Effectiveness of lipid-lowering therapy on mortality and major adverse cardiovascular event outcomes in patients undergoing percutaneous coronary intervention: a network meta-analysis of randomised controlled trials

Abstract

Background

Emergency percutaneous coronary intervention (PCI) can quickly restore myocardial perfusion after acute coronary syndrome. Whether and which lipid-lowering regimens are effective in reducing major adverse cardiovascular events (MACEs) and mortality risk after PCI remain unclear.

Objective

This study assessed the benefits of different lipid-lowering regimens on the risk of MACEs and mortality in the post-PCI population by network meta-analysis.

Methods

Public databases, including PubMed, Embase and the Cochrane Library, were searched from inception to August 2022. Randomised controlled trials (RCTs) on lipid-lowering regimens in post-PCI populations were included and analysed. The outcomes were the incidence of all-cause mortality and MACEs, whether reported as dichotomous variables or as HRs.

Results

Thirty-nine RCTs were included. For MACEs, alirocumab plus rosuvastatin (OR: 0.18; 95% CI: 0.07 to 0.44), evolocumab plus ezetimibe and statins (OR: 0.19; 95% CI: 0.06 to 0.59), eicosapentaenoic acid (EPA) plus pitavastatin (HR: 0.67; 95% CI: 0.49 to 0.96) and icosapent ethyl plus statins (HR: 0.73; 95% CI: 0.62 to 0.86) had significant advantages and relatively high rankings. For mortality, rosuvastatin (OR: 0.30; 95% CI: 0.11 to 0.84), ezetimibe plus statins (OR: 0.55; 95% CI: 0.43 to 0.89) and icosapent ethyl plus statins (OR: 0.66; 95% CI: 0.45 to 0.96) had significant advantages compared with the control.

Conclusion

EPA, especially icosapent ethyl, plus statins had a beneficial effect on reducing the risk of MACEs and mortality in post-PCI patients. Proprotein convertase subtilisin/kexin type-9 inhibitors plus statins were able to reduce the risk of MACEs, but the risk of mortality remained unclear.

PROSPERO registration number

CRD42018099600.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Only randomised controlled trials with high overall design quality were considered for inclusion.

• Major adverse cardiovascular event (MACE) and mortality were adopted as outcomes with little influence from subjective factors. This meta-analysis was based on the study level instead of the individual level.

• The criteria for defining MACEs varied among studies.

• Many included studies only reported dichotomous outcomes but did not report the HR results.

Introduction

Acute coronary syndrome (ACS) is a term used to refer to a range of conditions associated with acute myocardial ischaemia and/or infarction, which are usually due to coronary artery occlusion and acute ischaemic necrosis of the myocardium due to the progression of coronary atherosclerotic lesions.^{1 2} Emergency percutaneous coronary intervention (PCI) can quickly restore myocardial perfusion.³ Although the development of technological and procedural PCI has resulted in substantial improvements in clinical outcomes, recurrent coronary events may still occur after PCI.⁴

The view of ‘residual cardiovascular risk’ was introduced because major adverse cardiovascular events (MACEs) still occur in some patients who underwent PCI during follow-up. PCI can treat focal manifestations of systemic progressive disease, but the residual risk of ACS is largely related to the systemic proatherosclerotic effect of poorly controlled cardiovascular risk factors.⁴ Lowering lipid levels, especially Low density lipoprotein - cholesterol (LDL-C), can halt the progression of coronary atherosclerosis and improve cardiovascular outcomes. Based on this view, it is believed that long-term optimal lipid-lowering therapy is effective in reducing long-term cardiovascular events after PCI. However, this view was still subject to challenges.

Based on data from the ‘Korea Acute Myocardial Infarction Registry’, the proponents concluded that patients treated with statins had significantly lower rates of MACEs, all-cause death and cardiac death during the 2-year follow-up period after PCI application.⁵ However, a study of postoperative follow-up of patients with PCI enrolled in the Melbourne Interventional Group registry concluded that statins have no significant beneficial effect on MACEs after PCI.⁶ The controversy may be explained by two concepts: on the one hand, the optimal lipid reduction target may not be achieved by using single statins.^{7 8} On the other hand, long-term high-dose application of statins increases the risk of intracerebral haemorrhage and other side effects.^{9 10}

There is a consensus on preloading high-dose statins to reduce MACEs in the perioperative period with PCI.^{11 12} However, there is still insufficient evidence for the continued application of lipid-lowering drugs to reduce the risk of long-term MACEs and mortality. This study assessed the benefits of different lipid-lowering regimens on the risk of MACEs and mortality in the post-PCI population by network meta-analysis (NMA).

Methods

This study was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. The study was registered with PROSPERO.

Patient and public involvement

None.

Search strategy

Public literature databases, including PubMed, Embase and the Cochrane Library, were searched from inception to August 2022 without language restrictions using the following search terms: (lipid-lowering or statin or simvastatin or rosuvastatin or atorvastatin or fluvastatin or lovastatin or pravastatin or pitavastatin or mevastatin or ezetimibe or “eicosapentaenoic acid” or “icosapent ethyl” or “bempedoic acid” or fibrate or bezafibrate or gemfibrozil or fenofibrate or ciprofibrate or evolocumab or alirocumab or evinacumab or volanesorsen or vupanorsen or pelacarsen or olezarsen or inclisiran or olpasiran) and (“percutaneous coronary intervention” or “coronary angioplasty”) and (random* or randomized or randomized). The details of the full search strategy are listed in the online supplemental file 1. The references of relevant systematic reviews and meta-analyses were also searched to avoid omissions. The two authors conducted literature retrieval independently, and any conflicts were resolved through discussion with the third author.

Inclusion and exclusion criteria

The literature was included if it met the following criteria: (1) the study adopted a randomised controlled study design; (2) the study included patients who underwent PCI surgery or reported the subgroup of the population that underwent PCI; (3) the lipid-lowering regimen was applied to the population of the intervention group; (4) the control group used a different lipid-lowering agent or regimen and (5) the study reported the outcome of mortality and/or MACEs. The exclusion criteria were as follows: (1) as preloading of statins before PCI was shown to have clear benefits, to determine whether application of lipid-lowering drugs after PCI also had beneficial effects, this work excluded studies on the preloading application of lipid-lowering drugs before PCI; and (2) although high-dose lipid-lowering agents, such as statins, have a better lipid-lowering effect, long-term application may bring potential side effects.^{9 13} Therefore, only studies in which all agents were considered to be applied at reasonable doses were included, and dose–response studies were excluded. In addition, repeatedly published studies, protocols, conference abstracts, reviews, comments and editorials were also excluded.

Data extraction and quality assessment

Two authors independently extracted the information from the included studies. The contents include the name of the first author, publication year, study location, sample size (population that underwent PCI), study abbreviation and registration number, lipid-lowering intervention and control and follow-up time.

The outcomes analysed were the incidence of all-cause mortality and MACEs, whether reported as dichotomous or HR statistics based on Cox regression. The MACE outcome was selected to most closely approximate the composite endpoint, including mortality, Myocardial infarction (MI), stroke, coronary revascularisation and restenosis. Study quality was assessed by two investigators using the Cochrane risk of bias assessment tool, which included random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other potential biases.

Statistical analysis

We conducted a frequentist NMA using random-effects models weighted by the inverse variance method. ORs and 95% CIs were used for dichotomous outcomes. The HRs and 95% CIs based on Cox regression results were also pooled for reporting. If the HR value was not reported but there was a Kaplan-Meier survival curve, the HR value was extracted from the curve by GetData Graph Digitizer software V.2.24.

In network plots, the direct comparisons among treatment arms are shown, the end of each line indicates a treatment arm, and the thickness of the lines indicates the number of studies comparing the two treatments. Forest plots were used to describe the network comparison results between each treatment and the control.

The restricted maximum likelihood estimation was used to quantify network heterogeneity. The Q statistic was used to assess the sum of statistics for heterogeneity (within designs) and for overall inconsistency (between designs).¹⁴

The ranking probabilities of each regimen were estimated using the surface under the cumulative ranking curve (SUCRA), which was the ratio of the area under the curve to the entire area. A comparison-adjusted funnel plot was used to examine potential publication biases in the NMA. P values of <0.05 were considered to indicate statistical significance. The NMA was performed using R language with the ‘netmeta’ package.

Results

After removing duplicates, we obtained 1588 literature items. After screening the titles and abstracts, 1515 irrelevant studies were excluded. Seventy-three articles were screened for full text. The following articles were excluded: dose–response studies (8); those where no PCI population or subgroup was reported (6); those where no mortality or MACE-related outcomes were reported (6); repeated publications (5); studies related to preloading of lipid-lowering agents (4); studies unrelated to lipid-lowering agents (3); a protocol study (1) and a study with a non-randomised controlled trial (RCT) design (1). Finally, 39 articles were included, containing 54 478 post-PCI patients^15–53 (figure 1).

Figure 1.

Flowchart of the study selection process for eligible studies. *Consider, if feasible to do so, reporting the numbers of records identified from each database or register searched (rather than the total number across all database/registers). **If automation tools were used, indicate how many records were exculded by a human and how many were exculded by automation tools. MACE, major adverse cardiovascular event; PCI, percutaneous coronary intervention; RCT, randomised controlled trial.

Among the included studies, the publication period ranged from 1991 to 2022. The research locations were mainly in Asia (China, Japan and South Korea), Europe (Netherlands, France and Italy), America and multiple centres. There were 10 studies with sample sizes greater than 1000 patients. There were also 22 studies with publicly available clinical study registration numbers (table 1). In terms of design quality, all included studies were RCTs. Therefore, the design quality was generally high. The main factors potentially affecting design quality were the blinding of participants and personnel and blinding of outcome assessment (figure 2). However, as the desired outcomes were mortality and MACEs, the subjective factors of the investigator had little influence on the outcomes.

Table 1.

The characteristics of included studies

Study	Location	Sample size	Abbreviation	Register ID	Intervention	Control	Follow-up*
Räber et al15 2022	European	300	PACMAN-AMI	NCT03067844	Alirocumab; rosuvastatin	Placebo; rosuvastatin	52W
Peterson et al16 2022	Multicentre	3408	REDUCE-IT PCI	NCT01492361	Icosapent ethyl; statins	Placebo; statins	4.8Y
Furtado et al17 2022	Multicentre	17 073	FOURIER	NCT01764633	Evolocumab; statins	Placebo; statins	2.2Y
Okada et al18 2022	Japan	102	–	UMIN000028729	Evolocumab; pitavastatin	Pitavastatin	4W
Hao et al19 2022	China	136	–	–	Evolocumab; atorvastatin; ezetimibe	Ezetimibe; atorvastatin	3M
Deng et al20 2021	China	90	–	–	Ezetimibe; atorvastatin	Atorvastatin	1Y
Sun et al21 2021	China	171	–	ChiCTR-IPR-17012219	Ezetimibe; rosuvastatin	Rosuvastatin	3M
He et al22 2020	China	192	–	–	Atorvastatin vs Rosuvastatin vs Simvastatin	–	6M
Hibi et al23 2018	Japan	128	Ezetimibe-ACS	NCT00549926	Ezetimibe; pitavastatin	Pitavastatin	1Y
Im et al24 2017	Korea	2000		NCT01557075	Atorvastatin	Pravastatin	1Y
Hagiwara et al25 2017	Japan	1734	HIJ-PROPER	UMIN000002742	Ezetimibe; pitavastatin	Pitavastatin	36M
Guo et al26 2017	China	137	–	–	Rosuvastatin	Control	1Y
Wang et al27 2017	China	132	–	ChiCTR-IPR-15007035	Pitavastatin	Atorvastatin	6M
Watanabe et al28 2017	Japan	193	CHERRY	UMIN000002815	EPA; pitavastatin	Pitavastatin	6–8M
Liu et al29 2017	China	102	–	–	Ezetimibe; atorvastatin	Atorvastatin 20 mg/day	1Y
Nosaka et al30 2016	Japan	241	–	UMIN000016723	EPA; pitavastatin	Pitavastatin	1Y
Matsushita et al31 2016	Japan	118	Yokohama-ACS	NCT00549926	Atorvastatin vs Pitavastatin vs Pravastatin vs Fluvastatin	–	10.3M
Cannon et al32 2015	Multicentre	12 941	IMPROVE-IT	NCT00202878	Ezetimibe; simvastatin	Simvastatin	6M
Tsujita et al33 2015	Multicentre	246	PRECISE-IVUS	NCT01043380	Ezetimibe; atorvastatin	Atorvastatin	1Y
Nicholls et al34 2015	Multicentre	3295	VISTA-16	NCT01130246	Varespladib; atorvastatin	Placebo; atorvastatin	6M
Zhang et al35 2015	China	104	–	–	Atorvastatin	Rosuvastatin	6M
Leoncini et al36 2014	Italy	333	PRATO-ACS	NCT01185938	Rosuvastatin	Control	6M
Takano et al37 2013	Japan	458	PEARL	UMINC000000428	Pitavastatin	Control	35.5M
Nozue et al38 2015	Japan	164	TRUTH	UMIN000004627	Pitavastatin	Pravastatin	2Y
Lablanche et al39 2010	Multicentre	887	CENTAURUS	NCT00296387	Rosuvastatin	Atorvastatin	3M
Gibson et al40 2009	USA	2868	PROVE IT-TIMI 22	NCT00382460	Atorvastatin	Provastatin	2Y
Han et al41 2009	China	1275	–	NCT00405717	Atorvastatin	Provastatin	1Y
Hiro et al42 2009	Japan	307	JAPAN-ACS	NCT00242944	Pitavastatin	Atorvastatin	1Y
Dohi et al43 2009	Japan	180	Extended-ESTABLISH trial	–	Atorvastatin	Control	4Y
Toi et al44 2009	Japan	160	–	–	Pitavastatin	Atorvastatin	17D
Xu et al45 2007	China	648	–	–	Atorvastatin	Control	2Y
Bae et al46 2004	Korea	205	–	–	Atorvastatin	Control	6M
Serruys et al47 2002	Multicentre	1677	LIPS	–	Fluvastatin	Placebo	3.9Y
Mulder et al48 2000	Netherland	201	REGRESS	–	Pravastatin	Placebo	2Y
Flaker et al49 1999	Multicentre	1154	CARE trial	–	Pravastatin	Placebo	6Y
Bertrand et al50 1997	France	695	PREDICT	–	Pravastatin	Placebo	6M
O'Keefe Jr et al51 1996	USA	200	APPLE	–	Probucol; lovastatin	Placebo	6M
Onaka et al52 1994	Japan	66	–	–	Pravastatin	Control	5M
Sahni et al53 1991	USA	157	–	–	Lovastatin	Control	6M

*Follow-up period.

D, days; EPA, eicosapentaenoic acid; M, months; W, weeks; Y, years.

Figure 2.

Methodological quality assessment of included studies.

As the two studies did not specify the types of statins, the NMA was divided into two parts. One part was analysed based on specific types of statins, and the other was based on taking statins as a whole. For the dichotomous results of MACEs, the NMA based on specific types of statins included 18 lipid-lowering regimens. The Q test for heterogeneity (p=0.07) and inconsistency (p=0.16) was non-significant, indicating no evidence of heterogeneity or inconsistency in the NMA.

In pairwise comparisons with the control, alirocumab plus rosuvastatin (OR: 0.18; 95% CI: 0.07 to 0.44; SUCRA: 0.94), evolocumab plus atorvastatin and ezetimibe (OR: 0.18; 95% CI: 0.05 to 0.63; SUCRA: 0.90) and ezetimibe plus rosuvastatin (OR: 0.29; 95% CI: 0.11 to 0.76; SUCRA: 0.80) had significant advantages and relatively high SUCRA rankings. No potential publication bias was found according to the comparison-adjusted funnel plot (figure 3).

Figure 3.

Network plots of comparisons for major outcomes included in the analyses. (A) Dichotomous results of MACE based on specific types of statins. (B) Dichotomous results of MACE based on taking statins as a whole. (C) HR results of MACE based on specific types of statins. (D) HR results of MACE based on taking statins as a whole. (E) Dichotomous results of mortality based on specific types of statins. (F) Dichotomous results of mortality based on taking statins as a whole. MACE, major adverse cardiovascular event.

In the NMA based on taking statins as a whole, 10 regimens were analysed. Evolocumab plus ezetimibe and statins (OR: 0.19; 95% CI: 0.06 to 0.59; SUCRA: 0.92), alirocumab plus statins (OR: 0.27; 95% CI: 0.13 to 0.59; SUCRA: 0.87) and icosapent ethyl plus statins (OR: 0.39; 95% CI: 0.25 to 0.62; SUCRA: 0.72) had significant advantages and relatively high SUCRA rankings. No potential publication bias was found.

For the HR results of MACEs, the NMA based on specific types of statins included nine regimens. The Q test for heterogeneity was non-significant (p=0.964) because the network comparisons lacked loops. Therefore, the results were considered consistent. Compared with the control, eicosapentaenoic acid (EPA) plus pitavastatin (HR: 0.67; 95% CI: 0.49 to 0.96; SUCRA: 0.91), atorvastatin (HR: 0.76; 95% CI: 0.63 to 0.90; SUCRA: 0.83) and varespladib plus atorvastatin (HR: 0.77; 95% CI: 0.61 to 0.97; SUCRA: 0.77) had significant advantages and relatively high SUCRA rankings. Potential publication bias was not analysed due to the small number of included studies.

In the NMA based on taking statins as a whole, seven regimens were analysed. EPA plus statins (HR: 0.60; 95% CI: 0.42 to 0.85; SUCRA: 0.96) and icosapent ethyl plus statins (HR: 0.73; 95% CI: 0.62 to 0.86; SUCRA: 0.81) had significant advantages over the control.

For the dichotomous mortality results, the NMA based on specific types of statins included 17 lipid-lowering regimens. The Q test for heterogeneity (p=0.78) and inconsistency (p=0.99) was non-significant. Due to the rare occurrence of events, the results of the comparison had low precision with a large SE. Compared with the control, only rosuvastatin (OR: 0.30; 95% CI: 0.11 to 0.84; SUCRA: 0.79) showed a significantly better effect. Ezetimibe plus rosuvastatin had a relatively high SUCRA ranking, but there was no significant difference compared with the control (OR: 0.14; 95% CI: 0.02 to 1.26; SUCRA: 0.86). No potential publication bias was found (figure 4).

Figure 4.

Forest plots of lipid-lowering therapy compare to control for outcomes in network meta-analysis with SUCRA ranking results. (A) Dichotomous results of MACE based on specific types of statins. (B) Dichotomous results of MACE based on taking statins as a whole. (C) HR results of MACE based on specific types of statins. (D) HR results of MACE based on taking statins as a whole. (E) Dichotomous results of mortality based on specific types of statins. (F) Dichotomous results of mortality based on taking statins as a whole. MACE, major adverse cardiovascular event; SUCRA, surface under the cumulative ranking curve.

In the NMA based on taking statins as a whole, nine regimens were analysed. Ezetimibe plus statins (OR: 0.55; 95% CI: 0.43 to 0.89; SUCRA: 0.75) and icosapent ethyl plus statins (OR: 0.66; 95% CI: 0.45 to 0.96; SUCRA: 0.63) had significant advantages compared with the blank control group. No potential publication bias existed. NMA was not performed due to the small number of studies reporting HRs for mortality (figure 5).

Figure 5.

The comparison-adjusted funnel plot for assessing all main outcomes. (A) Dichotomous results of MACE based on specific types of statins. (B) Dichotomous results of MACE based on taking statins as a whole. (C) Dichotomous results of mortality based on specific types of statins. (D) Dichotomous results of mortality based on taking statins as a whole. MACE, major adverse cardiovascular event.

Discussion

This study analysed the benefits of lipid-lowering therapy on mortality and MACE outcomes in patients who underwent PCI by NMA. The results showed that several lipid-lowering regimens could reduce the risk of MACEs compared with the blank control. Icosapent ethyl plus statins had the benefit of reducing both the risk of MACEs and mortality. However, EPA plus statins had more advantages in reducing the risk of MACEs. Of note, based on the current evidence, alirocumab and evolocumab plus statins had obvious advantages in reducing the risk of MACEs but had no obvious benefit in reducing the risk of mortality.

EPA is a long-chain omega-3 polyunsaturated fatty acid. Long-term intake of EPA can reduce the residual cardiovascular risk to reduce the risk of MACEs.⁵⁴ In terms of pathological mechanisms, EPA combined with pitavastatin was shown to reduce the lipid volume of coronary artery plaques and total atherosclerotic plaque volume in patients who underwent PCI, which may be the reason for the reduced risk of MACEs.⁵⁵

Icosapent ethyl is a highly purified and stable EPA ethyl ester that has potential higher anti-inflammatory, antioxidant, plaque stability and cell membrane stability effects.⁵⁶ In the NMA results, icosapent ethyl plus statins had significant benefits for reducing the risk of either mortality or MACEs in patients who underwent PCI, which was an ideal regimen for the population.

Ezetimibe inhibits the absorption of cholesterol and has a synergistic lipid-lowering pharmacological effect with statins to further reduce the risk of death and MACEs. In particular, when combined with rosuvastatin, ezetimibe has a stronger lipid-lowering effect with a high safety profile without the risk of drug interactions.⁵⁷ Our NMA results also showed that ezetimibe can reduce the risk of MACEs and mortality. According to the guidelines for the management of dyslipidaemia from the European Society of Cardiology and the European Atherosclerosis Society, ezetimibe was recommended if the LDL-C target was not reached.^{58 59} The American College of Cardiology guidelines also recommend adding ezetimibe when using maximally tolerated statin therapy and if LDL-C levels remained ≥70 mg/dL.⁶⁰ These benefits have also been demonstrated in the secondary prevention of PCI.

Alirocumab and evolocumab are both proprotein convertase subtilisin/kexin type-9 inhibitors (PCSK9is), which can increase the level of LDL receptor in the liver, thus improving the ability of the liver to bind LDL-C and reducing the level of peripheral LDL-C.⁶¹ There was also a synergistic lipid-lowering pharmacological effect when PCSK9is were combined with statins that resulted in a significantly reduced LDL-C concentration and atherosclerosis event risk; however, there was still controversy regarding the mortality risk reduction.⁶² It has been suggested that the powerful effect of PCSK9is on reducing LDL-C predisposes patients to hypocholesterolaemia, which will not increase the risk of cerebral haemorrhage. PCSK9is may be the preferred lipid-lowering agents in patients with elevated Intra-Cerebral Hemorrhage (ICH) risk.^63–65 On the other hand, PCSK9is did not reduce serum inflammatory factors in one study, suggesting that they may not reduce the risk of residual inflammation in the post-PCI population.⁶⁶

In the results of this study, lipid-lowering therapy strategies had general advantages in reducing MACE risk. However, for all-cause mortality, the advantage of lipid-lowering therapy was not obvious. Based on dichotomous outcomes of mortality, some strategies may even have a tendency to increase the mortality risk. This challenges the opinion that lipid-lowering therapy is recommended after PCI.⁶⁷ A large sample size retrospective study suggests that statins can reduce the risk of all-cause death in patients with coronary artery disease undergoing PCI, regardless of individual cholesterol levels.⁶⁸ Alternatively, the ‘lipid paradox’ view has been proposed and indicates that higher levels of LDL-C and triglycerides on admission are associated with better clinical outcomes. Especially in patients with ST-elevation myocardial infarction, lower LDL-C levels were associated with worse mortality outcomes.⁶⁹ However, this view is also controversial.⁷⁰

On the other hand, it is possible that the contribution of LDL-C reduction to the risk of mortality outcomes is obscured by other confounding factors. For example, inflammatory status may also have had an important impact on patient mortality risk. In a cohort of post-PCI patients with low LDL-C levels, residual inflammatory risk also had a significant effect on overall mortality.⁷¹ C reactive protein can also predict long-term mortality in post-PCI patients independent of LDL-C levels.⁷² In addition, cardiac remodelling also has an important impact on the survival outcome of post-PCI patients.⁷³

There are several limitations in this study. First, this analysis was based on the study level instead of the individual level, making it difficult to consider the individual confounding factors in the analysis. Second, the two included studies did not specify the type of statins, so our study had to be analysed separately according to whether all statins were considered as a whole. Third, the criteria for defining MACEs varied among studies, which contributed to heterogeneity among the study results. Fourth, many included studies only reported dichotomous outcomes but did not report the HR results, resulting in missing relevant data for the analysis.

In conclusion, the results of this study suggested that EPA, especially icosapent ethyl, plus statins had a beneficial effect on reducing the risk of MACEs and mortality in post-PCI patients. PCSK9is plus statins were able to reduce the risk of MACEs, but the effects on the risk of mortality remained unclear.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

Not applicable.