Author + information
- Received May 27, 2010
- Revision received June 22, 2010
- Accepted July 6, 2010
- Published online February 8, 2011.
- Chris C.S. Lim, MBBS⁎,†,‡,
- William J. van Gaal, MBBS, MSc, MD†,‡,
- Luca Testa, MD∥,
- Florim Cuculi, MD⁎,
- Jayanth R. Arnold, MD⁎,
- Theodoros Karamitsos, MD¶,
- Jane M. Francis, DCR(R), DNM¶,
- Steffen E. Petersen, MD⁎,
- Janet E. Digby, PhD⁎,
- Stephen Westaby, PhD⁎,
- Charalambos Antoniades, MD, PhD⁎,
- Rajesh K. Kharbanda, PhD⁎,
- Louise M. Burrell, MD‡,§,
- Stefan Neubauer, MD¶ and
- Adrian P. Banning, MD⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Adrian P. Banning, Consultant Cardiologist, The John Radcliffe, Headley Way, Oxford OX3 9DU, United Kingdom
Objectives We aimed to assess the differential implications of creatine kinase-myocardial band (CK-MB) and troponin measurement with the universal definition of periprocedural injury after percutaneous coronary intervention.
Background Differentiation between definitions of periprocedural necrosis and periprocedural infarction has practical, sociological, and research implications. Troponin is the recommended biomarker, but there has been debate about the recommended diagnostic thresholds.
Methods Thirty-two patients undergoing multivessel percutaneous coronary intervention and late gadolinium enhancement (LGE) cardiac magnetic resonance (CMR) imaging in a prospective study had cardiac troponin I, CK-MB, and inflammatory markers (C-reactive protein, serum amyloid A, myeloperoxidase, tumor necrosis factor alpha) measured at baseline, 1 h, 6 h, 12 h, and 24 h after the procedure. Three “periprocedural injury” groups were defined with the universal definition: G1: no injury (biomarker <99th percentile); G2: periprocedural necrosis (1 to 3 × 99th percentile); G3: myocardial infarction (MI) type 4a (>3 × 99th percentile). Differences in inflammatory profiles were analyzed.
Results With CK-MB there were 17, 10, and 5 patients in groups 1, 2, and 3, respectively. Patients with CK-MB–defined MI type 4a closely approximated patients with new CMR-LGE injury. Groups defined with CK-MB showed progressively increasing percentage change in C-reactive protein and serum amyloid A, reflecting increasing inflammatory response (p < 0.05). Using cardiac troponin I resulted in 26 patients defined as MI type 4a, but only a small minority had evidence of abnormality on CMR-LGE, and only 3 patients were defined as necrosis. No differences in inflammatory response were evident when groups were defined with troponin.
Conclusions Measuring CK-MB is more clinically relevant for diagnosing MI type 4a, when applying the universal definition. Current troponin thresholds are oversensitive with the arbitrary limit of 3 × 99th percentile failing to discriminate between periprocedural necrosis and MI type 4a. (Myocardial Injury following Coronary Artery bypass Surgery versus Angioplasty: a randomised controlled trial using biochemical markers and cardiovascular magnetic resonance imaging; ISRCTN25699844)
Periprocedural myocardial injury (PMI) can result from procedural complications of percutaneous coronary intervention (PCI), such as distal embolization, side-branch occlusion, coronary dissection, and disruption of collateral flow (1). In some of these cases a complication is clinically evident, but evidence of myocardial injury can also be detected after routine uneventful PCI procedures. The increasing sensitivity of blood biomarkers, particularly troponin, has reinvigorated debate about the significance of elevations in troponin, creatine kinase (CK), and creatine kinase-myocardial band (CK-MB) (2–7), but it is clear that elevation of CK-MB >3 to 8× upper reference limit (URL) has a prognostic implication especially if accompanied by the development of Q waves on the electrocardiogram (8,9). Troponin is a particularly sensitive biomarker, introduced predominantly for risk stratification in patients with acute coronary syndrome (10). Studies measuring troponin after PCI with cardiac magnetic resonance (CMR) with late gadolinium enhancement (LGE), have demonstrated a relationship between moderate to high troponin elevations and CMR evidence of new myocardial necrosis (11,12).
Recognizing the need for a universal definition of periprocedural myocardial infarction (MI), the joint European Society of Cardiology/American College of Cardiology Foundation/American Heart Association/World Heart Federation task force recently established definitions for myocardial injury after PCI. Periprocedural myocardial necrosis (PMN) is defined as elevation of cardiac biomarkers above the URL of the 99th percentile of the normal population, assuming normal baseline troponin levels and an assay coefficient of variation <10% at this limit (13). The Universal Definition states that troponin is the preferred biomarker and an elevation of more than 3× the 99th percentile URL is defined as a PCI-related MI (MI type 4a) (13). There are currently no data to demonstrate whether this arbitrary limit in the absolute level of the biomarker appropriately divides patients with PMN and those with MI type 4a. Indeed the definition of MI type 4a has been considered to be too sensitive for reporting in clinical trials and consequently, investigators have advocated use of the less-sensitive CK-MB assay instead of troponin when applying this new definition of MI (14,15).
Percutaneous coronary intervention induces plaque disruption and local vessel wall trauma that can induce an inflammatory response measurable by a rise in circulatory inflammatory cytokines (16–20). Additionally myocyte necrosis will induce a larger inflammatory response, which can also be measured (21,22).
With data from a prospective randomized trial of complex PCI we aimed to analyze evidence of myocardial injury between patients with diagnosis of PMN and those with MI type 4a.
We hypothesized that the current arbitrary definitions with CK-MB might be superior to troponin in representing the continuum of periprocedural injury (no injury → PMN → MI type 4a) and measured inflammatory cytokine profiles and sought anatomic evidence of infarction with CMR-LGE.
The MICASA (Myocardial Injury following Coronary Artery Surgery versus Angioplasty) trial was a prospective, single-center, randomized (1:1) trial of myocardial injury after PCI compared with coronary artery bypass grafting. The primary end point was myocardial injury defined by troponin and CMR.
The study was approved by the local ethics committee, and informed written consent was obtained from each patient.
Treatment and procedures
A detailed description of the study procedures is described elsewhere (23). Briefly, patients undergoing PCI were pre-treated with aspirin 75 mg once daily for at least 2 days and clopidogrel 300 mg at least 6 h before PCI. Heparin was given intravenously to all patients. Upfront glycoprotein IIb/IIIa inhibitor (abciximab) followed by a 12-h infusion and use of drug-eluting stents were the preferred strategy. All patients underwent PCI in 1 session. Most bifurcations were managed with a provisional T strategy.
CMR imaging time points, protocol, post-processing, and data analysis
Patients were studied at 1.5-T (Sonata, Siemens Healthcare, Erlangen, Germany). Baseline CMR assessment was performed in the fortnight before revascularization. Repeat CMR was performed 7 days (range 4 to 10 days) after the revascularization. The LGE imaging was performed as previously described (11,12). For analysis of LGE, 2 experienced observers who were blinded to patient data interpreted the images. When measurements were different, review was performed by a third observer and a consensus was obtained. Areas of LGE were quantified with customized software (MATLAB R2007b, Mathworks, Natick, Massachusetts) with computer-assisted planimetry of short-axis images. With this method of LGE imaging, the minimum detectable limit of LGE necrosis is a group of 10 hyperenhanced pixels—a voxel of 1.9 × 1.4 × 7 mm (12).
Plasma samples were obtained at baseline and at 1, 6, 12, and 24 h after the procedure. Samples were also stored in a −81°C locked freezer. Cardiac troponin I (cTnI) and CK-MB were quantified with automated chemiluminescent immunoassay techniques on the Siemens ADVIA Centaur (cTnI-“Ultra” for the majority) and Siemens IMMULITE, respectively (both Siemens Healthcare Diagnostics, Frimley, United Kingdom). The ADVIA Centaur cTnI-Ultra assay has a 10% imprecision at 0.05 μg/l, with a 99th percentile URL of 0.06 μg/l. The IMMULITE CK-MB assay has a 99th percentile URL of 4.8 μg/l, with <10% imprecision at this level.
Initial analyses were performed with the Siemens ADVIA Centaur assay before the release of the new, highly sensitive ADVIA Centaur cTnI-Ultra. To meet guideline requirements of <10% imprecision at all reported biomarker levels, re-analysis of frozen stored plasma samples with the “Ultra” assay was performed when cTnI was <0.20 μg/l.
Detection of inflammatory markers with the Luminex multiplex assay
Measurement of inflammatory markers from frozen serum was performed with a multiplex assay with Luminex xMAP technology (Luminex Corporation, Austin, Texas). Tumor necrosis factor (TNF) alpha, myeloperoxidase (MPO), high-sensitivity C-reactive protein (CRP), and serum amyloid A (SAA) were measured in duplicate with Milliplex MAP kit (Millipore, Billerica, Massachusetts). Intra- and inter-assay co-efficient of variation (CV) for all biomarkers were between 6% and 13%. The peak value is defined as the maximum value reached in the 24 h after PCI or, in the case of negative change, the minimum value reached. The maximum change is defined as the peak value minus the baseline value.
The 12-lead electrocardiograms were obtained before and after the PCI procedure and were analyzed for the presence of new Q waves.
Definition of PMI groups
Patients were stratified into 3 groups with CK-MB: group 1: no injury (normal CK-MB); group 2: PMN (CK-MB elevation between 1× and 3× 99th percentile URL); group 3: MI type 4a (CK-MB >3× 99th percentile). For comparison, we similarly defined the 3 universal definition cTnI groups: group 1 (no injury - no elevation of cTnI), group 2 (PMN–cTnI elevated between 1× and 3× 99th percentile URL); group 3 (MI type 4a–cTnI >3× 99th percentile. Changes in CRP, SAA, MPO, and TNF-alpha levels were compared between the CK-MB– and troponin-defined groups. Additionally a comparison was performed between patients with evidence of CMR-LGE and those without.
Testing for normal distribution was performed with the Kolmogorov-Smirnov test for normality. Normally distributed variables are presented as mean ± SD unless stated otherwise. For comparisons between the 3 groups, we used 1-way analysis of variance for multiple comparisons followed by Bonferroni post hoc correction. Non-normally distributed variables are presented as median (25th to 75th percentile), and for comparisons between groups we used either the Kruskal-Wallis test (between 3 groups) or the Mann-Whitney U test (between 2 groups). For categorical outcomes, chi square or Fisher exact test was performed where appropriate. A probability of p < 0.05 was considered statistically significant. Receiver-operator characteristic (ROC) curves were plotted to identify the optimal biomarker cutoff point, which was defined as the point with the shortest distance from point (0,1) (top left corner). MedCalc version 11.3 software (MedCalc Software, Mariakerke, Belgium) and the DeLong et al. (24) method was used for the comparison of ROC curves. SPSS version 17.0 (SPSS Inc., Chicago, Illinois) statistical package was used for all other analyses.
The PCI cohort of the MICASA study consisted of 40 patients, and a total of 32 were included in this analysis. Eight patients were excluded from the study: 2 due to elevated baseline cTnI, 1 had a pre-procedure upper respiratory infection with markedly raised baseline CRP of 75,000 ng/ml, 1 underwent rotational atherectomy, and 4 had incomplete blood sampling (Fig. 1).
The baseline characteristics of the patients are shown in Table 1. The mean age was 65 ± 8.8 years in group 1, 58.3 ± 6.6 years in group 2, and 62.3 ±7.6 years in group 3 (p = NS). There was no evidence of difference in cardiovascular risk factors and pattern of coronary artery disease between the groups (Table 1).
There were no significant differences in baseline troponin, CK-MB, and inflammatory marker levels. The mean number of implanted stents was significantly higher in groups 2 and 3 compared with group 1: 5.2 ± 1.9 and 5.0 ± 1.2 versus 3.5 ± 1.7, p = 0.03.
The procedure was uneventful in 27 (84%) patients. Occlusive dissection was successfully treated in 3 patients, and in 1 patient there was hemodynamic disturbance probably caused by distal embolization of plaque material. One patient failed complete revascularization due to inability to cross a chronic total occlusion and later underwent coronary artery bypass grafting. Patients undergoing PCI received an average of 4.3 ± 1.8 stents, with an average stent length of 78.9 ± 34.0 mm/patient. All patients received drug-eluting stents: Taxus Express (44%) and Promus (19%) (Boston Scientific, Natick, Massachusetts); Xience V (32%) (Abbott Vascular, Redwood City, California); Cypher Select (5%) (Cordis, Johnson and Johnson, Warren, New Jersey). Two patients also received a bare-metal stent each.
Twenty-three (72%) patients received abciximab, and 21 of these patients also received a 12-h infusion after PCI.
Comparison of cTnI- and CK-MB–defined periprocedural injury groups
Baseline CK-MB and cTnI values were not significantly different among the 3 PMI groups (Table 2).Figure 2 demonstrates the correlation between peak CK-MB and peak cTnI (r2 = 0.79, p < 0.001). A strong correlation was also present in patients without new LGE (n = 27, r2 = 0.61, p < 0.001).
Five of 32 patients (15%) had evidence of CMR-LGE. The mean peak cTnI value in these patients was 13.4 μg/l versus 0.93 μg/l in patients without CMR-LGE (p < 0.001), and mean peak CK-MB value was 28.3 versus 4.7 μg/l (p < 0.001). All 5 patients with CMR-LGE fulfilled the criteria for cTnI-defined MI type 4a. However when using CK-MB–based definitions 2 of 5 patients with CMR-LGE are categorized as PMN (group 2), and 3 of 5 patients have CK-MB–defined MI type 4a.
Comparison of the observed thresholds for PMN and MI type 4a suggests that many more patients are categorized with MI type 4a with troponin than when the CK-MB measurement is used (Fig. 2, dashed lines depict the cutoff for cTnI- and CK-MB–defined MI type 4a). Figure 3 compares the spectrum of patient distribution when the PMI groups are defined with CK-MB (Fig. 3A) and cTnI (Fig. 3B).
If grouping is performed with cTnI, 3 of 32 patients (9%) have no evidence of PMI, 3 of 32 patients (9%) have PMN, and 26 of 32 patients (82%) are considered to have MI type 4a. However, only 5 of those 26 patients had evidence of CMR-LGE giving a positive predictive value of only 19% and a poor specificity of 22% (Table 3).
When the CK-MB definition was applied, 17 of 32 patients (53%) had no evidence of injury, 10 of 32 patients (32%) had PMN, and 5 of 32 patients (15%) were considered to have MI type 4a. In contrast, there was a high specificity of 93% for CMR-LGE, and 3 of 5 patients meeting MI type 4a criteria had new LGE giving a positive predictive value of 60% (Table 3).
Inflammatory markers according to PMI groups
Patients with evidence of LGE on CMR had significantly greater elevations of CRP (18.6 [11.0 to 45.1] mg/ml vs. 3.6 [1.7 to 8.5] mg/ml, p < 0.01) and SAA (34.5 [11.4 to 239.0] mg/ml vs. 5.3 [2.3 to 16.8] mg/ml, p < 0.05) but not MPO (190 [34 to 549] ng/ml vs. 90 [22 to 210] ng/ml) or TNF-alpha.(0.28 [−0.36 to 6.1] pg/ml vs. 0.69 [0.18 to 2.6] pg/ml) compared with those without LGE.
Figure 4 demonstrates the changes of the 4 inflammatory markers according to CK-MB– and cTnI-defined injury groups. The percentage changes of both CRP and SAA serum levels progressively increased across the CK-MB-defined injury groups (p < 0.05 for both) (Fig. 4). However, the changes in serum levels of MPO and TNF-alpha across the CK-MB–defined injury groups did not reach statistical significance.
In contrast, there was no significant difference in the levels of the examined inflammatory markers across the 3 cTnI-defined groups (Fig. 4). Therefore, the change in inflammatory marker profiles reflect the differentiation of injury made evident by measurement of CK-MB. No differentiation in the magnitude of injury was evident when troponin categorizations were used.
ROC analysis for biomarker detection of new CMR-LGE
The ROC analysis for detection of CMR-LGE shows areas under the curve of 0.985 (95% confidence interval: 0.864 to 1.000) for cTnI versus 0.970 (95% confidence interval: 0.840 to 0.999) for CK-MB, with no significant difference between the 2 areas (p = 0.411). This indicates that cTnI is not inferior to CK-MB for diagnosis of LGE necrosis; however, at the current cutoff of 3× 99th percentile URL cTnI suffers from very poor specificity.
The optimal cTnI cutoff from ROC analysis of our data would be a cTnI of >2.4 μg/l (100% sensitivity, 93% specificity), which is 40× its 99th percentile URL of 0.06 μg/l. The optimal cutoff for CK-MB is >9.5 μg/l (100% sensitivity, 93% specificity), which is double its 99th percentile URL of 4.8 μg/l.
Our data demonstrate that cTnI and CK-MB have an almost perfect linear correlation. Consequently both cTnI and CK-MB can be used to detect periprocedural injury, but the use of the current cutoff multipliers of the 99th percentile result in differential results and categorization. By altering the thresholds for cTnI, PMI group categorizations similar to those described with CK-MB can be obtained. For example, with our data, the equivalent thresholds for cTnI would be:
G2: CK-MB–defined PMN: cTnI >0.8 μg/l (12× 99th percentile URL)
G3: CK-MB–defined MI: type 4a cTnI >3.0 μg/l (50× 99th percentile URL)
The ideal definition thresholds
With a PMN cutoff approximating that of CK-MB–defined PMN and an MI Type 4a cutoff calculated from ROC analysis with CMR-LGE injury, we can propose “ideal” cTnI cutoffs of >12× 99th percentile (0.8 μg/l) for PMN and >40× 99th percentile (2.4 μg/l) for MI type 4a. Figure 3C illustrates the periprocedural injury group distribution of such a cTnI definition.
This study confirms that a considerable number of patients undergoing complex PCI have elevation of biomarkers despite a successful PCI procedure. Marked variation occurs in proportions reaching universal definition arbitrary diagnostic levels for necrosis or infarction depending on whether CK-MB or troponin is measured. Our data suggest that, with this definition, CK-MB more readily differentiates necrosis and infarction than troponin. With CK-MB slightly more than one-half of the patients have no evidence of periprocedural injury, one-third of them have evidence of periprocedural necrosis, and roughly 15% can be classified as MI type 4a. Using troponin radically alters this distribution, with the vast majority of the patients (>80%) fulfilling the criteria for MI type 4a. However, redefining the troponin I thresholds to 0.8 μg/l (12× 99th percentile) for necrosis and 2.4 μg/l (40× 99th percentile) for MI type 4a allows a comparable categorization to those identified by measurement of post-procedural CK-MB.
Ten years ago, the CK-MB assay was regarded as the best biomarker for detection of myocardial injury. The replacement of CK-MB with troponin as the cardiac biomarker of choice was primarily due to its superiority in acute coronary syndrome risk stratification, because it was able to detect patients that had experienced minor degrees of ischemia and were at risk despite having CK-MB levels below the URL. This was a consequence of the enhanced sensitivity and specificity of troponin, resulting in improved differentiation at low level, and also the higher population variability of CK-MB levels, resulting in a higher absolute 99th percentile CK-MB URL. Clinical experience with CK-MB elevation after PCI demonstrated its ability to predict mortality, but troponin elevation (defined as >99th percentile) was not predictive (25). In this study by Cavallini et al. (25), elevation of troponin occurred in almost 3 times as many patients as CK-MB. Consequently it is likely that many patients categorized with troponin elevation were in the range <0.8 μg/l, but their CK-MB was normal. This over-stringent categorization of troponin abnormality would result in significant dilution of any potential observed effect. Although it is clear that higher levels of troponin elevation after PCI accurately reflect new myocardial necrosis, there has been persistent debate and concern about the clinical significance of lower levels of troponin elevation. Most recently this concern has prompted prominent clinical trial investigators to suggest avoiding the universal definition of periprocedural MI using troponin in favor of CK-MB, stating their concern over its impact on analysis and reporting of safety and efficacy data in stent trials (14,15).
Accurate differentiation between definitions of periprocedural necrosis and MI has practical and sociological implications as well as its implications for research. Lowering the threshold for MI will increase the epidemiological incidence of MI and has significant psychological impact for patients. The prospective inclusion of a heterogenous patient group with a different prognosis might potentially confound outcome trials of therapeutic strategies. In contrast, an accurate diagnosis of periprocedural necrosis has little emotive implication and might more readily be accepted as a benign procedural consequence.
In this study PCI patients with evidence of LGE on CMR have significant elevation of inflammatory markers compared with those patients with lesser biomarker abnormality and no evidence of myocardial necrosis on CMR. Concerns that the current thresholds for troponin might be oversensitive for stratification of periprocedural injury are supported by the results of the inflammatory cytokines in our study. It is known that myocardial injury initiates an inflammatory cascade resulting in proportionate inflammation, measurable by circulating markers (22). The inflammatory cytokines, especially CRP and SAA, demonstrate a progressive rise across the PMI groups when defined by CK-MB (Fig. 4). This step-wise progression was not evident when cTnI was used to define injury groups (Fig. 4).
CRP is probably the best-studied inflammatory marker in acute MI and PCI, and the proportionate rise of this marker with increasing CK-MB emphasizes its role. Serum amyloid A shares production in the liver and release kinetics characteristics with CRP, and SAA elevation after tissue damage is well-established (26). Recent studies show that SAA is a clinically useful marker of inflammation, and its levels are strongly associated with cardiovascular events (27). Myeloperoxidase is a neutrophil activation marker, with circulatory levels that peak acutely at 1 to 3 h after PCI (28).
Our study is limited by its small sample size.
The timing of biomarker samples (last time point at 24 h) might also have resulted in nonsignificant results for TNF-alpha, which might demonstrate continual release kinetics beyond 24 h.
When applying the current universal definition of MI to periprocedural injury, CK-MB should currently be the preferred biomarker. Under the current definition criteria, troponin is oversensitive with the arbitrary limit of 3× 99th percentile failing to discriminate between necrosis and MI type 4a. Redefining the troponin I thresholds for necrosis and MI type 4a will allow a more-standardized diagnostic approach to diagnosing PMI.
The authors are grateful to their colleagues who allowed their patients to participate in the study.
This work was partially supported by an unrestricted research donation from Boston Scientific. Dr. van Gaal has received research grants from Cordis and Pfizer. Dr. Banning and some of the study costs were partially funded by the National Institute for Health Research Oxford Biomedical Research Centre Programme, United Kingdom. Dr. Antoniades is funded by the European Association of Percutaneous Coronary Interventions. Prof. Neubauer, Dr. Karamitsos, and Dr. Petersen are partially funded by the Medical Research Council (United Kingdom) and British Heart Foundation. Dr. Banning has received unrestricted grants from Cordis and Boston Scientific. All other authors have reported that they have no relationships to disclose.
- Abbreviations and Acronyms
- creatine kinase-myocardial band
- cardiac magnetic resonance imaging
- C-reactive protein
- cardiac troponin I
- late gadolinium enhancement
- myocardial infarction
- percutaneous coronary intervention
- periprocedural myocardial injury
- periprocedural myocardial necrosis
- serum amyloid A
- tumor necrosis factor
- upper reference limit
- Received May 27, 2010.
- Revision received June 22, 2010.
- Accepted July 6, 2010.
- American College of Cardiology Foundation
- Cuculi F.,
- Lim C.C.,
- Banning A.P.
- Lim C.C.,
- van Gaal W.J.,
- Banning A.P.
- Prasad A.,
- Gersh B.J.,
- Bertrand M.E.,
- et al.
- Testa L.,
- van Gaal W.J.,
- Biondi Zoccai G.G.,
- et al.
- Prasad A.,
- Singh M.,
- Lerman A.,
- Lennon R.J.,
- Holmes D.R. Jr..,
- Rihal C.S.
- Brener S.J.,
- Ellis S.G.,
- Schneider J.,
- Topol E.J.
- Kini A.,
- Marmur J.D.,
- Kini S.,
- et al.
- Newby L.K.,
- Goldmann B.U.,
- Ohman E.M.
- Selvanayagam J.B.,
- Petersen S.E.,
- Francis J.M.,
- et al.
- Selvanayagam J.B.,
- Porto I.,
- Channon K.,
- et al.
- Thygesen K.,
- Alpert J.S.,
- White H.D.
- Cutlip D.E.,
- Windecker S.,
- Mehran R.,
- et al.
- Vranckx P.,
- Cutlip D.E.,
- Mehran R.,
- et al.
- Zairis M.N.,
- Ambrose J.A.,
- Manousakis S.J.,
- et al.
- Schillinger M.,
- Exner M.,
- Mlekusch W.,
- et al.
- Frangogiannis N.G.,
- Smith C.W.,
- Entman M.L.
- DeLong E.R.,
- DeLong D.M.,
- Clarke-Pearson D.L.
- Cavallini C.,
- Savonitto S.,
- Violini R.,
- et al.