Real-world clinical response and efficacy of tacrolimus-based maintenance therapy for Korean patients with lupus nephritis
Article information
Abstract
Background/Aims
We evaluated the efficacy and safety of tacrolimus as maintenance therapy in Korean patients with lupus nephritis (LN).
Methods
A total of 179 patients with biopsy-proven LN were included, of whom 92 received tacrolimus and 87 did not. Clinical parameters were assessed at six months and at one, two, three, and five years. Complete renal response (CR) and partial response (PR) were defined based on established criteria. Adverse events, renal flares, and poor outcomes have been reported.
Results
Baseline characteristics were similar, except for a higher prevalence of class V LN in the tacrolimus group. At six months, the CR rate was 49.5% in the tacrolimus group and 56.6% in the non-tacrolimus group (p = 0.308), with PR rates of 33.0% and 24.1% (p = 0.213). At one year, the non-tacrolimus group had a significantly higher CR rate (73.1% vs. 52.3%, p = 0.006), whereas the overall response rates were similar (p = 0.15). By two years, the CR rates were 71.8% in the non-tacrolimus group and 58.2% in the tacrolimus group (p = 0.031). At three years, the overall response was found 75.4% with tacrolimus and 83.1% without (p = 0.252); and at five years, these rates were 72.9% and 87.3% (p = 0.1). No significant differences in renal flares, poor outcomes, or adverse events were observed.
Conclusions
This study has demonstrated that tacrolimus is an effective and safe maintenance therapy for achieving renal response and slowing disease progression in patients with LN who have not achieved remission.
INTRODUCTION
Lupus nephritis (LN), a significant complication of systemic lupus erythematosus (SLE), is characterised by immune complex-mediated glomerulonephritis. Within five years of diagnosis, approximately 30–60% of adults and up to 70% of children develop LN. Progression to end-stage renal disease (ESRD) significantly increases mortality rates in SLE, posing a three-fold higher risk than in those without LN [1,2].
Advances in the understanding of the pathogenesis of LN have led to notable improvements in treatment strategies [3]. Although standard immunosuppressive therapy with mycophenolate mofetil (MMF) or cyclophosphamide combined with glucocorticoids remains the standard approach for LN, treatment outcomes may be inconsistent, leading to the exploration of alternative strategies [4]. Recent SLE treatment guidelines advocate the early addition of belimumab or calcineurin inhibitors (CNIs), particularly voclosporin [5]. Voclosporin, a newer CNI with a favourable metabolic profile and fewer adverse events, is not widely available in regions such as Korea, necessitating reliance on older CNIs [6]. Among these, tacrolimus is preferred for LN treatment because of its potency compared to cyclosporine and lower risk of hypertension and dyslipidaemia [7].
Although previous studies have reported the therapeutic effects of tacrolimus in LN, its ability to reduce proteinuria through non-immune mechanisms may lead to misconceptions regarding its efficacy in improving disease activity [8]. Therefore, a comprehensive evaluation of tacrolimus efficacy and outcomes is crucial to improve disease activity in LN. Recent treatment guidelines have emphasised the increasing role of tacrolimus; however, long-term data on its efficacy and safety in Korean patients are limited. Given that CNIs are primarily used for maintenance therapy, this study aimed to evaluate the efficacy and safety of tacrolimus as a maintenance treatment and to investigate treatment outcomes in Korean patients with LN.
METHODS
Study population
We retrospectively reviewed medical records from Ajou University Hospital, Suwon, Republic of Korea, from January 1999 to September 2020. Approval from the Institutional Review Board was obtained, and informed consent was waived owing to the retrospective nature of the study (AJOUIRB-DB-2022-364). Inclusion criteria comprised patients aged ≥ 18 years diagnosed with SLE based on either the 1997 American College of Rheumatology criteria or the 2012 Systemic Lupus International Collaborating Clinics classification criteria, with biopsy-proven LN [9,10]. Histological classification of LN was based on the International Society of Nephrology/Renal Pathology Society standards, with activity and chronicity indices calculated [11]. Out of 323 screened patients, 179 patients were eligible for inclusion, with 92 in the tacrolimus group and 87 in the non-tacrolimus group (Fig. 1).
Patient flow diagram. This flowchart illustrates the inclusion and exclusion criteria as well as the number of patients at each stage of the study. ICD-10, International Classification of Diseases, 10th Revision; SLE, systemic lupus erythematosus.
Patient data collection
Demographic, laboratory, and clinical data were collected. Laboratory findings included complete blood count, erythrocyte sedimentation rate, renal function (assessed by estimated glomerular filtration rate [eGFR] using the Modification of Diet in Renal Disease formula), and 24-h urine protein measurements [12]. Complement 3 (C3) and complement 4 (C4) levels were measured using Cobas (Roche Diagnostics, Basel, Switzerland), with normal ranges of C3 (90–180 mg/dL) and C4 (10–40 mg/dL). Autoimmune antibodies such as anti-nuclear antibody (ANA) and anti-double-stranded DNA antibody (anti-dsDNA Ab) levels, were quantified at the time of LN diagnosis, with anti-dsDNA Ab levels > 7 IU/mL considered abnormal, using an anti-dsDNA kit (Trinity Biotech, Bray, Ireland). Disease activity was evaluated using the SLE Disease Activity Index (SLEDAI)-2K, specifically focusing on renal involvement (rSLEDAI), ranging from 0 to 16. Parameters such as > 500 mg/day proteinuria, haematuria (> 5 red blood cells [RBC]/high-power field [HPF]), pyuria (> 5 white blood cells/HPF), and cellular casts each contributed 4 points to the rSLEDAI score [13].
Treatment regimens, including the use of hydroxychloroquine, angiotensin-converting enzyme inhibitors (ACEi), angiotensin receptor blockers (ARBs), and glucocorticoids, were documented. Immunosuppressants such as tacrolimus, MMF, cyclophosphamide, azathioprine, and mizoribine were administered for both introduction and maintenance therapy.
Clinical outcome
Clinical parameters (urine protein levels, eGFR, serum C3, C4, anti-dsDNA Ab, and rSLEDAI) were assessed at six months and one, two, three, and five years after maintenance therapy. Patients who changed their treatment regimen during maintenance therapy, such as through discontinuation or replacement of medications, were excluded from the clinical outcome analysis, thus resulting in a gradual decline in the number of patients over time. The medication maintenance rate was determined by dividing the number of patients who remained on treatment at the 5-year follow-up by the initial number of patients who started treatment.
Renal response was defined according to criteria established in the Aspreva Lupus Management Study and Lupus Nephritis Assessment with Rituximab trials [14,15]. Complete response (CR) was defined as follows: normalisation of serum creatinine (SCr) if abnormal at baseline, or SCr ≤ 15% above baseline if baseline creatinine was within the normal range, along with an inactive urine sediment (< 5 RBC/HPF and absence of RBC casts), and urinary protein to creatinine ratio (UPCR) < 0.5. Partial response (PR) was defined by SCr ≤ 15% above the baseline value, RBC/HPF ≤ 50% above baseline, and at least a 50% decrease in UPCR to < 1.0 if the baseline UPCR was ≤ 3.0, or a UPCR to ≤ 3.0 if the baseline UPCR was > 3.0. Patients not meeting CR or PR criteria were classified as non-responders.
A renal flare was defined as a ≥ 30% increase in SCr from the last recorded value, with active urine sediment (hematuria or RBC casts) or increased proteinuria (double or more increase in proteinuria if the last visit proteinuria was in the sub-nephrotic range, or at least a 2 g increase in proteinuria per 24 hours if the last visit value was in the nephrotic range). Adverse events, particularly infections, cardiac disorders, and malignancies, during LN treatment were assessed using the Medical Dictionary for Regulatory Activities (version 20.0). Additionally, ESRD or death was considered a poor outcome.
Statistical analysis
Data analysis was performed using the IBM SPSS software (version 25.0; IBM Corporation, Armonk, NY, USA). Continuous variables were tested for normality using the Kolmogorov–Smirnov test. Normally distributed data were compared using Student’s t-test, and non-normally distributed data were compared using the Mann–Whitney U-test. Categorical variables were compared using the chi-square test or Fisher’s exact test, as appropriate. Kaplan–Meier survival analysis was used to assess the clinical outcomes of tacrolimus for LN, with differences in survival curves evaluated using the log-rank test. To address potential lead time and immortal time biases, an additional analysis was conducted using the Cox proportional hazards regression model. The start of tacrolimus treatment was treated as a time-varying covariate in order to accurately reflect the time-dependent nature of tacrolimus exposure and to mitigate immortal time bias. Relevant covariates, including age, sex, disease duration, and concomitant glucocorticoid use, were included in the model to adjust for potential confounders. A p value of < 0.05 was considered statistically significant. No specific subgroup or interaction analyses were performed as the study focused on the overall comparison between the groups. Sensitivity analyses were not conducted, because the primary aim was to evaluate the overall treatment effects of tacrolimus. No missing data were present as only patients with complete datasets were included in the study, thus ensuring unbiased results.
RESULTS
Baseline characteristics of patients with LN
In this study, 179 patients with biopsy-proven LN were enrolled, with 92 in the tacrolimus group and 87 in the non-tacrolimus group. Patient demographics and characteristics are presented in Table 1. Both groups had a similar mean age (approximately 41 years), with females comprising 88.0% and 85.1% of each group, respectively. Proliferative LN (classes III and IV) were prevalent in over 60% of both groups, but class V was significantly higher in the tacrolimus group (30.4% vs. 12.6%; p = 0.004). At the time of renal biopsy, the mean rSLEDAI scores was 8.3 ± 3.6 in the tacrolimus group and 9.2 ± 3.0 in the non-tacrolimus group, with no statistical significance. The mean activity and chronicity scores were nearly identical between groups. When comparing the tacrolimus and non-tacrolimus groups, the SCr level was 0.9 ± 0.3 mg/dL and 1.1 ± 1.2 mg/dL, the eGFR was 98.6 ± 33.6 mL/min/1.73 m2 and 89.6 ± 35.0 mL/min/1.73 m2, and the 24 h-UPCR was 3.7 ± 3.9 g/d and 2.9 ± 2.1 g/d, respectively, showing no differences. No significant differences were found between the groups in terms of ANA positivity, anti-dsDNA Ab positivity, or proportion of patients with low C3 and C4 levels.
Clinical and laboratory findings and treatment regimens at the time of lupus nephritis diagnosis
Treatment details of patients with LN
Treatment regimens are listed in Table 1. Tacrolimus was administered as maintenance treatment for LN, with a mean of 32.6 ± 42.6 months of LN. Induction treatments, including MMF, cyclophosphamide, and glucocorticoids, were similar between the groups. For maintenance treatment, 79.3% of patients in the tacrolimus group received multitarget therapy, with the combination of tacrolimus and MMF being the most common, followed by tacrolimus with azathioprine, and tacrolimus with mizoribine. Conversely, monotherapy was more common in the non-tacrolimus group, with MMF in 53 patients (60.9%), azathioprine in 23 patients (26.4%), and mizoribine in 7 patients (8.0%). The cumulative glucocorticoid doses were similar between the tacrolimus and non-tacrolimus group = 26.2 g vs. 21.8 g; p = 0.101). The use of hydroxychloroquine, ARB (or ACEi), and vitamin D was comparable.
Comparison of changes in clinical parameters between patients with LN with and without tacrolimus
Clinical parameters were assessed over five years, starting with maintenance treatment (Table 2). At baseline, the tacrolimus group had significantly higher proteinuria (2.2 ± 2.4 g/day) compared to the non-tacrolimus group (1.6 ± 1.0 g/day). By six months, both groups experienced a significant reduction in proteinuria (tacrolimus: 1.0 ± 1.6 g/day; non-tacrolimus: 0.7 ± 0.8 g/day), but the difference was not significant (p = 0.054). No significant differences in proteinuria were observed at one, two, three, and five years:
Change in clinical parameters over the 5 years of tacrolimus and non-tacrolimus group in maintenance therapy for lupus nephritis
• One year: tacrolimus 0.8 ± 1.2 g/day vs. non-tacrolimus 0.5 ± 1.5 g/day; p = 0.218.
• Two years: tacrolimus 0.7 ± 0.9 g/day vs. non-tacrolimus 0.5 ± 1.0 g/day; p = 0.207.
• Three years: tacrolimus 0.7 ± 1.0 g/day vs. non-tacrolimus 0.5 ± 0.6 g/day; p = 0.057.
• Five years: tacrolimus 0.7 ± 0.7 g/day vs. non-tacrolimus 0.5 ± 0.8 g/day; p = 0.232.
Patients with eGFR < 60 showed no significant differences during the follow-up period. Anti-dsDNA Ab decreased initially but increased over time, with no significant difference after five years (tacrolimus 49.4 ± 72.8 IU/mL vs. non-tacrolimus 32.7 ± 36.0 IU/mL; p = 0.128). The serum C3 and C4 levels increased similarly. The baseline rSLEDAI scores were significantly higher in the tacrolimus group. By the sixth month, while a considerable reduction was observed, it remained significantly higher in the tacrolimus group (tacrolimus 4.4 ± 4.5 vs. non-tacrolimus 3.2 ± 3.5; p = 0.043). However, from one year onwards, the differences in rSLEDAI scores were not significant at two, three, and five years.
Comparison of renal responses in patients with LN with and without tacrolimus
No significant differences were observed in the overall (complete and partial) responses between the two groups over five years (Fig. 2A). At six months, 49.5% of the tacrolimus group and 56.6% of the non-tacrolimus group achieved CR, whereas PR rates of 33.0% and 24.1%, respectively, were not significantly different (p = 0.308 and p = 0.213, respectively). By one year, the non-tacrolimus group had a significantly higher CR rate (73.1% vs. 52.3%; p = 0.006), whereas the overall response rates were similar (p = 0.15). After two years, the CR rates were 71.8% in the non-tacrolimus group and 58.2% in the tacrolimus group (p = 0.031), with PR rates of 13.9% and 12.0% (p = 0.723), respectively, resulting in a significantly higher overall response in the non-tacrolimus. However, after three years, the overall response rates were 75.4% and 83.1% in the tacrolimus and non-tacrolimus group, respectively (p = 0.252). In addition, after five years, the overall response rates were similar (72.9% and 87.3%, respectively).
Five-year renal response and Kaplan–Meier curves of renal flare-free survival in patients with lupus nephritis. (A) Renal response to tacrolimus vs. non-tacrolimus over 5 years in patients with lupus nephritis. (B) Kaplan–Meier analysis of renal flare-free and poor outcome-free probabilities in patients with lupus nephritis. CR, complete response; PR, partial response. *p < 0.05.
Renal flares and poor outcomes in patients with LN over time
Figure 2B presents Kaplan–Meier curves showing the probability of remaining renal flare-free in patients with LN. The probability of remaining renal flare-free was similar between the tacrolimus and non-tacrolimus groups (log-rank test, p = 0.414). The incidence of ESRD and mortality rates did not differ between the two groups (Table 3), and the probability of remaining poor outcome-free showed no significant difference (log-rank test, p = 0.837).
Comparison of adverse events between tacrolimus and non-tacrolimus groups in patients with lupus nephritis
We performed Cox proportional hazards regression analysis in order to evaluate the impact of tacrolimus on flare occurrence and poor outcomes in patients with LN after adjusting for potential confounders (Fig. 3). For renal flare occurrence, significant factors included age (hazard ratio [HR] = 1.09, 95% confidence interval [CI] = 1.02–1.16; p = 0.013), tacrolimus use (HR = 0.60, 95% CI = 0.37–0.97; p = 0.045), and glucocorticoid use (HR = 3.28, 95% CI = 1.45–7.41; p = 0.004), as shown in Figure 3A. Regarding poor outcomes, significant factors were tacrolimus use (HR = 0.28, 95% CI = 0.08–0.91; p = 0.034) and glucocorticoid use (HR = 3.61, 95% CI = 1.08–12.03; p = 0.036), as shown in Figure 3B. These findings indicate that tacrolimus treatment is associated with a reduced risk of flare occurrence and poor outcomes, even after adjusting for the relevant covariates.
Factors including tacrolimus associated with clinical outcomes in patients with lupus nephritis (LN) analysed by Cox proportional hazards regression. (A) Factors including tacrolimus associated with renal flare occurrence. (B) Factors including tacrolimus associated with poor outcomes. Bold indicated p < 0.05. rSLEDAI, renal Systemic Lupus Erythematosus Disease Activity Index; dsDNA Ab, double stranded deoxyribonucleic acid antibody; HR, hazard ratio; CI, confidence interval.
Adverse events in patients with LN with and without tacrolimus
Table 3 summarises the adverse events observed in patients treated with tacrolimus compared to those treated with non-tacrolimus. The incidence of infection was similar between the two groups (21.7% vs. 23.0%; p = 0.841). The incidence of herpes zoster infection was comparable between the two groups (6.5% vs. 8.0%; p = 0.695). The rates of Mycobacteria tuberculosis (3.3% vs. 1.1%; p = 0.621), fungal infections (2.2% vs. 2.3%; p > 0.999), bacterial pneumonia (3.3% vs. 4.6%; p = 0.715), bacterial sepsis (4.3% vs. 6.9%; p = 0.527), and Clostridium difficile infections (2.2% vs. 0%; p = 0.498) were not significantly different. The malignancy rates were similar between the two groups (9.8% vs. 4.6%; p = 0.182). The incidences of cardiac disorders were comparable (4.3% vs. 4.6%; p > 0.999), including myocardial infarction (2.2% vs. 2.3%; p > 0.999), and heart failure (3.3% vs. 2.3%; p > 0.999). Ischaemic cerebrovascular disorders occurred in 2.2% of the tacrolimus group and none in the non-tacrolimus group (p = 0.498).
DISCUSSION
The 2023 EULAR guidelines emphasise the inclusion of CNIs in LN treatment, irrespective of the classification, highlighting the growing interest in these agents [4]. Our study aligns with the trend towards multitarget therapy and provides valuable insights into the long-term efficacy and safety of tacrolimus in Korean patients with LN. Although previous studies analysed the efficacy of tacrolimus in Korean patients with LN, our study is significant because we monitored the effects and safety of tacrolimus over the longest period in this population [16].
Over the 5-year maintenance period in our study, tacrolimus exhibited a favourable medication maintenance rate of 64%, similar to the findings of a large-scale Japanese study [17]. Importantly, the overall response rate after five years of maintenance therapy exceeded 70%, indicating substantial therapeutic efficacy. This response rate surpasses that reported in a multicentre Korean study focusing on the one-year efficacy of MMF and tacrolimus combination therapy in patients with LN [18]. The superior outcomes observed in our study may be attributed to the lower baseline proteinuria levels and higher proportion of pure Class V cases, which typically show better responsiveness to tacrolimus. Historically, tacrolimus has been predominantly used for membranous LN (Class V) because of its distinct immunopathogenesis, which contributes to its effectiveness in controlling nephrotic syndrome and explains the higher prevalence of Class V patients in this long-term analysis [19]. Importantly, our study included both Class V and proliferative LN (Class III and IV), highlighting the effectiveness of tacrolimus across different disease subtypes, particularly in patients with severe disease.
Multitarget therapy is commonly used for refractory or recurrent LN cases, and our findings indicate that patients initially receiving tacrolimus had higher proteinuria and rSLEDAI scores, reflecting more severe disease. Consequently, although the CR rate was initially lower in the tacrolimus group, response rates converged over time, resulting in comparable rSLEDAI scores. This improvement underscores the effectiveness of tacrolimus in stabilising disease progression, particularly in patients with more severe initial disease. These results are consistent with observations from the clinical trials of voclosporin, another CNI that has emerged as a promising LN treatment [20]. Despite the recent approval of voclosporine for active LN, head-to-head comparisons remain scarce. However, our study and meta-analysis confirmed that adding tacrolimus substantially improves LN outcomes when monotherapy with other immunosuppressants is inadequate [21].
Another advantage of tacrolimus is its demonstrated efficacy and safety in SLE pregnancies, making it a viable treatment option with relatively high efficacy in reducing the risk of ovarian failure [22]. Unlike conventional older treatment approaches, adding tacrolimus to MMF monotherapy is more advantageous than switching to cyclophosphamide when the latter is ineffective [21]. Furthermore, recent studies have shown that tacrolimus is an effective multitarget induction therapy, and reports are available demonstrating its efficacy as a monotherapy [23]. Considering these results, the role of tacrolimus in the treatment of LN has gradually expanded.
Reports suggest that tacrolimus can induce nephrotoxicity characterised by arteriolar hyalinosis, tubular atrophy, interstitial fibrosis, and glomerular sclerosis, which raises concerns about potential harm to patients with LN [24]. However, we observed no significant decline in eGFR over the five-year follow-up period, which is consistent with the findings of a 12-year observational study of tacrolimus in LN [25]. Real-world data indicate that CNI-induced nephrotoxicity occurs more frequently and severely with cyclosporine than with tacrolimus. This difference may stem from their distinct mechanisms as CNIs; tacrolimus binds to FKBP-12, while cyclosporine binds to cyclophilin. Tacrolimus’s selective suppression of T-helper cell function and reduced IL-2 production may contribute to reduced nephrotoxicity and better renal outcomes [26]. Maintaining appropriate blood tacrolimus concentrations can reduce the risk of nephrotoxicity. Therefore, while caution is warranted regarding nephrotoxicity when using tacrolimus in patients with LN, restriction in its use is not recommended based on current evidence [27].
Throughout the observation period, the rates of renal flares, progression to ESRD, and mortality did not differ significantly between the tacrolimus and non-tacrolimus groups. The five-year progression rate to ESRD was 8.7% with tacrolimus and 10% without tacrolimus, aligning closely with the outcomes reported in a decade-long Japanese study of tacrolimus in LN [28]. Our study included patients diagnosed as early as the 1990s, whereas the Japanese study focused on a cohort from the late 2000s onwards. Fortunately, tacrolimus did not increase the risk of mortality, cardiovascular events, or malignancies, which is consistent with the safety profiles observed in the LN and other South Korean autoimmune disease cohorts [29,30]. Furthermore, a detailed analysis revealed that the addition of tacrolimus did not increase infection risk, with serious infection rates approximating those reported in broader SLE cohorts [31].
Our study, which was one of the longest in Korea, has provided substantial evidence regarding the efficacy and safety of tacrolimus in treating LN, including its use as a monotherapy or in combination with other immunosuppressants, in real-world clinical practice. However, the predominance of Asian populations in studies on tacrolimus highlights the need for more diverse ethnic and geographic analyses. This study has several limitations. It was a single-centre, retrospective study, which may introduce selection bias related to tacrolimus-associated nephrotoxicity. The timing of tacrolimus initiation varied, with some patients starting after remission and a subsequent flare leading to different treatment durations. We could not standardise the tacrolimus dosages or account for changes during treatment, and blood tacrolimus concentrations were not measured, limiting our ability to assess patient compliance. Because the use of tacrolimus is primarily part of a multitarget therapy with MMF, this limits our ability to clearly evaluate the effects of tacrolimus when used alone. Additionally, the baseline differences between the tacrolimus and other treatment groups, such as higher proteinuria and more frequent class V disease before induction, may have affected the outcomes. Future studies using methods such as propensity score matching could therefore help better assess the impact of tacrolimus on treatment outcomes. Finally, the limited number of rebiopsies in patients with flares precluded a detailed analysis of the effects of tacrolimus on renal histopathology, highlighting the need for future studies involving histological evaluations.
In conclusion, this study has confirmed that tacrolimus is an effective and safe option for maintaining LN remission. Specifically, for patients with persistent proteinuria after induction therapy, the addition of tacrolimus to other immunosuppressive agents yielded outcomes comparable to those of patients who responded well to induction treatment alone. Notably, tacrolimus did not increase the risk of infection associated with immunosuppressive therapy. Therefore, selecting tacrolimus as maintenance treatment may be a highly effective strategy for managing disease activity and preventing renal flares in patients who do not adequately respond to conventional induction therapy or experience flares during maintenance therapy.
KEY MESSAGE
1. For patients with persistent proteinuria after induction therapy, the addition of tacrolimus to other immunosuppressive agents yielded outcomes comparable to those of patients who responded well to induction treatment alone.
2. The rates of adverse events, including renal flares and serious complications, were similar between tacrolimus-based and non-tacrolimus treatments.
3. These findings support the use of tacrolimus as a valuable option in personalised maintenance strategies for lupus nephritis.
Notes
CRedit authorship contributions
Ji-Won Kim: methodology, investigation, data curation, formal analysis, writing - original draft, writing - review & editing; Ju-Yang Jung: investigation, formal analysis, writing - review & editing; Hyoun-Ah Kim: investigation, formal analysis, writing - review & editing; Chang-Hee Suh: conceptualization, writing - review & editing, supervision, funding acquisition
Conflicts of interest
The authors disclose no conflicts.
Funding
This research was supported by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HR16C0001).