Article

Perspective on the Use of True Vessel Characterization Imaging in Interventional Cardiology Clinical Practice

Abstract

Intracoronary near infrared spectroscopy (NIRS), incorporated in the True Vessel Characterization (TVC) Imaging System™, is a novel method for the identification and quantification of lipid composition in coronary plaques. The main advantage of NIRS over other plaque characterisation methods is its ability to directly identify chemical composition – the primary use of spectroscopy for other applications. NIRS by itself does not display structures, but a combination device with co-registered intravascular ultrasound (IVUS) has been developed to give simultaneous and complementary structural and compositional information. Identification of lipid core plaque with NIRS hypothetically has the potential to optimise the length of vessel to stent and to lead to effective utilisation of embolic protection devices in the native coronaries, identifying the exact location of lipid-core lesions at high risk of distal embolisation. The NIRS-IVUS device also has promise in the identification of vulnerable plaque, which may lead to strategies to prevent future coronary events.

Disclosure: The authors have no conflicts of interest to declare.

Received:

Accepted:

Citation:Interventional Cardiology 2012;7(1):17–20

Correspondence: Patrick W Serruys, Head of the Interventional Cardiology Department, Erasmus MC, Thoraxcenter, 's Gravendijkwal 230,3015 CE, Rotterdam, The Netherlands. E: p.w.j.c.serruys@erasmusmc.nl

Acknowledgements: The authors would like to thank Sean Madden from InfraReDx for his help in the preparation of the article.

Support: The publication of this article was funded by InfraReDx. The views and opinions expressed are those of the authors and not necessarily those of InfraReDx.

Open access:

The copyright in this work belongs to Radcliffe Medical Media. Only articles clearly marked with the CC BY-NC logo are published with the Creative Commons by Attribution Licence. The CC BY-NC option was not available for Radcliffe journals before 1 January 2019. Articles marked ‘Open Access’ but not marked ‘CC BY-NC’ are made freely accessible at the time of publication but are subject to standard copyright law regarding reproduction and distribution. Permission is required for reuse of this content.

The identification of vulnerable plaques has been a longstanding challenge for interventional cardiologists. Plaque composition is regarded as an important feature for assessment of plaque vulnerability. Histopathologists have in particular shown that the type of plaque most commonly prone to rupture in vivo is the thin-cap fibroatheroma (TCFA), in which the presence of lipid core (also called necrotic core) is one of the most important compositional characteristics.1,2 Previous angiographic studies, however, failed to identify those quiescent plaques prone to progress or rupture, as most of them are non-flow limiting.3–5 These lipid core plaques (LCP) are indeed not detectable with traditional diagnostic methods: coronary angiography only depicts the coronary lumen and does not provide data on plaque size, remodelling characteristics, composition or its biological activity.6 In addition, it underestimates the magnitude of atherosclerotic burden, particularly in earlier stage disease in which positive vascular remodelling may allow ‘normal’ lumen calibre despite substantial vascular wall plaque.

This growing need for more information about coronary atherosclerosis in order to identify patients and lesions at risk for complications during percutaneous coronary intervention (PCI) and for future adverse cardiac events has been the primary impetus for the development of novel intra-coronary imaging methods, able to detect plaque composition, in particular presence of a lipid core, such as near infrared spectroscopy (see Table 1).

Near Infrared Spectroscopy System

The near infrared spectroscopy (NIRS) catheter system (True Vessel Characterization [TVC] Imaging System™, InfraRedx, Burlington, Ma) is comprised of a scanning near-infrared laser and a fibre-optic catheter similar in size and use to an Intravascular ultrasound (IVUS) catheter, and an automated pullback and rotation device (see Figure 1).7–10

The NIRS catheter is a monorail system, which can be advanced over a 0.014 guidewire up to a reference point distal to the target lesion, alike an IVUS catheter. Scanning with automated rotational pullback is performed at a speed of 0.5 mm/s. The system performs approximately 32,000 chemical measurements per 100 mm of artery scanned. A predictive algorithm calculates the probability that a LCP is present at each interrogated location in the artery. The data are immediately and automatically displayed in a two-dimensional map of the vessel called a ‘chemogram’. The x-axis of the chemogram represents mm of pullback in the artery and the y-axis represents degrees of rotation: a colour scale from red to yellow indicates increasing algorithm probability that a LCP is present (see Figure 1D). The ‘block chemogram’ provides a summary of the results for each 2 mm section of artery. The numerical value of each block in the block chemogram represents the 90th percentile of all pixel values in the corresponding 2 mm chemogram segment. The block chemogram is mapped to the same colour scale as the chemogram, and the display is grouped into four discrete colours to aid in visually interpreting the algorithm probability that a LCP is present in that 2 mm block (red: p<0.57; orange: 0.57≤p≤0.84; tan: 0.84≤p≤0.98; yellow: p>0.98; (see Figure 1D). The lipid core burden index (LCBI) is provided as a quantitative summary metric of the LCP’s presence in the entire scanned region and is computed as the fraction of valid pixels within the scanned region that exceed a LCP probability of 0.6, multiplied by 1,000. A radio-opaque marker on the tip of the NIRS catheter can be used to identify the location of the catheter and imaging element in relation to target vessel fiduciary landmarks as detected by coronary angiography (e.g., stenoses of interest, branches, guide catheter) with the possibility of placing a mark on the chemogram to annotate anatomical landmarks at the locations of acquisition of the NIRS spectra.

In order to combine compositional and quantitative information on coronary plaques, a new generation combined NIRS-intravascular ultrasound catheter (TVC Insight™ Catheter) has been developed in partnership between InfraReDx and the Biomedical Engineering Group of the Thoraxcenter in Rotterdam9 (see Figure 2). This novel catheter allows a complete characterisation of coronary plaques, providing simultaneous and co-registered acquisition of their structural and compositional information. In particular, NIRS can confirm LCP presence when high plaque burden and hypo-echogenic regions are identified by IVUS, possibly indicating lesions at high risk of distal embolisation during balloon dilation and stenting.11,12 The NIRS-IVUS catheter has been approved by the US FDA (July 2010) and has received the CE mark (April 2011) for detection of LCP of interest in the coronary arteries and measurement of the LCBI.

Calibration and Validation of the System

Initial calibration and validation studies were performed in intact and blood-perfused human coronary artery autopsy specimens in order to create an advanced algorithm ultimately applicable to interpretation of NIR signals obtained in patients.7 NIR spectrographic data were compared to the histology gold standard of LCP, defined as a fibroatheroma containing a necrotic core at least 0.2 mm thick with a circumferential span of at least 60° on a cross-section. In this doubleblind, prospective validation of the system for detection of LCP in nearly 2,000 individual blocks of 51 hearts, NIRS yielded an area under the ROC curve of 0.80 (95 % CI: 0.76–0.85) for up to 3.0 mm average lumen diameters.7 The system gained a specific US FDA label claim for the detection of LCP, the first and currently only label claim of its kind. Concurrently, the SPECTroscopic Assessment of Coronary Lipid (SPECTACL) pivotal clinical study was conducted to determine if the spectra recorded in patients (in whom tissue is not available for validation) were equivalent to the spectra recorded in autopsy specimens (in whom tissue is available for histologic validation).8

This study successfully demonstrated that high-quality NIRS signals, similar to those validated in human coronary autopsy specimens, could be obtained from the coronary arteries of living patients, supporting the feasibility of invasive detection of coronary LCP with this novel system. Excellent reproducibility of the NIRS findings during repeat pullbacks has been recently demonstrated.13

We recently showed that LCP longitudinal distribution detected by NIRS is not homogenous and in particular that it is independently affected by the distance from coronary ostium, resembling the distribution of the culprit lesions in ST-elevation myocardial infarction (STEMI) patients.14 The presence of tortuosity or frequent branching close to the ostium may explain this finding: significant variation in shear stress has been indeed demonstrated at the level of bifurcations or bending, and low shear stress may be implicated in the migration of lipids and monocytes into the vessel wall, processes that could accelerate the progression of an atherosclerotic lesion towards a vulnerable plaque.15 These findings obtained by NIRS are in agreement with prior histopathological, angiographic and IVUS observations.16–18 Overall, results from these studies showed that the NIRS catheter is very safe, accurate and reproducible for localised detection of LCP, as well as for determination of overall lipid burden of a scanned artery, in patients undergoing coronary angiography. NIRS is well-suited for analysis of LCP in a coronary vessel tissue since it: can penetrate blood and several millimetres into the tissue; can overcome the problem of cardiac motion, through an ultrafast scanning laser; is capable of acquiring the thousands of spatial measurements required to create an image of the artery; and provides an automated positive and specific chemical measure of LCP (it does not rely on reader-interpreted drop-out of signal as do IVUS and optical coherence tomography), since cholesterol has prominent spectral features in the NIR region that can distinguish it from other tissue constituents such as collagen.

Recently, the NIRS findings have been compared with IVUS-VH (VH stands for Virtual Histology™, developed by Volcano Corporation) findings, which have been extensively and previously used for identification of vulnerable plaques.10 Larger coronary plaques, identified by greyscale IVUS, were more likely to be recognised as LCP and as necrotic-core rich plaques by NIRS and VH, respectively: however, the correlation between NIRS and VH was poor. In a recent paper, it has been further shown that this correlation is worse in calcified than in non-calcified plaques.19 The fundamental differences in the principles of each technique (VH is based on a pattern classification of the backscattering ultrasound signal, whereas NIRS is based on near-infrared spectral signals) and their respective limitations should be taken into account in the interpretation of the differing results between NIRS and IVUS-VH.10

Clinical Applications of Intracoronary Near Infrared Spectroscopy

Early experience with the NIRS device has revealed multiple potential uses, all of which will require validation in future studies. In general, the potential clinical applications of NIRS-IVUS delineation of plaque architecture and composition can be summarised as follows:

  • prevention of peri-procedural myocardial infarction (MI). Large LCP are known to be at high risk of distal embolisation and peri-procedural MI. In this case pre-PCI identification of such plaques can lead to the use of preventive strategies, employing distal embolic protection devices;
  • delineation of length of vessel to stent with adequate lesion coverage and optimal stent expansion; and
  • identification of vulnerable plaques.

Prevention Of Peri-procedural Myocardial Infarction

Early users of NIRS have observed cases in which balloon inflation at the LCP site resulted in no or slow reflow with associated myocardial necrosis markers elevation.9 Pre- and post-scans showing postintervention disappearance of LCP lend further evidence to the hypothesis that peri-procedural MI can develop in association with post-stent disappearance of yellow LCP, presumably due to rupture and release of LCP contents with distal embolisation. In several cases, placement of filters distal to such LCP has resulted in collection of yellow material following balloon dilatation.11

The COLOR Registry, an ongoing prospective observational study of patients undergoing NIRS prior to PCI, provided additional evidence of this capability of NIRS.12 In a recent report, the extent of LCP in the treated region was calculated as the maximal LCBI for each 4 mm longitudinal segment (maxLCBI4mm) within the treated region. Results showed that a peri-procedural MI developed in 50 % of the patients with LCP with a maxLCBI4mm of ≥500 (22.6 % of the lesions) and in only 4.2 % of the patients with LCP with a low maxLCBI4mm (p=0.0002). The knowledge that dilation of a stenosis containing a large LCP, identified by NIRS, carries a 50 % risk of causing a peri-procedural MI further indicates the need for preventive therapy. The use of an embolic protection device (EPD) to prevent distal embolisation of plaque contents, already used in PCI of vein grafts and carotid arteries, is a particularly promising approach. Since peri-procedural MI can prolong hospital stay and is an impediment to more frequent performance of stenting in a less costly outpatient setting, pre-PCI identification of such plaques at high risk of embolisation with use of preventive strategies (e.g., distal embolic protection device) may enhance the safety, efficacy and cost-effectiveness of stenting.20–29 Of note is that previous studies on the use of EPDs during stenting of lesions in the native coronary arteries failed to show clinical benefit. However, these studies did not characterise, by intra-coronary imaging, the types of plaques most likely to embolise, and therefore to benefit from the use, of an EPD.30 The potential ability of NIRS-guided use of an EPD to prevent peri-procedural MI is being tested in a prospective randomised trial, CANARY (Coronary Assessment by Near-infrared of Atherosclerotic Rupture-prone Yellow; ClinicalTrials.gov identifier NCT01268319).

Optimisation of Percutaneous Coronary Intervention

Another possible use of NIRS-IVUS in clinical practice would be to assist in decisions regarding the length of artery to be stented. During stent placement, it is generally considered to implant the stent from sites demonstrating normal distal and proximal reference diameters, selected on the basis of angiographic information alone. Although use of angiography alone generally produces good outcomes, it has been suggested that stent complications (dissection, early or late stent thrombosis, or restenosis) might result from a geographical miss in stent placement (stent ends in an area with high plaque burden) or from incomplete stent expansion or malapposition.31–34 IVUS imaging frequently shows that at the proximal and distal reference segments there could be extensive plaque with expansive remodelling.35

Angioscopy studies also revealed that placement of a drug-eluting stent over yellow lipid core plaque resulted in high frequency of stent thrombosis, due to delayed re-endothelialisation.22,34,36 Dixon et al. have recently shown that although LCP is usually located within the angiographic margins of a target lesion, in approximately 16 % of those it may extend beyond these margins into adjacent vessel segments with minimal angiographic disease.37 In this regard, NIRS-IVUS makes possible the avoidance of placement of the ends of a stent in a LCP by accurately and precisely revealing its location. Longer stents can be chosen to extend into vessels free of LCP or conversely the choice of a shorter stent could be supported by the absence of LCP in a given landing zone. Long term and additional studies are required to determine whether NIRS-guided stent length could result in better clinical outcomes.

Identification Of Vulnerable Plaque

The identification of TCFAs as ‘suspected’ vulnerable plaques provided a useful target that might be identified by novel intra-coronary imaging methods.38 Given the validated capabilities of NIRS for detection of LCP, NIRS-IVUS has also the potential for delineation of vulnerable plaques and in particular for guiding the treatment of those lesions causing an intermediate degree of coronary stenosis. As early evidence mounts for the greater propensity for rupture and more rapid progression when LCPs are present, the presence of LCPs in angiographically intermediate stenosis may support a tailored stenting. It should be considered, however, that it is unknown if we should screen for the presence of TCFAs and if we should treat them in order to prevent future adverse events. A randomised trial of stenting of such LCP lesions of intermediate stenosis needs to be conducted to test this hypothesis. An observational study of cholesterol in coronary arteries (COLOR Registry, ClinicalTrials.gov identifier NCT00831116), aimed to enroll 1,000 patients in 14 centres in the US, has been initiated by InfraReDx Inc. in order to determine the relationship between NIRS findings and subsequent events over a two-year period.

Other Possible Clinical Applications

NIRS is also likely to be useful in the choice of more intensive lipid modification therapy in some subgroups of patients. The presence of extensive LCP may, indeed, indicate the need for more intensive or different types of lowering LDL therapy. NIRS will be also useful in the development of anti-atherosclerosis medication by providing a surrogate endpoint in plaque regression/stabilisation studies. In particular, the ability of NIRS to assess the lipid content of plaques may be a more effective means of identifying a beneficial effect of an agent than IVUS. For testing these hypotheses, the IBIS-3 (integrated biomarker and imaging study) trial has begun enrolling patients at the Thoraxcenter in Rotterdam in order to determine the effect of intensive rosuvastatin therapy on the content of necrotic core (IVUS-VH) and lipid-containing regions (NIRS) at 52 weeks in a non-intervened coronary artery. Another potential future use could be to more fully inform the decision to perform PCI versus coronary artery bypass graft (CABG), based on the presence, patterns and extent of LCP. Most notably changes could be envisioned in the determination of the presence of ‘three vessel disease’ or patients who are ‘not candidates for PCI’. For example, it is reasonable to hypothesise that LCP-free, fibrotic stenoses in all three coronary vessels may in fact be better treated with stents. In contrast, diffuse lipid in all three vessels, either stenotic or non-stenotic, may turn out to be better treated by CABG.

References

  1. Virmani R, Burke AP, Farb A, Kolodgie FD, Pathology of the vulnerable plaque, J Am Coll Cardiol, 2006;47:C13–8.
    Crossref | PubMed
  2. Naghavi M, Libby P, Falk E, et al., From vulnerable plaque to vulnerable patient: a call for new definitions and risk assessment strategies: Part II, Circulation, 2003;108:1772–8.
    Crossref | PubMed
  3. Ambrose JA, Tannenbaum MA, Alexopoulos D, et al., Angiographic progression of coronary artery disease and the development of myocardial infarction, J Am Coll Cardiol, 1988;12:56–62.
    Crossref | PubMed
  4. Hackett D, Davies G, Maseri A, Pre-existing coronary stenoses in patients with first myocardial infarction are not necessarily severe, Eur Heart J, 1988;9:1317–23.
    Crossref | PubMed
  5. Little WC, Constantinescu M, Applegate RJ, et al., Can coronary angiography predict the site of a subsequent myocardial infarction in patients with mild-to-moderate coronary artery disease? Circulation, 1988;78:1157–66.
    Crossref | PubMed
  6. Topol EJ, Nissen SE, Our preoccupation with coronary luminology. The dissociation between clinical and angiographic findings in ischemic heart disease, Circulation, 1995;92:2333–42.
    Crossref | PubMed
  7. Gardner CM, Tan H, Hull EL, et al., Detection of lipid core coronary plaques in autopsy specimens with a novel catheterbased near-infrared spectroscopy system, JACC Cardiovasc Imaging, 2008;1:638–48.
    Crossref
  8. Waxman S, Dixon SR, L'Allier P, et al., In vivo validation of a catheter-based near-infrared spectroscopy system for detection of lipid core coronary plaques: initial results of the SPECTACL study, JACC Cardiovasc Imaging, 2009;2:858–68.
    Crossref | PubMed
  9. Schultz CJ, Serruys PW, van der Ent M, et al., First-in-man clinical use of combined near-infrared spectroscopy and intravascular ultrasound: a potential key to predict distal embolization and noreflow? J Am Coll Cardiol, 2010;56:314.
    Crossref | PubMed
  10. Brugaletta S, Garcia-Garcia HM, Serruys PW, et al., NIRS and IVUS for characterization of atherosclerosis in patients undergoing coronary angiography, J Am Coll Cardiol Imaging, 2011;4:647–55.
    Crossref | PubMed
  11. Goldstein JA, Grines C, Fischell T, et al., Coronary embolization following balloon dilation of lipid-core plaques, JACC Cardiovasc Imaging, 2009;2:1420–4.
    Crossref | PubMed
  12. Goldstein JA, Maini B, Dixon SR, et al., Detection of lipid-core plaques by intracoronary near-infrared spectroscopy identifies high risk of periprocedural myocardial infarction, Circ Cardiovasc Interv, 2011;4:429–37.
    Crossref | PubMed
  13. Garcia BA, Wood F, Cipher D, et al., Reproducibility of nearinfrared spectroscopy for the detection of lipid core coronary plaques and observed changes after coronary stent implantation, Circ Cardiovasc Interv, 2010;76:359–65.
    Crossref | PubMed
  14. Brugaletta S, Garcia-Garcia HM, Serruys PW, et al., In vivo distribution of lipid core containing plaque according to distance from the ostium by bear infrared spectroscopy in non-culprit coronary arteries, J Am Coll Cardiol Imaging, 2011. In press.
  15. Cunningham KS, Gotlieb AI, The role of shear stress in the pathogenesis of atherosclerosis, Lab Invest, 2005;85:9–23.
    Crossref | PubMed
  16. Cheruvu PK, Finn AV, Gardner C, et al., Frequency and distribution of thin-cap fibroatheroma and ruptured plaques in human coronary arteries: a pathologic study, J Am Coll Cardiol, 2007;50:940–9.
    Crossref | PubMed
  17. Wang JC, Normand SL, Mauri L, Kuntz RE, Coronary artery spatial distribution of acute myocardial infarction occlusions, Circulation, 2004;110:278–84.
    Crossref | PubMed
  18. Valgimigli M, Rodriguez-Granillo GA, Garcia-Garcia HM, et al., Distance from the ostium as an independent determinant of coronary plaque composition in vivo: an intravascular ultrasound study based radiofrequency data analysis in humans, Eur Heart J, 2006;27:655–63.
    Crossref | PubMed
  19. Pu J, Mintz GS, Brilakis ES, et al., In vivo characterization of coronary plaques: novel findings from comparing greyscale and virtual histology intravascular ultrasound and near-infrared spectroscopy, Eur Heart J, 2011. Epub ahead of print.
  20. Heusch G, Kleinbongard P, Bose D, et al., Coronary microembolization: from bedside to bench and back to bedside, Circulation, 2009;120:1822–36.
    Crossref | PubMed
  21. Mauri L, Rogers C, Baim DS, Devices for distal protection during percutaneous coronary revascularization, Circulation, 2006;113:2651–6.
    Crossref | PubMed
  22. Saucedo JF, Mehran R, Dangas G, et al., Long-term clinical events following creatine kinase – myocardial band isoenzyme elevation after successful coronary stenting, J Am Coll Cardiol, 2000;35:1134–41.
    Crossref | PubMed
  23. Prasad A, Singh M, Lerman A, et al., Isolated elevation in troponin T after percutaneous coronary intervention is associated with higher long-term mortality, J Am Coll Cardiol, 2006;48:1765–70.
    Crossref | PubMed
  24. Selvanayagam JB, Porto I, Channon K, et al., Troponin elevation after percutaneous coronary intervention directly represents the extent of irreversible myocardial injury: insights from cardiovascular magnetic resonance imaging, Circulation, 2005;111:1027–32.
    Crossref | PubMed
  25. Mehran R, Dangas G, Mintz GS, et al., Atherosclerotic plaque burden and CK-MB enzyme elevation after coronary interventions: intravascular ultrasound study of 2256 patients, Circulation, 2000;101:604–10.
    Crossref | PubMed
  26. Hong YJ, Mintz GS, Kim SW, et al., Impact of plaque composition on cardiac troponin elevation after percutaneous coronary intervention: an ultrasound analysis, J Am Coll Cardiol Imaging, 2009;2:458–68.
    Crossref | PubMed
  27. Kotani J, Nanto S, Mintz GS, et al., Plaque gruel of atheromatous coronary lesion may contribute to the no-reflow phenomenon in patients with acute coronary syndrome, Circulation, 2002;106:1672–7.
    Crossref | PubMed
  28. Kawamoto T, Okura H, Koyama Y, et al., The relationship between coronary plaque characteristics and small embolic particles during coronary stent implantation, J Am Coll Cardiol, 2007;50:1635–40.
    Crossref | PubMed
  29. Uetani T, Amano T, Ando H, et al., The correlation between lipid volume in the target lesion, measured by integrated backscatter intravascular ultrasound, and post-procedural myocardial infarction in patients with elective stent implantation, Eur Heart J, 2008;29:1714–20.
    Crossref | PubMed
  30. Lee MS, Park SJ, Kandzari DE, et al., Saphenous vein graft intervention, JACC Cardiovasc Interv, 2011;4:831–43.
    Crossref | PubMed
  31. Liu X, Tsujita K, Maehara A, et al., Intravascular ultrasound assessment of the incidence and predictors of edge dissections after drug-eluting stent implantation, JACC Cardiovasc Interv, 2009;2:997–1004.
    Crossref | PubMed
  32. Oemrawsingh PV, Mintz GS, Schalij MJ, et al., Intravascular ultrasound guidance improves angiographic and clinical outcome of stent implantation for long coronary artery stenoses: final results of a randomized comparison with angiographic guidance (TULIP Study), Circulation, 2003;107:62–7.
    Crossref | PubMed
  33. Sakhuja R, Suh WM, Jaffer FA, Jang IK, Residual thrombogenic substrate after rupture of a lipid-rich plaque: possible mechanism of acute stent thrombosis? Circulation, 2010;122:2349–50.
    Crossref | PubMed
  34. Waxman S, Freilich MI, Suter MJ, et al., A case of lipid core plaque progression and rupture at the edge of a coronary stent: elucidating the mechanisms of drug-eluting stent failure, Circ Cardiovasc Interv, 2010;3:193–6.
    Crossref | PubMed
  35. Nissen SE, Application of intravascular ultrasound to characterize coronary artery disease and assess the progression or regression of atherosclerosis, Am J Cardiol, 2002;89:24B–31B.
    Crossref | PubMed
  36. Oyabu J, Ueda Y, Ogasawara N, et al., Angioscopic evaluation of neointima coverage: sirolimus drug-eluting stent versus bare metal stent, Am Heart J, 2006;152:1168–74.
    Crossref | PubMed
  37. Dixon SR, Grines CL, Munir A, et al., Analysis of target lesion length before coronary artery stenting using angiography and near-infrared spectroscopy versus angiography alone, Am J Cardiol, 2012;109;60–6.
    Crossref | PubMed
  38. Stone GW, Maehara A, Lansky AJ, et al., A prospective naturalhistory study of coronary atherosclerosis, N Engl J Med, 2011;364:226–35.
    Crossref | PubMed