histological chronicity score on repeat renal biopsy (n=29) was not significantly different from between the CSA and AZA groups at year 2.40 Further studies with protocol-based repeat renal biopsy and a longer period of TAC maintenance therapy are necessary to confirm the long-term safety of the CNIs in LN.
There is still no consensus on the definition of renal response to LN therapies, although this may have impact on the success of treatment trials.41 Most clinical trials of induction therapy of LN adopted a proteinuria value of <0.5 g/day and stabilisation of renal function as the efficacy end point. However, the time to renal response is variable and may be delayed in patients with LN with membranous histology. Using ROC analysis, we demonstrated in our study that a uPCr≤0.75 g/day and eGFR≥80 mL/min at month 18 best predicted the long-term renal outcome. These clinical parameters might be considered as appropriate targets of induction/consolidation therapy of LN. Our results are in line with the Euro-Lupus CYC and MAINTAIN trials42 43 in which proteinuria values of <0.8 g/day and <0.7 g/day, respectively, at 12 months post therapy were found to best predict the long-term renal prognosis.
Our data were derived from Chinese patients and should not be extrapolated to other ethnic groups. Although TAC/AZA was non-inferior to MMF/AZA, the overall rates of renal flares and CKD development were relatively high. This underscores the unmet need for better therapies in LN. Our study was first designed 16 years ago and switching to AZA after TAC or MMF was protocol based. Should we choose MMF for switching, the long-term outcome might improve because MMF was shown to reduce renal flares more effectively than AZA in the ALMS study published subsequently.44 Belimumab is a promising agent in reducing renal flares and enhancing renal response in LN when added to the standard of care. Results of ongoing studies on novel biological and targeted agents in LN are eagerly awaited.
In summary, 10-year data of our RCT showed that TAC remained non-inferior to MMF as induction therapy of LN in terms of renal flares and renal function decline. Relapsed LN, lower eGFR and refractory disease post induction therapy were associated with a poorer outcome. A uPCr≤0.75 and eGFR of≥80 mL/min at 18 months best predicted the 10-year outcome and may be a suitable target for induction/consolidation therapy. Maintenance therapy for at least 5 years may be necessary to maximise the reduction in renal flares.
References
Footnotes
Handling editor Josef S Smolen
Contributors CCM: study design, patients’ assessment, data collection and analysis. LYH, SKY, MCL, CHT and WLN: patients’ assessment and follow-up, data collection.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Patient consent for publication Not required.
Ethics approval The study protocol was approved by the Research & Ethics Committee of Tuen Mun Hospital, which consists of lay persons from the public as the core members.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information.