Restenosis revisited: Insights provided by quantitative coronary zingiography

Patrick W. Serruys, MD, PhD,a David P. Foley, MB, ChB,a Richard L. Kirkeeide, PhD,b and Spencer B. King III, MD” Rotterdam, The Netherlands, Hbuston, Texas, and Atlanta, Ga.

Since its inception as a specialist discipline, inter- rence rate” of 12% to 55% within 6 months (Fig. 1) ventional cardiology has been and still is preoccupied is already widely accepted, especially because there is with the achievement of a greater understanding and no hard evidence that longevity can be improved. ultimately control of the biologic healing process af- (Only since 1987 have formal, randomized prospec- ter percutaneous transluminal coronary angioplasty tive studies been initiated to compare PTCA and (PTCA),l? 2 the apparent enigma of “restenosis.” In- coronary artery bypass grafting [CABG] in terms troduction and application of a wide variety of alter- of effect on long-term morbidity and mortality; native and adjunctive treatment modalities (endo- thus objective results will not be available for some luminal coronary stent implantation,3 directional, time.) extractional, or rotationa14-6 atherectomy, laser bal- Innovators and exponents of new treatment mo- loon angioplasty,7 and laser angioplasty8T g and a wide dalities may allow their enthusiasm to compromise variety of pharmacologic agents have, thus far, failed their objectivity, which probably explains why many to suppress this ubiquitous processlo-l6 interventionalists armed with the latest device ac- Proceeding with trepidation through the veritable quire the expertise to use it largely through self- jungle of already available and rapidly proliferating training techniques and assess its impact by clinical literature on this subject, it is noteworthy that there observations in the short term, allowing their judg- are perhaps only three observations that are univer- ment to be influenced more by anecdotal experience sally held: (1) the attribution of initiation of percu- than by the results of carefully controlled prospective taneous coronary intervention to Dr. Andreas Gru- clinical studies. The value of all new treatment mo- entzig; (2) that this approach to therapy of coronary dalities must be submitted to-objective and critical artery obstructions has become “widely accepted” assessment through such studies by using the best despite that its inaugural use was only in September available methodological analytical techniques. The 1977r7; and notwithstanding (3) the apparently ines- advent of computer-assisted quantitative coronary capable biologic phenomenon of restenosis (the third angiography18-20 has demonstrated the fallibility of universal tenet). The unbiased reader will consider it traditional visual and user-dependent techniques for surprising that a,treatment modality that was tenta- assessment of the coronary cineangiogram,21 on which tively incorporated into clinical practice marginally most of, if not all, the early reports on restenosis af- more than a decade past and is attended by a “recur- ter angioplasty are based. Tenets founded on the re- sults of early clinical studies must, therefore, be re- examined in the light of new revascularization and From the aDepartment of Interventional Cardiology, Thoraxcentre, Eras- imaging technology, and we must be prepared to mus University, Rotterdam; bDivision of Cardiology, Department of Inter- nal Medicine, University of Texas Health Science Center, Houston; and consider potential changes in basic philosophic and CAndreas Gruentzig Cardiovascular Center, Department of Cardiology, methodological approaches to both the treatment of Emory University School of Medicine, Atlanta. coronary disease and evaluation of outcome after Reprint requests: Patrick W. Serruys, MD, PhD, Erasmus University Ee treatment as a consequence of fresh insights pro- 2332, P.O. Box 1’738, 3000 DR Rotterdam, The Netherlands. A~H~~~~J1993;126:1243-67. vided. Copyright @ 1993 by Mosby-Year Book, Inc. In the following paragraphs we review restenosis 0002~8703/93/$l.b0+ .lO 4/l/49916 from the angiographic point of view because it is

1243 November 1993 1244 Sermys et al. American Heart Journal

HONAN BENGSTON NOBUYOSHI FLECK LEVINE SCHWARTZ HOLMES VANDORMAEL BERTRAND : THORNTON - TIME WTERVAL DE FEYTER UEBIS PTCA-F/U :1-Q MTHS SERRUYS ANGIO F/U: 57-100% MEYER ! CORCOS LE’MGRuBER/iKALTENBACH /

0 5 10 15 20 25 30 35 40 45 50 55 60 RESTENOSIS RATE IN THE LITERATURE (%) Fig. 1. Nonscientific figure depicting restenosisrates from selection of published studies with different angiographic follow-up rates (57%-loo%), follow-up intervals (1-9 months), 11 different restenosiscrite- ria, and various angiographic analysis techniques.

Table I. Angiographic definitions of restenosisthat have vein we briefly outline the main discrepancies in the been usedin various clinical studies. NHLBI 1,2,3, and 4 restenosis literature and discuss the impact of quan- are criteria for angiographic restenosis,as laid out by the titative coronary angiographic studies on under- National Heart, Lung, and Blood Institute of the United standing of the restenosis process. We describe some States new conceptual approaches to the study of resteno- 1. A diameterstenosis ~50% at follow-up.22 sis that we have developed through our own experi- 2. An immediatepost-PTCA diameter stenosis <50% ence in clinical studies, with serial computer-assisted that increases to 150% at follow-up.23,26 quantitative coronary angiographic analysis (in pa- 3. As for 2, but a diameter stenosis 270% at follow-up tients treated by balloon angioplasty, directional (NHLBI 2).24 4. Loss during follow-up of at least 50 510 of the initial gain atherectomy, and stent implantation) and compare at PTCA (NHLBI 4).25 these approaches with findings of other groups. 5. A return to within 10% of the pre-PTCA diameter Through these new approaches we have attempted to stenosis (NHLBI 3).26 bridge the gap from angiography to pathology and 6. Loss ~20% diameter stenosis from post-PTCA to fol- have explored the theoretic relationship between low-up.27 7. Loss 230% diameter stenosis from post-PTCA to fol- vessel wall injury at intervention and subsequent low-up (NHLBI 1).23 restenosis. The ultimate aim of this article is to pro- 8. A diameter stenosis 270% at follow-up.28 pose the use of the approach described here as a kind 9. Area stenosis ~85% at follow-up.2g of unifying strategy to studying the restenosis phe- 10. Loss 21 mm2 in stenosis area from post PTCA to fol- nomenon through coronary angiography to resolve low-up.ss 11. Loss ~0.72 mm in minimal luminal diameter from the currently differing methods. post-PTCA to follow-up.31 12. Loss ~0.5 mm in minimal luminal diameter from RESTENOSIS: UNCERTAINTIES AND DISCREPANCIES post-PTCA to follow-up.32 IN DEFINITION CRITERIA 13. Diameter stenosis >50% at follow-up with >lO% de- The first step to improved understanding of the terioration in diameter stenosis since PTCA of a previ- ously successfully dilated lesion (defined as diameter restenosis phenomenon is an accurate, meaningful, stenosis <50% with a gain of >lO% at PTCA).33 and universally accepted definition. Unfortunately,, at this time no such agreement exists, and there have been at least 13 different definitions based on coro- nary angiographic findings and applied by various through this medium that most research has been clinical investigators attempting to address the prob-, performed and angiography is still the only univer- lem of restenosis through clinical studies in recent sally used objective clinical tool for evaluation of im- years (Table I).22-33 Most of these are arbitrary cat- mediate and long-term results of intervention. In this egoric cut-off points and, although some are based one Volume 126, Number 5 American Heart Journal Serruys et al. 7245

Diameter (mm) 3 RD 2.5

PRE POST 7 d 7 m 2 3 4 5 6 72 PTCA PTCA mths Fig. 2. Graphic representation of MLD and reference diameter (RD) values as reported in studies of Nobuyoshi et a1.32and Serruys et al.31 showing virtually identical time trends during 6-month follow-up. (From Serruys et al. J Intervent Cardiol 1991;4:265-76.) historical physiologic concepts,34 the measurement To circumvent the potential for inaccuracy with used is percent diameter stenosis, which is inherently respect to percent diameter stenosis measurements flawed by the method of its computation, as we de- calcmated by using an arbitrarily selected reference scribe later. Furthermore, a single cut-off point can- segment by the observer, the computer-based Car- not accurately describe what is essentially a “moving diovascular Angiographic Analysis System (CAAS) target” (Fig. 2), as has been independently dem- (described further later) generates the interpolated onstrated almost simultaneously by our group3r reference diameter.20y 44, 45 The contour detection al- and Nobuyoshi et al. 32 By using predetermined seri- gorithm reconstructs how the arterial borders of the al angiographic follow-up at 1, 2, 3, and 4 months segment of interest should appear in the disease-free (Nobuyoshi et al. carried out additional angiography state by the technique of interpolation. According to at 6 and 12 months), both groups showed that some this process, the actual lesion itself (obstructed degree of renarrowing occurs in most dilated lesions region/segment) is excluded by using the curvature and is a time-related phenomenon developing in the analysis, which detects the proximal and distal ends first 4 months after therapy and rarely progressing of the lesion (this process may be less accurate in dif- after 6 months. fusely diseased vessels than where there is a discrete Percent diameter stenosis is traditionally calcu- stenosis). Then, in a continuous fashion, on the basis lated by assuming a normal diameter value for a seg- of the detected contours of the proximal and distal ment of coronary vessel immediately proximal or segments and allowing for anatomic vessel tapering, distal to area of interest as a reference point. This by using a second-degree polynomial function, the assumption has been shown to be erroneous, partic- arterial contours over that segment are interpolated. ularly in the context of multivessel disease, when The measurement taken as the reference diameter, there is virtually always diffuse intimal and/or sub- then, is the interpolated reference diameter (from the intimal thickening35> 36 and variable age-related or so-called diameter function curve) at the site of the compensatory ectasia,37 and after interventions when minimal luminal diameter (Fig. 3). The theoretic ba- the reference ‘diameter becomes involved in the res- sis of and actual mathematic steps involved in this tenosis process. 38-40 In recent years quantitative an- process have been described in detail in technic.al giographic studies have clearly and definitively re- publications in the past, and their intricacies are be- vealed that labsolute luminal measurements such as yond the scope of this article. minimal luminal diameter (MLD) or minimal cross- Fig. 4 ,illustrates the potential, pitfalls of the arbi- sectional area of coronary narrowings provide more trary selection of proximal and/or distal coronary reliable and meaningful information than percent segmente as a reference point and how a more objec- diameter stenosis with regard to hemodynamic sig- tive measurement of percent diameter stenosis can be nificance of an obstructive coronary lesion.41-45 derived by using the interpolated reference diameter. November 1993 1246 Serruys et al. American Heart Journal

d@-. NT~RPCILAT~DREFERENCE DIAMETER

DETERMINED BY CURVATURE ANALYSIS

STANDARD CAAS ANALVSIS Fig. 3. Graphic representation of CAAS measurement of interpolated reference diameter. Actual luminal contour is detected by edge detection technique. Proximal and distal extremities of obstructive lesion are determined from curvature analysis of detected contour; thus identified lesion is then excluded from de- termination of interpolated reference diameter. Second-degree polynomial function is applied to diametric measurements made from each scan line (every 0.1 mm) of segment proximal and distal to lesion; anatomic vessel tapering is taken into consideration and vessel contours in area of lesion are reconstructed and in- terpolated into diameter function curve (shown as dashed line in analysis and corresponding upper and lower vessel contours). Actual interpolated reference diameter used then is diametric measurement, from diameter function curve, at point of minimal luminal diameter, as shown.

Whatis the InterpolatedReference Diameter(lRD)?

XD Sten I 77%DSten USER DEFINED ” USER DEFINED PROX. REF. DIAM. DtST. REF. DIAM.

Mean user-defined=76%DS Fig. 4. Variability in percent diameter stenosis measurements of same lesion as a result of arbitrary selection of reference diameter by “user,” and more objective derivation of percent diameter stenosis by user-independent method of interpolated reference diameter (IRD). For stenosis where minimal luminal diameter is measured at 0.8 mm, if user-defined proximal or distal reference segment or mean of both is selected then resultant measure of percent diameter stenosis for obstructing lesion is 75 % , 77 % , or 76 % , respectively. If computer-determined interpolated reference diameter (shown asupper and lower thick dark lines) is used, diameter stenosis measurement of 66% is obtained. Prox ref diam, Proximal reference di- ameter; dist, distal; %DSten, percent diameter stenosis. (From Serruys P, et al. J Intervent Cardiol 1991;4:265-76.) Volume 126, Number 5 American Heart Journal Serruys et al. 1247

REST RATE (%) 35

3o i, + NHLBI 4

-0 1M 2M 3M 4M 5M A FOLLOW-UP (MTHS)

NUMBER OF LESIONS FULFILLING RESTENOSIS CRITERIA POPULATION = 490 LESIONS

2 50% DS (~~67)

? 0.72 mm B (N=~o) Fig. 5. A, Cumulative incidence of restenosis during 5-month (MTHS) follow-up period as defined by three different criteria among 490 successfully dilated lesions. NHLBI 4, Criterion 4 of National Heart, Lung and Blood Institute of America: loss of >50% of initial gain at angioplasty; 250%) ~50% diameter stenosis (OS) at follow-up; ~0.72 mm = 20.72 mm loss in minimal luminal diameter from after angioplasty to fol- low-up. (From Beatt et al. J Am Co11 Cardiol 1990;15:491-8.) B, Venn diagram showing distribution of le- sions fulfilling the different criteria applied in Fig. 3. Note that, of 153 lesions undergoing restenosis ac- cording to at least 1 of 3 criteria applied, only 43 fulfill all 3. It is appreciated how different definitions of restenosis define separate populations. (From Beatt et al. J Am Co11 Cardiol 1990;15:491-8.)

Nevertheless, even with use of interpolated reference tion to variable angiographic follow-up rates (57 % to diameter, inaccuracy introduced by the presence of 100 % ) and the continued use of visual assessmentof diffuse arterial disease (and by the effect of interven- the coronary angiogram by most investigators, the tion) is not surmounted, and the use of minimal lu- reported incidence of restenosis varies widely (Fig. 1). minal diameter appears more reliable for the purpose Applying three different and widely used definitions of important clinical research, as will be elaborated to a series of 398 lesions serially measured during further in later sections of this article. 6-month follow-up, our group3r demonstrated that: As a consequence of definition inaccuracy, in addi- (1) the greatest single determinant of the restenosis November 1993 1248 Sei-ruys et al. American Heart Journal rate is the choice of definition (Fig. 5, A), and (2) even event.l> 55, 57-5g~61, w 67 From this point in the para- if the eventual incidence of restenosis is similar, dif- digm a new aspect is introduced whereby the fresh ferent definitions identify different patient popula- mural thrombus becomes rapidly endothelialized (3 tions (Fig. 5, B), making risk factor determination to 4 days after injury) and is then infiltrated by (and indeed meaningful study of the natural history mononuclear cells from the luminal surface inward. of the restenosis process) impossible.46 These diver- The degenerating thrombus is subsequently colo- sities are almost certainly responsible for much of the nized by proliferating ol-actin staining cells (becom- confusion surrounding the concept of restenosis after ing visible from day 6 onwards) whose origin and ex- PTCA. act nature remains elusive (although they are possi- bly smooth muscle cells or myofibroblasts), again WHAT IS RESTENOSIS? from the luminal surface inward (in the opposite di- Restenosis refers to the combined biologic (throm- rection to that which is hypothesized in the current bo-rheo-fibro-proliferative) healing processes taking conventional model), with concomitant production place after the mechanical or physicai injury im- of extracellular matrix, eventual complete resorption parted by balloon inflation or other percutaneous of the thrombus, and the ultimate formation of ma- device, which ultimately lead to a progressive renar- ture neointima. rowing of the patent lumen to a greater or lesser de- Karas et a1.68 have also presented a model of res- gree during the immediate weeks and months after tenosis by using balloon dilatation or implantation of the therapeutic procedure. This process of response a balloon-expandable stent in normolipemic domes- to injury has been most extensively studied in the tic swine. They observed marked intimal smooth- context of balloon dilatation and, as such, appears to muscle cell proliferation and an increase in extracel- loosely consist of four elements: (1) Elastic recoil, a lular matrix, destruction of the internal elastica, and natural property of intact blood vessels to respond to thinning of the arterial medial layer, both in stented stretch. This phenomenon has been characterized by and balloon-dilated segments. Intimal proliferation quantitative angiography47-50 and by intravascular was more prominent and significantly greater in ultrasound (IVUS)51; although available data are stented segments by morphometric analysis; in addi- somewhat conflicting, on balance elastic recoil is de- tion, residual luminal area was significantly less in picted as occurring immediately after balloon defla- these segments. These histopathologic changes were tion 47, 4g with no significant further recoil during the described as being comparable with those observed in succeeding 24 hours52a 53 and thus has a doubtful real human restenosis. Furthermore, reactive inflamma- contribution to the process of late luminal renarrow- tory infiltrates were frequently observed in proxim- ing4gJ 54, 52; (2) subintimal platelet deposition, mural ity to the stent filaments, suggesting a foreign body thrombus formation, and consequent organization type of reaction. The authors hypothesize that, may result in rapid luminal obstruction or early res- despite provoking more intimal proliferation, stent- tenosis56 after the apparently unavoidable intimal ing maintains a greater morphometric luminal area disruption induced by balloon dilation@; further- than balloon dilatation through the achievement of a more, these early “hemorrheological responses” to larger lumen by continued mechanical opposition of vessel wall injury57 are likely to be an initial pathway elastic recoil. These findings present circumstantial to (3) fibrocellular neointimal hyperplasia, which is evidence that agrees with the description of a pro- widely regarded as the final pathologic process by portional relationship between the extent of neo- which progressive luminal renarrowing develops in intimal hyperplasia and the severity of vessel wall the months after PTCA1> 58-62; and (4) reaccumula- injury by Schwartz et al. 65 However, the authors dis- tion or acceleration-progression of classic atheroscle- agree with the hypothesis of Schwartz et al.63 regard- rotic plaque.@ ing the central role of thrombus in the process of res- Alternative pathologic partidigms. The conventional tenosis, on the basis of the rarity of angiographic ev- assumption of an outgrowth of proliferating cells idence of large space-occupying clots in clinical (presumably smooth-muscle cells) from the dam- practice, and suggest that excessive arterial injury aged vessel media to form the neointimal has lately was induced in their model, thus limiting its use as a been challenged by Schwartz et al.63 On the ba- screening tool. sis of extensive experience with a domestic swine Further observations of coronary artery dilatation mode1,64$65 they assign a, central role to mural in the porcine overstretch model without stenting thrombus formation at the site of injury, which has provided no evidence for a space-occupying throm- already previously been well recognized as a key bus6g as suggested by Schwartz et al.65 Serial exam- Volume 126, Number 5 American Heart Journal Serruys et al. 1249

ination of the dilated arteries at 1,3,7,14, and 28 days ASSESSMENT OF RESTENOSIS: SYMPTOMS, showed little or no thrombus but active migration FUNCTION, FLOW, OR LUMEN? and proliferation of cells from the area of the dis- Coronary obstructions, whether of primary ather- rupted media. Examination of these healed lesions omatous origin or arising as a response to the con- and similar lesions from patients70 failed to reveal the trolled mechanical or thermal injury inherent in the footprints of preexisting thrombus, that is, hemo- various currently available nonsurgical therapeutic siderin deposits, etc. These divergent findings may coronary interventions, can be described from a primarily relate to the extensive use of stents in the number of different viewpoints: (1) by the symptom- MayoClinicmodel (whichuniformlyproduces throm- atic sequelae. Although improvement in quality of bus) and the overstretch without stenting used in the life (and life expectancy) is the goal of any therapeu- Emory model (which was found not to produce sig- tic modality, it is also the least objective yardstick by nificant thrombus). which to evaluate the impact of treatment.74> 75 Intracoronary fiberoptic angioscopy allows safe (2) By the physiologic disturbance in myocardial and rapid assessment of the site of percutaneous in- perfusion caused. The physiologic effect of a coronary tervention.56 One recent clinical study small and stenosis may be most certainly the ultimate determi- preliminary, reports the presence of luminal throm- nant of management for the individual patient. The bus in two thirds of patients immediately after increasing battery of noninvasive investigations avail- angiographically successful balloon dilatation, with able for this purpose contribute vital information progression in thrombus volume, within 1 hour, in necessary for clinical decision making, but they can- half of these, while contrast angiography failed to not provide the reproducibility and objectivity77-7g provide clear evidence of intraluminal thrombus.71 required to assess the process of luminal renarrowing Further recent technologic advances in angioscopic after percutaneous revascularization in large patient hardware will undoubtedly facilitate even more de- groups in the context of clinical trials. In daily clin- tailed and “microscopic” evaluation of the coronary ical practice it may be considered most prudent to lumen in clinical studies.72 Thus the technology now regard restenosis according to the need for a repeat appears to be available to prospectively study the role revascularization procedure; thus it is important for of intraluminal thrombus in the restenosis process the reader to recognize that this treatise is primarily and in the clinical setting, and such studies may al- focused on the evaluation of outcome after interven- ready be underway in some institutions. tions in the context of large clinical trials of new de- Whatever the exact nature of the histopathologic vices or therapeutic approaches rather than as a di- process, it is evident that the corresponding angio- rect guide to general clinical practice. Nevertheless, graphic appearance of progressive luminal renarrow- it must not be forgotten that conclusions drawn from ing is not well encapsulated by traditional conven- the outcome of such trials provide us with the type of tional definitions or criteria such as “loss of greater information that is ultimately instrumental in our than 50 % of the gain” or “greater than 50 % diame- daily clinical decision making, hence the paramount ter stenosis at follow-up,” two of the most widely used importance of design, method, and approach in these definitions in daily clinical practice, and in resteno- trials and studies. sis prevention studies (Table I) and trials of new de- (3) By its hemodynamic consequences, that is, re- vices. Schwartz et al.65 and Beatt et a1.73 have each duction in coronary blood flow caused by the ob- highlighted this inherent limitation of conventional struction. Regional coronary flow reserve (CFR) (de- restenosis criteria and the consequent potential for fined as the ratio of maximal to resting coronary misinterpretation of results of clinical trials relying blood flow) may be directly measured at cardiac on such criteria. Similarly, Karas et al.68 also draw catheterization by a number of different techniques, attention to the possibility of considerable intimal or alternatively may be derived from values obtained proliferation without angiographic restenosis. by quantitative angiography.42 Absolute coronary Because current understanding of the restenosis blood flow itself can be measured by Doppler wire, process is at best fragmented, until our knowledge and transstenotic pressure gradients may be ob- becomes more complete, it is crucial to make sensi- tained by using ultra-thin fiberoptic or fluid-filled ble use of the most objective descriptive methodolog- catheters. Further sophisticated and precise hemo- ical approaches available, especially in the minefield dynamic measurements can be obtained during car- of clinical research. The question is, what are the diac catheterization. However, these are mainly re- most objective, reliable, and reproducible methods of search tools that are not at this time generally avail- assessing restenosis? able to the general nonacademic interventionist and November1993 1,250 Serruys et al. American Heart Journal are unapplicable for m’ulticenter clinical studies in a series of arbitrary centerpoints selected by the an- large patient populations. alyst, but for which the algorithm can retroactively (4) By its anatomic configuration, namely degree correct and regenerate the true centerline of the seg- of luminal narrowing. For the present, this aspect is ment image. Absolute diameter measurements are best assessed by conventional contrast angiographic determined in millimeters by using the outer border techniques, although intravascular ultrasound of the. contrast-free angiographic catheterg5 (the dis- (IVUS) is emerging as an exciting and promising im- tal 20 cm of each individual catheter used for contrast aging modality3s 51~s5 and will undoubtedly have a injection is retained and measured by micrometer) as useful application in the future in this area. As yet, a scaling or calibration device. A correction factor is IVUS is in the developmental stage and many tech- then introduced for the so-called pincushion distor- nical obstacles need to be surmounted, particularly tion introduced by the image intensifier (there are transducer size (models used in the most recently 557 different pincushion correction factors in the published studies are 1.83 mm in diameter, although CAAS database relating to all the catheterization a 1.15 mm prototype is being tested,36 cost, ethical laboratories from which we receive cinefilms for considerations, and objective delineation of the rela- analysis). Pincushion distortion is minimal with tionship between IVUS images and actual morphol- modern angiographic systems, and this step may not ogy of the blood vessel wa1Jg5 before it can be consid- be necessary in the coming years. The interpolated ered as a realistic or practical alternative to carefully reference diameter is obtained as previously out- controlled angiography. lined. From the absolute measurements (minimal lu- At this time the coronary cineangiogram is still the minal diameter, maximal luminal diameter, mean only universally available imaging modality for ex- luminal diameter, and lesion length) and the inter- amination of coronary anatomy; quantitative angio- polated measurements obtained by the computer graphic techniques as described later have emerged (symmetry, curvature, inflow-outflow angle, plaque as the gold standard for the accurate and objective area, roughness), many others may be derived, such analysis and description of the basic cineangiogram, as percent diameter stenosis, percent area stenosis, particularly in the context of large, multicenter res- theoretic transstenotic pressure gradient, calculated tenosis prevention clinical studies and trials of per- Poiseuille resistance, and calculated turbulence re- cutaneous coronary revascularization devices. sistance. Coronary luminal measurement by quantitative coro- Videodensitometry assesses the area of a stenosis nary angiography: The CAAS approach. Quantitative by comparing the density of contrast in the diseased coronary angiography (QCA) has been used at our and normal segment, which has the advantage that institution for a decade” and is becoming increas- only a single angiographic projection is required. ing available with the development of on-line quan- However, this must be perfectly perpendicular to the titative angiographic computer software for digi- long axis of the vessel, and there must be no overlap- tal cineangiographic imaging (DCI) equipment ping or closely parallel sidebranches or other dis- in the catheterization laboratory. The CAAS turbing radiopacque structures. Only relative values system has been rigorously and extensively vali- are provided so data obtained by contour detection dated, and the methodology is described in detail are also necessary to provide absolute measurements. elsewhere.1s-20~ 87s88g g5-g7 To explain the method A further drawback of this otherwise promising briefly: After a selected cineangiographic image has technique is its high sensitivity (it is even more sen- been converted to an optically magnified digital im- sitive than contour detection) to x-ray scatter, veiling age (digitization; this step may soon be rendered un- glare, beam hardening, and suboptimal contrast fill- necessary by the rapid development of on-line digi- ing of the vessels. These basic limitations have tal systems), the contours of the selected coronary deterred the application of the technique to impor- segment are detected automatically (so called con- tant clinical angiographic studies. In the latest up- tour or edge detection) by the computer algorithm dated CAAS system, however, steps have been taken applied to the brightness profile of the segment along to .overcome these drawbacks; results of validation scan lines that are perpendicular to the segment cen- studies are eagerly awaited. terline (scan lines are made every 0.1 mm the Minimal luminal diameter. Of all the measurements approximate size of 1 pixel). The centerline is deter- acquired by quantitative angiography, the absolute mined by the computer, which uses as starting points value of the minimal luminal diameter (MLD) has been shown to be the greatest single determinant of *18-20,31,38,42-48,86-108,111,112,115-120,127,135,136. the hemodynamic consequences of a stenosis because Volume 126, Number 5 American Heart Journal Serruys et al. 7257

this parameter affects blood flow by a fourth power ity cannot be inferred from the original study because term.42T 8o, 84ylog It is therefore the most unambigu- no intervention was carried out. ous, objective, and reproducible parameter to use for We have completed a pilot study53 that provides primary measurement of coronary arterial (or bypass clear angiographic data in this regard, whereby graft) luminal caliber and changes therein resulting among 110 lesions (mean vessel size of 2.67 mm) from interventions, that is, the angiographic gain in studied after balloon angioplasty and at 24 hours minimal luminal diameter after intervention and (under optimally standardized conditions, i.e., subsequent loss during follow-up and, as described matched angiographic projections, intracoronary ni- later, the proportional angiographic gain and loss (so trate before angiography, full therapeutic anticoagu- called relative gain and relative loss which normalize lation) there was no difference in MLD (0.007 mm, the gain and loss for the actual vessel size. p = 0.79), and the SD of the mean difference was 0.2 Criteria for definition of restenosis on the basis of mm. By extrapolation from these data, it can be con- change, in minimal luminat diameter during follow-up. cluded that the post-PTCA lesion measurement Important multicenter trials examining the impact of variability of the CAAS system is 0.2 mm; thus a various treatment strategies on restenosis have in re- change of 0.4 mm in MLD can be considered with cent years been availing of central, standardized, 95 % confidence to represent a real change and thus blinded, computer-assisted, quantitative angio- can be considered a potential criterion for detection graphic analysis in angiographic core laborato- of significant luminal loss or renarrowing. In the light riesg8g gg, no-l14 and have begun to use the minimal of the well-recognized difficulties of angiographic in- luminal diameter (MLD) as the most objective and terpretation of the postballoon angioplasty result, we useful measurement.l15 In the past, our group used believe this measurement variability to be eminently the long-term minimal luminal diameter measure- acceptable. ment variability of the CAAS system as a means of It must be noted that the criterion we propose for identifying lesions undergoing significant or detect- detecting significant change in MLD over a period of able luminal change during follow-up after balloon time is the measurement variability of the analytic angioplasty. 88 The SD of the mean difference be- system being used and not 0.4 mm per se, because this tween MLD measurements of the same lesions at is the variability of the CAAS system and may not be different points in time where no intervention was relevant to other systems. Ultimately, as alluded to carried out was measured under a worst-case scenario already and as discussed further later in this article, and found to be 0.36 mm. Two SDS would identify the application of dichotomous criteria to the de- with 95% confidence lesions undergoing a real, de- scription of long-term outcome after intervention is tectable, or significant change. By using the long- fraught with imprecision, conflict, controversy, and term lesion measurement variability it was believed dissension. might present an objective approach to dividing pa- tients monitored after PTCA into restenosis and NEW INSIGHT TO THE RESTENOSIS PROCESS nonrestenosis. FAClLlTATED BY QCA That study may now be considered somewhat ob- Gaussian distribution. Although most biologic phe- solete because, as highlighted by Ellis and Muller,l16 nomena are distributed in nature in normal or gaus- and as we ourselves had already recognized, a large sian fashion, the outcome of percutaneous coronary number of limitations to this deffinition of restenosis interventions and of pharmacologic restenosis pre- are now evident. First, the developmental study used vention trials have been traditionally assessed up to vessels with an average reference diameter of 3.7 now by using categoric cut-off criteria for the occur- mm,88 whereas in two recent large multicenter rest- rence (or not) of restenosis because, as previously enosis prevention studies the mean reference diam- mentioned, clinical decision making is ultimately a eter of treated vessels was 2.6 rnrn.llly 112 Further- binary process. However, for the purposes of large more, the initial study used a worst case scenario multicenter controlled clinical trials, estimation of whereas extensive standardization measures (use of sample size required to demonstrate the statistical intracoronary nitrates to control vasomotor tone, significance of a treatment effect has been based on performance of angiography in exactly matched mul- the assumption of a gaussian distribution for the loss tiple projections, careful identification and selection in MLD during follow-up after balloon angioplasty.l18 of an end-diastolic tine frame for QCA analysis, etc.) That this discussion is not merely a matter of are now carried out in modern multicenter studies semantics is illustrated in the example shown in Fig. and furthermore post-PTCA measurement variabil- 6.118 A significant beneficial treatment effect is de- November 1993 1252 Serruys et al. American Heart Journal

0.72 mm NO RESTENOSIS RW

“PLACEBO” m RESTENOSIS LOSS MLD m .40

n MLDimm)

“ACTIVE TREATMENT’ &$ij RESTENOSIS LOSS MLD = .25

n MLDlmm) I Fig. 6. Gaussianmodel of restenosisin reference and treatment groups. Lower curve represents30 % re- duction in minimal luminal diameter changeat follow-up (-0.25 mm vs -0.40 mm) in treated group, upper curve denotesdistribution of changein minimal luminal diameter (AMLD) found at follow-up in prospec- tive study at our institution. 31If changeof 0.72 mm is taken as cut-off point for restenosis,this categoric model would require 620 patients per group to have power of 90% . (From Serruys PW et al. Interventional cardiology. Stuttgart: Hogrefe and Huber, 1990.) fined as a reduction in the mean loss in minimal lu- corroborated these early reports, demonstrating sim- minal diameter during follow-up after balloon angi- ilar distribution patterns of luminal change after in- oplasty by 30 % , that is, from 0.40 mm to 0.25 mm. It tervention in patients undergoing directional athe- can be calculated that, assuming normal distribu- rectomy and stent implantation.122s 124 tions for the loss in MLD, 233 patients are required At a later date, however, the Emory group exam- in each treatment group to demonstrate the signifi- ined, by clinical estimation of percent diameter cance of this difference at the 95% confidence level stenosis, a large cohort of patients undergoing bal- with a power of 90 % . If a categoric approach (rest- loon angioplasty and follow-up angiography. They enosis yes/no) is used (applying loss in MLD of 10.72 found a bimodal distribution and concluded that mm as the criterion of detectable significant loss), there was either a physiologic bimodal distribution or then the 0.15 mm difference in MLD is equivalent to a systematic measurement error around the 50% a reduction in the restenosis rate from 25 % to 17.5 % ; diameter stenosis mark when clinically evaluating to demonstrate that this difference is significant, 620 cineangiograms. lz5 This finding, if confirmed, would patients will be required in each group, almost three therefore justify a categoric approach to the assess- times as many as required if a continuous ‘approach ment of angiographic outcome in clinical trials, is applied. thereby challenging the emerging assumptions of a Historically, in a number of clinical studies focus- continuous distribution arising from the separate ing on various aspects of the restenosis problem Be- findings of our group and the Beth Israel group. This att et a1.s8>lig-r2i demonstrated that quantitatively prompted our group to reinvestigate this phenome- measured changes in MLD and in reference diame- non in a much larger patient ,population than had ter during the months after PTCA are normally dis- been studied in the original studies,“8p iLg-r2i and un- tributed.’ This view of a continuous unimodal distri- der more standardized and ,&onsistent quantitative bution for luminal change after balloon angioplasty, angiographic conditions. In this study of 1234 #pa- although strongly challenged at that time, became tients it was demonstrated unequivocally that lumi- the nidus of our philosophy regarding methodologi- nal renarrowing after PTCA, whether assessed by cal approach to addressing the problem of resteno- using minimal luminal diameter or percent diameter sis.* The Beth Israel group subsequently strongly stenosis at follow-up or the change in these measure- *47,48, 54, 73, 94, 100-108, 111, 112, 118. ments during follow-up, clearly follows a gaussian or Volume 126, Number 5 American Heart Journal Serruys et al. 1253

DlSTRlBUTION of CHANGE ,in MLD from POST-PTCA to 6 MTHS F-UP

NO OF LESIONS 1OOr

60

-3 -2 -1 0 1 2 CHANGE IN MLD (mm) DF”,., Fig. 7. Histogram of change in minimal luminal diameter during 6-month (MTHS) follow-up (F-UP) of 1445 primary lesions treated by coronary balloon angioplasty during two large restenosis prevention trials. Theoretic gaussian curve, given mean and SD, is superimposed, clearly illustrating that luminal renarrow- ing is normally distributed phenomenon. (From Rensing et al. J Am Co11 Cardiol 1992;19:939-45.) normal distribution (Fig. 7g4) in agreement with our ing cumulative distribution curves displaying change own earlier findings,389 73 and the reports by Kuntz et in MLD during follow-up for treated versus placebo a1.124in patients treated by other devices. These cor- populations (Fig. 8)l11p 1121124 or indeed for PTCA roboratory findings appear to identify a basic flaw in versus stent or atherectomy13, lo47lo6 as discussed the clinical impression of a bimodal phenomenon. On later. this basis, it is our contention that a dichotomous Potential pitfalls of angiographic studies. Having laid view of restenosis is inappropriate and that categor- out this scheme of the angiographic representation of ically generated restenosis rates should no longer be the phenomenon of restenosis, it must be taken into the main focus of important scientific studies or dis- consideration that the entire hypothesis of a Gaus- course in this vital area. sian phenomenon hinges more or less on the accuracy There may be sound clinical reasons for selecting and reproducibility of quantitative angiographic particular angiograpbic definitions of restenosis, but measurements. Although it seems clear from the in the context of scientific studies or restenosis pre- foregoing evidence that luminal renarrowing or res- vention trials, the use of a blanket, categoric cut-off tenosis after interventions is a continuous or nor- point (e.g., X0% diameter stenosis) conveys no mally distributed phenomenon, it must be recognized measure of the extent of luminal renarrowing and that our observations do not always give a clear view therefore cannot provide a comprehensive assess- of reality. The measurement approaches used to ment of the effect of a particular therapeutic ap- quantify coronary luminal dimensions ‘from cinean- proach for the control of the biologic process of res- giograms have inherent associated measnrement vari- tenosis. Furthermore, because the threshold level for ability that may be attributed to a large number of absolute (or relative) luminal renarrowing that is causes. The importance of knowing this measure- physiologically or clinically significant is unknown, it ment variability for the purposes: of comparing the is much more realistic and meaningful (and requires results of studies carried out by using different mea- much fewer patientsI”) to study the overall effects of surement systems is self-evident. It could be hypoth- an intervention in terms of the mean change in min- esized that this measurement variability may be imal luminal diameter for the entire group.rlrp r12 We clouding our view of reality. believe, furthermore, that results of intervention tri- To demonstrate this point, let us construct a als may be simply presented in graphic form by us- hypothetical scenario whereby restenosis is in reality November 1993 1254 Serruys et al. American Heart Journal

Cumulative frequency % 100

MLD (mm) I I I I 0.0 0.5 1.0 1.5 210 215 310 Fig. 8. Cumulative frequency (distribution) curves of changein minimal luminal diameter from after an- gioplasty to follow-up (F-UP) in control and treatment groups for CARPORT restenosisprevention trial showing no treatment benefit. CUM%. Cumulative percentageof patients. (From Serruys et al. Circula- tion 1991;84:1568-81.)

a discrete disease process, so that by using a perfectly sisted image interpretation may not be exactly accu- accurate and precise quantitative angiographic mea- rate and reproducible. To account for such measure- surement system and plotting the change in minimal ment problems, panel B shows two solid gaussian luminal diameter in a frequency distribution plot two curves which were used to represent the statistical groups of patients could be delineated, one in which characteristics of two QCA measurement systems. restenosis occurred and the other in which no rest- Curve h denotes a highly precise and accurate system enosis developed (Fig. 9). In panel A, the no resteno- (mean difference between repeated measurements of sis group is denoted by the curve n, a relatively nar- MLD = 0 mm, SD of the mean difference = 0.1 mm); row gaussian function showing no overall loss in MLD curve m is a medium-precision but highly accurate (mean = 0, SD = 0.2 mm). The restenosis subgroup system (mean difference = 0 mm, SD = 0.3 mm). response curve r was also taken to be Gaussian but Measuring MLD loss by these systems is akin to broader (more variable) than the normal response, mathematically convolving the study population in A and it displayed a significant loss in MLD (mean = 0.5 with a measurement characteristic from B, the re- mm, SD = 0.4 mm). The difference in the assumed sults of which are shown in panels C and D. In panel variability of these groups merits an explanation. In C the combined population response (solid curve in the no restenosis group the variability in the mea- A) has been measured with the high-precision QCA sured loss in MLD was assumed to stem from uncon- system, resulting in observations that closely mirror trollable changes in lesion tone between the post- the true population response. Panel D shows the ob- PTCA and follow-up angiograms. Such variance was servations (solid heavy curve) resulting from less also assumed for the lesions undergoing restenosis precise measurement of the study population by but, in addition, the process of restenosis was also means of the medium-precision QCA measurement assumed to be variable, thereby exaggerating the system. The measurement process has blurred our variability in loss of MLD in this group. When the observations to the point that the measured loss in subgroup responses are added, the combined study MLD now approaches a gaussian distribution (the population is seen to be unimodal but skewed toward light-dashed line inD), which has the same mean, SD, the right as demonstrated by the heavy solid curve and number of observations. (panel A). This population response (the actual The above scenario proposes an alternative expla- response) must now be measured by a QCA system to nation for the observed gaussian distribution of COT- form our observations. This system, however, is not onary luminal measurements at.follow-up after cq- perfect because frame selection and computer-as- onary interventions and the change in dimensiorjs Volume 126, Number 5 American Heart Journal Serruys et al.

Simulated Study Response Measurement Systems

120

100 1 Em h c. I SW z 40 20 0 0.00 - -2.00 -1 .oo 0.00 1.00 2.00 -2.00 -‘I .oo 0.00 1.00 2.00 MID Loss Imml Differencs in repeat MLD measurements lmml

High Precision Observations I D Medium Precision Observations 120 100 t 80 5g 60. RY., - -.< f 40 ,/ .I \\ 20 1 .\. .

0 -2.00 -1 .oo 0.00 1 .oo 2.00 -2.00 -1 .oo 0.00 1 .oo 2.00 MU) Loss Imml MLD Loss lmml

Fig. 9. Simulated PTCA restenosisstudy. A, solid curve representsentire study population composedof two separatebut overlapping subgroups,both of which can be describedby gaussianfunction. Curve n de- scribesloss in MLD for 1000lesions that did not restenose(mean loss= 0 mm, SD = 0.2 mm). Curve r, loss among 500 lesionsthat did undergo restenosis(mean loss = 0.5 mm, SD = 0.4 mm). B, Measurement char- acteristics of two separateQCA systems, high accuracy and precision system (curve h, with accuracy of 0 mm and precision of 0.1 mm) and high accuracy but medium precision system (cut-w m, with accuracy = 0 mm, precision = 0.3 mm). Heavy curves in C and D display distributions of observations that would be produced by measuringactual lossin MLD in populations of lesionsin A by high-precisionsystem (C) and medium precision system (P) by using simulation processcalled curve convolving. Dashed curves in panels C and D show correspondinggaussian distributions. during follow-up (as reported by our group38, 73sg4y 121 endpoints. If, for example, clinical endpoints were and by the Beth Israel group,lz2p 124)namely impre- used to evaluate restenosis, we can only wonder how cise measurements of a mixed population of lesion blurred the observations might be as a result of inac- responses to intervention. However, without further cura.te or imprecise assignments of clinical events to objective investigation this proposal has no more va- lesion restenosis. Additionally, sensitive and flexible lidity than any other reasonable hypothesis that fits analyses of the observed data need to be explored. In the observations. this regard it might be worthwhile to use our knowl- The point is that it is important be careful in what edge of the measurement process to try to unblur our is concluded or excluded based on observations. Ac- observations. Deconvolution, curve-fitting, and cepting that there may be multiple explanations for curve-stripping procedures are often used in signal a given set of observations, is there anything that can and image processing to remove known artifacts, re- be done to sharpen our view of reality? We should duce noise, and separate overlapping phenomena. It make all reasonable efforts to minimize measurement may prove interesting to see whether such methods inaccuracy and imprecision. As shown in our exam- could be applied usefully in analysis of restenosis ple, even an imprecision of aO.3 mm in measuring trials. MLD loss may significantly contort the observations Clinical correlations of measured MLD. Danchin et and thus confound the conclusions drawn. Although al.126demonstrated a correlation between the thresh- this simulation featured QCA measurements as study old to exercise-induced myocardial ischemia (as dem- endpoints, the principles are applicable to all chosen onstrated by thallium-201 tomoscintigraphy) and an November 1993 1266 Serruys et al. American Heart Journal

% Angina pectgris % + Exercise test

go- ao- 70- 60. 50-

AO 0.5 1 1.5 2 2.5 3 B-O 0.5 ;LDl.;mm; 2.5 3 MLD (mm)

*-* SENSITIVITY - SPECIFICITY Fig. 10. A, Percentage correct classification of recurrence of angina (sensitivity) and absence of angina (specificity) as function of absolute value obtained for MLD by QCA at follow-up after PTCA in 350 pa- tients taking part in restenosis prevention trial. lrl Intersection of two curves represents cut-off point with greatest diagnostic accuracy. (From Rensing BJ et al. J Am Co11 Cardiol 1993;21:317-24.) B, Percentage correct classification of a’positive exercise test result (horizontal or down-sloping ST-segment depression . of >1 mm measured by callipers 80 msec after the J-point) and negative exercise test result (specificity) as function of absolute value obtained for MLD by QCA at follow-up in same patient population as for Fig. IO, A. Intersection of two curves represents cut-off point with greatest diagnostic accuracy.

absolute value for minimal luminal diameter from the sensitivity and specificity curves were again vir- which they conclude that an MLD r2 mm is suffi- tually identical for recurrence of angina and a posi- cient to provide freedom from myocardial ischemia in tive exercise test, at 1.38 mm in vessels <2.63 mm and 95% patients. Furthermore, Rensing et al.lu have 1.58 mm in vessels >2.63 mm in diameter. Thus it is reported that in 350 patients who underwent suc- clear that the vessel size does influence the minimal cessful PTCA for single-vessel disease (as part of a luminal diameter threshold for recurrence of angina large prospective multicenter restenosis prevention or exercise-inducible ischemia. Nevertheless, the ob- trial) and had exercise testing and repeat coronary servations suggest that the absolute value for MLD angiography at follow-up an MLD of 1.45 mm corre- at follow-up may ultimately prove to be a simple and lates with the threshold for recurrence of angina pec- clinically useful parameter both for scientific studies toris (sensitivity and specificity: 72%).127 Exercise- and practical clinical patient management, a claim induced ST-segment change was found to be a less that deserves further and more objective evaluation. reliable predictor of luminal renarrowing, although This implication supports the approach used in the point of greatest diagnostic accuracy for a posi- reporting two recent European multicenter resteno- tive exercise test corresponded with a measured sis prevention trialslrr> lr2 and that employed by MLD of 1.46 mm (Fig. 10). This information is other groups who have consistently focussed on the somewhat surprising because it would be expected MLD at follow up in reporting on angiographic out- that a large number of additional variables should come, in patients treated by DCA and stent implan- influence the relationship between minimal luminal tation 122-124,12%131 diameter and exercise-induced angina or ST-seg- As a measure of the extent of the hyperplastic ment depression such as vessel size, extent of myo- healing process itself, the change in MLD during fol- cardium supplied, viability of myocardial tissue, low-up is clearly the parameter of-choice. Rensing et presence of collateral circulation; use of antianginal al.rz7 also investigated the clinical value of measured medication, etc. When vessel size was taken into ac- change in MLD during follow-up in predicting the count by dividing the study group in half according physiologic significance of treated lesions 6 months to the median v&se1 size,. the point of intersectionof after sucdessful balloon angioplasty and:fou.nd it to Volume 126, Number 5 American Heart Journal Serruys et al. 1257

be only slightly less accurate than absolute MLD at sonably be used. There are three basic principles: (1) follow-up, a deterioration of >0.30 mm yielding sen- the angiographic dimensions of the matched lesions sitivity and specificity of almost 70% for prediction are assumed to be identical; (2) the observed differ- of recurrence of angina and almost 60 % for a positive ence between the two identical lesions must be within exercise test result. Corresponding values for per- range of reproducibility of the computer analysis centage diameter stenosis measurements are pro- system being used (for the CAAS system this is +-0.1 vided in this study for comparative purposes for the mm, i.e., 1 SD of the difference between repeated benefit of clinicians; however, for the extensive rea- measurements of the same angiogram); and (3) the sons given earlier in this article we discourage the use reference diameter of the vessels to be matched are of percent diameter stenosis in important interven- selected within a range of +-3 SD (0.3 mm), giving tional studies. confidence limits of 99 % .I32 lo4 Comparing the immediate angiographic results of COMPARATIVE ASSESSMENT OF NEW DEVICES PTCA, directional coronary atherectomy (DCA), and USING MLD AS THE CENTRAL MEASUREMENT intracoronary stenting with this technique illustrates New dilemmas have arisen as a result of the explo- that both DCA and stenting yield a more favorable sion of new interventional coronary treatments with early result than PTCA and that matching is a use- respect to comparison of results, particularly long- ful comparative method. lo3 Application of the match- term outcome. The unique mechanisms of action of ing principles to a direct comparison of immediate and subsequent pathophysiologic responses to the and long-term angiographic outcome after PTCA various devices renders broad comparison of these and DCA or stent implantation using cumulative treatment modalities basically invalid, especially be- distribution curves (Figs. 11 and 12),lo3s 135 is simi- cause it is generally recommended that atherectomy larly rewarding in its clarity and simplicity. Because devices should not be used, nor should endoluminal the lesions are matched for reference diameter, stents be, implanted, in coronary vessels <3 mm in approximate overall improvement in luminal diame- diameter. PTCA., however, can be (and is regularly) ter (gain) at intervention and loss in minimal lumi- carried out in arteries <2 mm in size, and rotational nal diameter during follow-up can be easily gleaned atherectomy and excimer laser therapy are best from the figure and directly compared. It is appreci- suited to smaller vessels.132 ated that although DCA is associated with a signifi- We consider that it may be reasonable to compare cantly greater initial gain (improvement) in MLD, the effects of interventions in terms of their relative the loss (restenosis) after DCA is also significantly merits or by confining comparisons to matched greater than that after PTCA, so that the ultimate lesions, i.e., lesions of similar severity, in vessel seg- outcome (MLD at follow-up) is similar for both ments of identical size and location (even though it treatment modalities. This technique to compare has lately been demonstrated by our grouploo and immediate and long-term angiographic results after others133 that, contrary to popular belief, restenosis PTCA and self-expanding stainless steel stent im- rates are not significantly different throughout the plantation in 93 matched lesions has revealed that, coronary tree). .At the Thoraxcenter, facilities are although associated with a greater loss in luminal di- available for the appropriate use of all of these ther- ameter during follow-up, stenting yields a signifi- apeutic techniques. The increasingly widespread ap- cantly larger vessel lumen (reflected by a larger plication of these devices, despite the lack of any hard MLD) than PTCA at follow-up.io4 evidence of greater long-term clinical benefit than The matching process, by its principles, may be balloon angioplasty, has prompted our search for a justifiably used at this time as a surrogate for unifying descriptive approach to the assessment and randomized studies,lo4 facilitating otherwise invalid comparison of immediate and long-term outcome comparisons between interventions in relatively small between devices and has led to the serial develop- patient groups. It is noteworthy that observations ment of two methods of comparing the theoretically emerging from the matching of patients undergoing incomparable. DCA and PTCA106 have been confirmed by prelim- Matching: A temporary but convenient surrogate for inary results of the CAVEAT trial,l13 thus demon- randomization. The first method enables us to actually strating a real and undeniable clinical use for this compare the comparable by matching the lesions in matching approach. Furthermore, superior angio- each treatment group for severity, location, and ves- graphic results in terms of MLD at follow-up, of sel size, thereby defining a population in which any DCA, and of stenting over historical PTCA results as of the treatment modalities to be compared may rea- reported by Kuntz et a1,1223124 are put in a slightly November1993 l$j8 Serruys et al. American Heart .kwrnal

0 0.6 1 1.6 2 2.6 3 3.6 4 0 0.6 1 1.6 2 2.6 3 3.6 4 Minimal iuminai diamety (mm) Minimal iuminai diameter (mm) Fig. 11. Cumulative frequency (distribution) curves to illustrate differential immediate (pre to post) and follow-up (post to follow-up) effects of PTCA versus directional coronary atherectomy (DCA) on matched coronary lesions with regard to absolute MLD measured by QCA. (From Umans VA et al. J Am Co11 Car- diol 1993;21:1382-90.)

90 - st 80 I pm 70-- 9tPW

-0 0.6 1 l.6 2 2.6 3 3.6 4 0 0.6 1 l.6 2 2.6 3 3.6 4 Minimal iuminai diameter (mm) Minimal iuminai diameter (mm) Fig. 12. Immediate (pre to post) and long-term (post to follow-up) angiographic outcome of matched study in 93 patients having balloon angioplasty @A) or self-expanding stainless steel stent (St) implantation. Superior initial gain by stenting is somewhat counterbalanced by greater luminal loss during follow-up @up). Nevertheless, luminal diameter at follow-up remains significantly greater than for balloon angioplasty. (From de Jaegere P. J Am Co11 Cardiol 1992;19(suppl):277A.)

different perspective by results obtained from match- in the matching process; therefore the effect of angi- ing. It is clear that the mean vessel size in patients nal status, medication, diabetes, lesion length, ec- treated by DCA and stent implantation are consid- centricity, calcification, etc. on the comparative out- erably greater (3.09 mm and 3.35 mm, respective- come of the treatment modalities is ignored. How- 1~~~~)than in PTCA studies (2.6 mm)111,112.There- ever, the matching study of stent implantation and fore direct comparison of absolute angiographic re- balloon angioplasty lo4 addressed this issue of poten- sults obtained by these devices with those obtained tial disparity between patient groups with regard to by balloon angioplasty becomes somewhat irrelevant these other variables and found no significant differ- without either matching or normalizing for the indi- ences in their distribution between the groups being vidual vessel size, as described in the next section. compared. Furthermore, the matching comparison of The limitations of the basic matching approach to balloon angioplasty with directional atherectomylo6 the comparison of interventianal therapies are, of also took account of age, gender, diabetes, and angi- course, that other potentially influential clinical and nal status and found that this additional consider- angiographic parameters are not taken into account ation did not affect the ultimate findings as already Volume 126, Number 5 American Heart Journal Serruys et al. 1259

RELATIVE LOSS

-. -0.2 0.1 0.4 0.7 1.0

RELATIVE GAIN Fig. 13. Scatter-histogram of all values obtained for relative gain after PTCA and relative loss (RL) dur- ing follow-up in large European multicenter restenosis prevention trial ‘I2 for both placebo and treatment groups. Linear relationship identical for both groups emerges, although coefficient of correlation is low at 0.4. Full line, control group; dashed line, treatment group.

described,lo7 thus perhaps vindicating the applica- sis process, respectively, 38,3g, 4o although there was no tion of the simple basic angiographic matching prin- statistically significant difference in interpol’ated ciples used. reference diameter between pre- and post-PTCA and Relative gain and relative loss in minimal luminal at follow-up in the two previously mentioned Euro- diameter. The second proposed method of compari- pean multicenter restenosis prevention trials.rll, II2 son of therapeutic devices arose originally from the Kuntz et a1.123previously presented a relationship need to create some type of sliding-scale criteria to between absolute gain at intervention and late loss circumvent the previously described limitations of during follow-up in their patients treated by direc- the categoric loss ~0.72 mm criterion for assessing tional atherectomy and stent implantation. However, the outcome of balloon angioplasty in vessels of dif- as a result of the wide variability in reference vessel ferent sizes. The concepts of relative gain and relative size among lesions treatable by current interven- loss in MLD were therefore introduced to adjust lu- tional devices, we believe the application of relative minal changes for individual vessel size by normaliz- gain and relative loss to be more appropriate and in- ing the absolute change in ML.D after intervention formative for comparative purposes. By using data and during follow-up for the reference diameter of accumulated prospectively during each of these res- the coronary segment in question54>73$ 1oo-108 in a tenosis prevention trials, we plotted the relative gain continuous approach. The net difference between and relative loss values for all treated lesions and relative gain and relative loss is termed the net gain found a direct linear relationship between relative index and is a measure of the ultimate net benefit of gain and relative loss (even though the coefficient of intervention. These simple calculations may be pre- correlation was low at 0.4) for each patient popula- sented as follows: Relative gain = MLD post inter- tion, with virtual superimposition of the regression vention - MLD preintervention/vessel size; Relative lines for placebo and treatment groups in each trial loss = MLD post intervention - MLD at follow-up/ (one of these trials is shown in Fig. 13; the other is vessel size; and Net gain index = MLD at follow- practically identical). These graphic representations up - MLD pre intervention/vessel size. confirmed the outcome of the studies with regard to The vessel size is represented by the interpolated there being no demonstrable benefit of the agent un- reference diameter of the lesion before intervention der evaluation in terms of reduction in the loss in because this is the closest possible objective angio- MLD during follow-up, as had been previously es- graphic approximation of the normal, disease-free tablished by using cumulative distribution curves. vessel size. After intervention and at follow-up the Perhiaps more importantly, however, was the rela- interp,olated reference diameter is subject to greater tionship between relative gain and relative loss, potential for measurement variability as a direct which is not dissimilar from the relationship:demon- consequence of the intervention and of the resteno- strated by Schwartz et ah6 between depth of vessel November 1893 1260 Serruys et al. American Heart Journal

RELATIVE GAIN AND RELATIVE LOSS PER REFERENCE DIAMETER GROUP (n = 1452)

RELATIVE GAIN * . RELATIVE LOSS 0.4 - 0.2

050 < 2 2.5 3.0 3.5 4.0 ’ 4.0 REF.DIAM. (mm) ADAM,, Fig. 14. Relative gain and relative loss (on the Y axes) plotted against reference diameter in increments of 0.5 mm on the X axis. Parallel trend of decreasein both variables is observedwith increasing reference diameter.

wall injury and thickness of the subsequent neoin- limitations. In addition, perhaps balloon (or device) timal hyperplasia in an experimental porcine model oversizing is more likely or frequent in small vessels. and that described by Kuntz et al. in their patients With the greater relative gain, more extensive wall undergoing DCA or stent implantation. injury is imparted, provoking a more intense healing response with formation of thicker neointimal layer THE INFLUENCE OF VESSEL SIZE ON THE that is reflected by greater angiographic relative loss RESTENOSIS PROCESS in lumen during follow-up. This may be a simplistic Exploration of the relationship of the vessel size but practical speculation on what is undoubtedly a itself on the process of luminal renarrowing reveals complex and multifactorial phenomenon, but one that the relative loss (proportional loss of lumen message is clear: the inescapable fact that greater during follow-up) decreases significantly as vessel proportional luminal gain at intervention induces size-increases in increments of 0.5 mm, as shown in greater subsequent proportional loss during follow- Fig. 14. However, when it is similarly found that rel- UP. ative gain shows a parallel pattern, it becomes The restenosis paradox. This apparent paradox of evident that it is the relationship between relative greater luminal renarrowing associated with more gain and relative loss, as already described, that is of substantial luminal improvement at balloon angio- central importance to addressing the injury-hyper- plasty has now been demonstrated in several clinical plasia phenomenon from an angiographic viewpoint. studies from our group by multiple regression anal- When the relative gain-relative loss relationship was ysis applied to large patient populations with respect investigated according to these 0.5 mm increments of to many potentially important predictors of resteno- vessel diameter, it was apparent that the relationship sis.54>lo53 117,136 We have also examined the rela- was exactly similar for all vessels (Fig. 15). Thus what tionship between relative luminal gain at inter- is described here appears to be a real phenomenon vention and relative loss during follow-up for other that is.independent of vessel size. We could speculate percutaneous coronary revascularization devices that the reason for the greater relative gain in small and preliminary results also demonstrate a direct vessels is the result of the clinical requirement for a linear relationship. lo1 In light of available evidence good angiographic result in the catheterization room. from previous clinical studies from our own institu- This demands considerable luminal gain in small tionll-l&l& 9%105 and othersl4>% 39,124,experiment& vessels given the usual angiographic magnification reports,65’ 13sand the commonly held belief that res- Volume 1’26, Number 5 American Heart Journal Serruys et al. 1261

Relative Gain Relative Gain Fig. 15. Linear regressionof relative loss on relative gain according to increments af vesselsize as shown in Fig. 14. Relationship between variables is similar for each group. It is also possibleto glean two other messagesfrom receding scatterplots with increasingvessel size: (1) much greater frequency with which bal- loon angioplasty procedures are carried out in smaller vessels,and (2) degreeof relative gain achieved at balloon angioplasty is less in larger vessels,as shown in Fig. 14. tenosis is a tissue response to vessel wall injury, the succinctly put by Schwartz et a1.,65between looking demonstration of such a relationship between rela- at the “doughnut” or the “doughnut hole.” There is tive luminal gain and loss is not all that surprising. little doubt that a larger lumen at follow-up is better The Mayo Clinic report65 (of a proportional neoint- for the patient and that this parameter, as discussed imal response to graded vessel wall injury) observing previously, is of paramount importance in assessing that the extent of coronary’ artery injury was more the long-term clinical successof therapy. However, in closely related to the actual thickness of the neoint- large clinical trials directed at prevention of the pro- imal layer than to percent luminal area stenosis ces.sof restenosis, the effect of therapy must be mea- highlights the impoitance in clinical angiographic sured by its ability to restrict or control the thickness restenosis studies of attempting to measure the vol- of the doughtnut, which we believe is best encapsu- ume of the “doughnut” and the “doughtnut hole.” lated angiographically by the relative gain/relative Collectively, all of these studies support previous ex- loss relationship. In Fig. 16 the actual relative gain/ perimenta1137 and autopsy-based138 claims that the relative loss relationships of patients undergoing intensity of neointimal proliferation after balloon therapy by four different iriterventional devices are dilatation is dependent on the depth of vessel wall shown with the line of identity and an imagin&y re- injury. Furthermore, they sustain the contention that gression line for the ideal interventional device3 is categoric restenosis definitions are inherently limited clear that a device whose regression line crosses the in their ability to describe the ubiquitous process of identity line can be considered to be associated with luminal renarrowing after interventions. aworse angiographic outcome than one with a gentle The deduction from the previously mentioned line slope. The ideal device has a horizontal relation- study by Kuntz et al. 124that by achieving a greater ship between relative gain and loss such that despite gain in lumen newer devices may reduce angiographic increasing relative gain there is no increase in relative restenosis, is apparently contradictory to our con- loss. Such a device may be considered as the magic tention. Although our group focuses mainly on the bullet or golden flee’ce. This finding is in clear relationship between the relative gain/relative loss contrast to the published findings of Kuntz et relationship by which to judge the effectiveness of a a1.,131,13g who h ave reported no difference in loss in- therapeutic intervention as a reflection of its impact dex (acute gain/late loss) between patients treated by on the injary/hyperpla@ia relationship, Kuntz et al. directional atherectomy, stent implantation, or bal- focused on the final MLD at follow-up as the ultimate loon angioplasty and concluded that the most im- outcome variable. This is the difference, as has been portant determinant of favorable long-term angio- November 1993 1262 Serruys et al. American Heart Journal

considerable scatter of data points in the regression analyses shown in Figs. 13 and 15, the phenomenon Relative Loss * DCA lr :' of wall injury and healing response is clearly multi- factorial. In addition, it must be recognized that pro- -* MESH gressively increasing relative gain will ultimately /” yield a greater proportional net angiographic benefit L * PTCA (despite provoking concomitantly greater relative luminal loss) because the relationship is always below and diverging from the line of identity. Therefore it would be fallacious to attempt to give individual guidelines as to the ideal relative gain for which to Relative Gain aim. In the final analysis, achievement of the great- est luminal improvement possible by the least trau- Fig. 16. Linear regression relationship of relative gain/ matic means and avoiding precipitation of acute relative loss of patients who underwent therapy by four complications must be the ultimate goal of percuta- different interventional devices are shown with line of identity. Imaginary regressionline for ideal interventional neous intervention. device is included. It is clear that device whose regres- The ultimate end-point. The ultimate test of new sion line crossesidentity line can be consideredto be asso- treatments is, of course, the randomized clinical trial, ciated with worse angiographic outcome than one with of which many are now in progress. The issues raised gentle line slope.Ideal device has horizontal regressionline in this article identify a vital aspect of randomized slope so that with increasing relative gain there is no increase in relative loss. DCA, Directional atherectomy trials, that is, how will the results be presented? As (n = 123 lesions);COIL, balloon-expandabletantalum coil has been mentioned, angiographic endpoints are now stent (n = 101 lesions); MESH, self-expanding stainless evaluated by quantitative analysis in terms of changes steel mesh stent (n = 110 lesions); PTCA, percutaneous in MLD from immediately after intervention to fol- transluminal coronary (balloon) angioplasty (n = 1435 le- low-up. The particular characteristic of the random- sions). ized population is that baseline demographic and angiographic characteristics in the various treatment graphic outcome is a large luminal diameter after in- groups are assumed to be similar.135 However, the al- tervention, regardless of which device is used to ready emphasized differences between devices with achieve this. It is worth noting that despite these regard to the immediate luminal improvements at- apparently conflicting viewpoints, both the Beth tainable at intervention and the subsequent luminal Israelg4 and Thoraxcentre124 groups agree that in- loss during follow-up suggest that the within patient stead of reducing intimal hyperplasia, newer devices change in minimal luminal diameter may not be the actually provoke increased hyperplasia, and that the measurement of choice to assess the comparative process of luminal renarrowing is a ubiquitous and value of the various interventions. The most objec- normally distributed phenomenon and should be tive and clinically meaningful parameter to use in described as such rather than according to arbitrary rarzclomized clinical trials is the minimal luminal di- binary criteria.g43 124 ameter at follow-up as the ultimate endpoint of Clinical implications and applications of relative gain treatment, taking all aspects of the trial into account. and relative Wss. This direct relationship between This simple approach may be equally usefully ap- restenosis, as represented by relative loss in luminal plied to trials of pharmacologic agents for the control diameter during follow-up, and luminal improve- of restenosis. Change in MLD during follow-up is ment or relative gain at intervention, has important undoubtedly the clearest angiographic measure of ramifications, not only for clinical trials but also per- the hyperplastic healing process, but it can only be haps for clinical decision making in individual pa- usefully applied if, in addition to baseline clinical and tients. With the impending widespread availability angiographic features, luminal gain at intervention is of quantitative coronary angiographic facilities for also likely to be similar in the groups being compared the catheterization room, precise measurements will thus where the same mechanical intervention is used. be readily accessible on-line, allowing a step-by-step The therapeutic effectiveness of interventional de- objective and accurate assessment of progress during vices in achieving satisfactory luminal increase with- intervention rather than the current practice of eye- out provoking excessive hyperplasia among different balling the video screen with its inherent limitations. patient groups may best be assessed through com- This should facilitate appropriate selection of bal- parison of the relative gain/relative loss relationship loon and device sizes to avoid excessive vessel wall (Fig. 16). The application of this approach to random- injury caused by oversizing. As confirmed by the ized trials will provide its ultimate test of usefulness. Volume 126, Number 5 American Heart Journal Serruys et al. 1263

Conchsions. We submit, in agreement with others, Hecht HS, Ryan C, Briskin JG, Dunlap RW, HansonCL, Zapolanski A, Cumberland DC. There is no such thing as that angiographic restenosis as a process of luminal “restenosis.” J Inv Cardiol 1992;4:282-290. renarrowing should be considered as a continuous 3. Sigwart U, Puel J, Mircowitch V, Joffre F, Kappenberger L. phenomenon and be so described in clinical trials. In Intravascular stents to prevent occlusion and restenosis after transluminal angioplasty. N Engl J Med 1987;316:701-6. addition, a proportional relationship is described be- 4. Simpson JB, Hinohara T, Selmon MR, Robertson G, White tween luminal increase at intervention and subse- N, Rorve M, Braden L. Comparison of early and recent ex- quent renarrowing during follow-up for a number of perience in percutaneous coronary atherectomy [Abstract]. J Am Co11 Cardiol 1989;13:109A. interventional devices. It is clear therefore that res- 5. Stack RS, Califf RM, Phillips HR, Pryor DB, Quigley PJ, tenosis is an unavoidable consequence of any therapy Boumann RP, Tcheng - JE, Greenfield JC Jr. Advances in that inflicts injury on the arterial wall. In contrast to cardiovascular technologies: interventional cardiac catheter- ization at Duke Medical Cent& Am J Cardiol1988,62:1-44F. the findings of other groups, clear differences are ob- 6. Ahn SS, Auth D, Marcus DR. Removal of focal atheromatous served between the devices in the relative gain/rela- lesions by angioscopically guided high speed rotary atherec- tive loss relationship, which may reflect inherent de- tomy. J Vase Surg 1988;7:292-300. 7. Spears JR, Reyes V, Sinclair N, Hopkins B, Schwartz L, Al- vice-specific characteristics of the injury/hyperplasia dridge H, Plokker HWT. Percutaneous coronary laser bsl- phenomenon. We have no erudite solutions to offer to loon angioplasty: preliminary results of a multicenter trial this persistent limitation of all interventional devices [Abstract]. J Am Co11 Cardiol 1989;13(suppl A):61A. 8. Abela GS, Seeger JM, Barbieri E, Franzini D. Laser angio- except to suggest that the search for a magic bullet plasty with angioscopic guidance in humans. J Am Co11 Car- now seems more compelling than ever. &01 i986;8:184-92. - - 9. Karsch KR, Haase KK, Mauser M, Ickrath 0, Voelker W, SUMMARY Duda S. Percutaneous coronary excimer laser angioplasty: initial clinical results. Lancet 1989;2:64’7-50. In this editorial, the problem of restenosis after 10. Hermans WRM, Rensing BJ, Strauss BH, Serruys PW. Pre- coronary balloon angioplasty and other transluminal vention of restenosis after percutaneous transluminal coro- nary angioplasty (PTCA): the search for a “magic bullet.” AM interventions is reviewed from the perspective of HEART J 1991;122:1:171-87. quantitative coronary angiography. The review is 11. Serruys PW, Strauss BH,, Beatt KJ, Bertrand ME, Puel J, largely based on the experience of theThoraxcentre Rickards AF, Meier B, Goy JJ, Vogt P, Kappenberger L, Sigwart U. Angiographic follow-up after placement of a self- in the application of quantitative angiography to the expandingcoronaryarterystent.NEngl JMed1991;324:13-7. study of restenosis over the past decade, with incor- 12. Serruys PW, Strauss BH, van Beusekom HM, van der Gies- poration and discussion of relevant and significant sen WJ. Stentins of coronarv arteries: has a modern Pando- contributions from other groups. Current discrepan- ra’s box been o&red? J Am Co11 Cardiol 1991;17:143-54B. 13. Umans VA, Beatt KJ, Rensing BJ, Hermans WR, de Feyter cies in the angiographic definition of restenosis are PJ, Serruys PW. Comparative quantitative angiographic highlighted and the use of percent diameter stenosis analysis of directional coronary atherectomy and balloon an- gioplasty: a new methodologic approach. Am J Cardiol or MLD as the measurement parameter of choice is 1991:68:X56-63. objectively addressed. Perspectives on the pathologic 14. Karsch KR, Haase KH, Voelker W, Baumbach A, Mauser M, paradigm of restenosis are briefly reviewed as a basis Seipel L. Percutaneous coronary excimer laser angioplasty in patients with stable and unstabla angina pectoris: acute re- from which to evaluate quantitative angiographic in- sults and incidence of restenosis during 6-month follow-up. formation provided by various studies. Particular at- Circulation 1990;81:1849-59. tention is then paid, in chronologic fashion, to 15. Bertrand ME, Lablanche JM, Leroy F, Bauters C, de Jaegere discussion and elaboration of insights to the resteno- P, Serruys PW, Meyer J, Dietz U, Erbel R. Percutaneous transluminal coronary rotational ablation with rotablator sis process provided by quantitative angiographic (European experience). Am J Cardiol 1992;69:470-4. studies, which have led to the introduction of some 16. De Jaegere P, Serruys PW, Bertrand M, Wiegand V, Gisbert new methodological approaches to the comparison of K, Marquis JF, Valeix B, Uebis R, Piessens J. Wiktor stent implantation in patients with restenosis following balloon short- and long-term angiographic luminal changes angioplasty of a native coronary artery lesion: immediate and after various interventions. A word of caution on the long-term clinical and angiographic results of the first fifty potential pitfalls of quantitative angiographic stud- patients. Am J Cardiol 1992;69:598-602. 17. Gruentzig AR, Myler RK, Hanna ES, Turina MI. Translu- ies is provided and counterbalanced with a discussion minal angioplasty of coronarv-arterv stenosis [Abstract]. of clinical correlations of quantitative angiographic Circulation 1977;56(suppl II):HI-84. - measurements. Finally, a proposa1 is made for the 18. Reiber JHC. Booman S. Tan HS. Slaver CJ, Schuurbiers JCH, Gerbrands JJ, Meester GT. A cardiac image analysis application of quantitative angiographic measure- system. Objective quantitative processing of angiocardio- ments to randomized clinical trials for the purpose of grams. Proc Comp Cardiol, 1978:239-42. comparing new interventional devices. 19. Reiber JHC, Booman S, Tan HS, Gerbrands JJ, Slager CJ, Schuurbiers JCH, Meester GT. Computer processing of cor- onary occlusions from x-ray arteriograms. Proceedings VI International Conference on Information Processing in Med- REFERENCES ical Imaging (Paris), 1979:79-92. 1. Forrester JS, Fishbein M, Helfant R, Fagin J. A paradigm for 20. Serruys PW, Booman F, Troost GJ, Reiber JHC, Gerbrands restenosis based on cell biology: clues for development of new JJ, van den Brand M, Cherrier F, Hugenholtz PG. Comput- preventive therapies. J Am Co11 Cardiol 1991;17:758-69. erized quantitative coronary angiography applied to percu- 2. Myler RK, Shaw RE, Stertzer SH, Ziphin RE, Rosenblum J, taneous transluminal coronary angioplasty: advantages and November 1993 1264 Serruys et al. American Heart Journal

limitations. In: Kaltenbach M, Rentrop P and Gruentzig AR coronary heart disease: comparison of cineangiographic and eds. Transluminal ‘coronary angioplasty and intracoronary necropsy findings. Ann Intern Med 1979;91:350-6. thrombolvsis. Berlin: Stringer-Verlag. 1982:110-24. 36. Nissen SE, Gurlev JC. Grines CL. Booth DC. McClure R. 21. Fleming RM, Kirkbeide RLY Smallin; RW, Gould KL. Pat- Berk M, Fischer C, DeMaria AN. Intravascular ultrasound terns in visual interpretation of coronary angiograms as de- assessment of lumen size and wall morphology in normal tected by quantitative coronary angiography. J Am Co11 Car- subjects and patients with coronary artery disease. Circula- diol 1991;18:945-51, 72. tion 1991;84:1087-99. 22. Hirshfeld JW, Schwartz SS, Jugo R, Macdonald RG, Gold- 37. Glagov S, Weisenberg E, Zarins CK, Stankunavicius K, Ko- berg S, Savage MP, Bass TA, Vetrovec G, Cowley M, Taus- lettis GJ. Compensatory enlargement of various human ath- sig AS, Withworth HB, Margolis JR, Hill JA, Pepine CJ, and erosclerotic coronary arteries. New Engl J Med 1987;316: the M-Heart Investigators. Restenosis after coronary angio- 1371-5. plasty: a multivariate statistical model to relate lesion and 38. Beatt KJ, Luijten HE, de Feyter PJ, van den Brand M, procedural variables to restenosis. J Am Co11 Cardiol 1991; Reiber JH. Serruvs PW. Change in diameter of coronarv ar- 18647-56. tery segments adjacent to &en&is after percutaneous trans.- 23. Holmes DR Jr, Vliestra RE, Smith HC, Vetrovec GW, Kent luminal coronary angioplasty: failure of the percentage of di- KM, Cowley MJ, Faxon DP, Griintzig AR, Kelsey SF, Detre ameter stenosis measurement to reflect morphologic changes KM, van Raden MJ, Mock MB. Restenosis after percutane- induced bv balloon dilation. J Am Co11 Cardiol 1988:12:315- ous transluminal coronary angioplasty: a report from the 23. ” PTCA Registry of the National Heart, Lung and Blood In- 39. Smucker ML, Kil D, Howard PF, Sarnat WS. “Whole artery stitute. Am J Cardiol 1984;53:77-8IC. restenosis” after coronary atherectomy: a quantitative an- Corocos T. David PR. Val PG. Renkin J. Dantroisse V. Rau- giographic study [Abstract]. AHA 64th scientific sessions; 24. - old HG, Bourassa MG. Failure of diltiazem to prevent rest- Circulation 1991;84:11-81. enosis after percutaneous transluminal coronary angioplasty. 40. Foley DP, Hermans WRM, Rutsch BJ, Emanuelsson H, AMHEARTJ 1985;109:926-31. Danchin N. Wiins W. Chaouuis F. Serruvs PW. for the Mer- 25. Thornton MA, Gruentzig AR, Hollman Y, King BS, Douglas cator group. Restenosis after PTCA affects the entire vessel JS. Coumadin and aspirin in the prevention of recurrence af- segment that is being dilated [Abstract]. J Am Co11 Cardiol ter transluminal coronary angioplasty: a randomized study. 1993;21:322A. Circulation 1984;69:721-7. 41. White CW, Wright CB, Doty DB, Hiratza LF, Eastham CL, 26. Leimgruber PP, Roubin GS, Hollman J, Cotsonis GA, Meier Harrison DG, Marcus ML. Does visual interpretation of the B, Douglas JS, King SB III, Gruentzig AR. Restenosis after coronary arteriogram predict the physiologic importance of successful coronary angioplasty in patients with single-vessel a coronary stenosis? N Engl J Med 1984;310:819-24. disease. Circulation 1986;73:710-7. 42. Zijlstra F, van Ommeren J, Reiber JHC, Serruys PW. Does 27. Vandormael MG, Deligonul U, Kern MJ, Harper M, Presant quantitative assessment of coronary artery dimensions pre- S, Gibson P, Galan K, Chaitman BR. Multilesion coronary dict the physiological significance of a coronary stenosis? angioplasty: clinical and angiographic follow-up. J Am Co11 Circulation 1987;75:1154-61. Cardiol 1987;10:246-52. 43. Serruys PW, Wijns W, Geuskens R, de Feyter P, van den 28. Reis GJ, Boucher TM, Sipperly ME, Silverman DJ, MeGabe Brand M, Reiber JHC. Pressure gradient, exercise thallium CH, Baim DS, Sachs FM, Grossman W, Pasternak RC. Ran- 201 scintigraphy, quantitative coronary cineangiography: in domised trial of fish oil for prevention of restenosis after cor- what sense are these measurements related? In: Reiber JHC, onary angioplasty. Lancet 1989;2:177-81. Serruys PW, eds. State of the art in quantitative coronary 29. Meyer J, Schmitz HJ, Kiesslich T, Erbel R, Krebs W, Shultz arteriography. Lancaster: Martinus Nijhoff Publishers, W, Bardos P, Minale C, Messmer BJ, Effert S. Percutaneous 1986251-70. transluminal coronary angioplasty in patients with stable 44. Wijns W, Serruys PW, Reiber JHC, van den Brand M, Si- and unstable angina pectoris: analysis of early and late moons ML, Kooijman CJ, Balakumaran K, Hugenholtz PG. results. AM HEARTJ 1983;106:973-80. Quantitative angiography of the left anterior descending 30. Fleck E, Dacian S, Dirschinger J, Hall D, Rudolph W. Quan- coronary artery: correlation with pressure gradient and titative changes in stenotic coronary artery lesions during results of exercise thallium scintigraphy. Circulation 1985; follow-up after PTCA [Abstract]. Circulation 1984;7O(suppl 71:273-g. II):Il-176. 45. Reiber JHC, Serruys PW. Quantitative coronary angiogra- 31. Serruys PW, Luijten HE, Beatt KJ, Geuskens R, de Feyter phy. In: Marcus ML, Schelbert HR, Skorton DJ, Wolf GI, PJ. van den Brand M. Reiber JHC. Ten Katen F. van Es GA, eds. Cardiac imaging, a companion to Braunwalds heart dis- Hugenholtz PG. Incidence of rest&o& after successful an- ease. New York:WB Saunders, 1991:211-80. -gioplasty: _ a time related phenomenon. A quantitative angio- 46. Beatt KJ, Serruys PW, Hugenholtz PG. Restenosis after graphic-study in 342 consecutive patients at 1, 2, 3 and 4 coronary angioplasty: new standards for clinical studies. J months. Circulation 1988:77:361-71. Am Co11 Cardiol 1990;15:491-9. 32. Nobuyoshi M, Kimura T,‘Nosaka H, Mioka S, Ueno K, Ha- 47. Rensinz BJ. Hermans WRM. Beatt KJ. Laarman GJ. Surv- masaki N, Horiuchi H, Oshishi H. Restenosis after success- apranaia Hi van den Brand M, de Feyter PJ, Serruys PW. ful percutaneous transluminal coronary angioplasty: serial Quantitative angiographic assessment of elastic recoil after aneioarauhic follow-up of 299 patients. J Am Co11 Cardiol percutaneous transluminal coronary angioplasty. J Am Co11 1988;i2:616-23. Cardiol 1991;17:34-8B. 33. Bourassa MG. Lesnerance J. Eastwood C, Schwartz L, Gote 48. Rensing BJ, Hermans WR, Strauss BH, Serruys PW. Re- G, Kazim F, Huddn G. Clinical, physiologic, anatomic, and gional differences in elastic recoil after percutaneous trans- procedural factors predictive of restenosis after percutaneous luminal coronary angioplasty: a quantitative angiographic transluminal coronary angioplasty. J Am Co11 Cardiol 1991; study. J Am Co11 Cardiol 1991;17:34-8B. l&368-76. 49. Hanet C, Wijns W, Michel X, Schroeder E. Influence of bal- 34. Gould KL, Lipscomb K, Hamilton GW. Physiological basis loon size and stenosis morphology on immediate and delayed for assessing critical coronary stenosis: instanteous flow elastic recoil after percutaneous transluminal coronary angi- response and regional distribution during coronary hyper- oplasty. J Am Co11 Cardiol 1991;18:506-11. aemia as measures of coronary flow reserve. Am J Cardiol 50. Hermans WR. Rensina BJ. Strauss BH, Serruvs FW. Meth- 1974;33:87-97. odological problems rerated to the quantitativeassessment of 35. Arnett EN, Isner JM, Redwood DR, Kent KM, Baker WP, stretch, elastic recoil, and balloon-artery ratio. Cathet Car- Ackerstein H, Roberts WC. Coronary artery narrowing in diovasc Diagn 1992;25:174-85. Volume 126, Number 5 American Heart Journal Serruys et al. 1265

51. Isner JM, Rosenfield K, Losordo DW, Rose L, Langevin RE proliferation following coronary angioplasty in a swine rest- Jr. Razvi S. Kosowskv BD. Combination balloon-ultrasound enosis model [Abstract]. Clin Res 1992;40:364A. imaging catheter for percutaneous transluminal angioplasty. 70. Gravanis MB, Roubin GS. Histopathologic phenomenona at Validation of imaging, analysis of recoil, and identification of the site of percutaneous transluminal coronary angioplasty: plaque fracture. Circulation 1991;84:739-54. the problem of restenosis. Hum Path01 1989:20:477-85. 52. Hanet C, Michel X, Schroeder E, Wijns W. Lack of detect- 71. denHeijer P, van Dijk RB, Pentinga ML, Lie KI. Serial an- able delayed elastic recoil during the first 24 hours after cor- gioscopy during the first hour after successful PTCA [Ab- onary balloon angioplasty [Abstract]. Eur Heart J 1992;13 stract]. Circulation 1992;86:1*458. (suppl):2228. 72. Uchida Y, Tomaru T, Fujimora Y, Miwa A, Nakamura F, 53. Foley DP, Rensing BJ, Vos J, Deckers J, Serruys PW. Is there Yamada K. Observation of luminal changes in human core: a need for angiography after 24 hours to assess the result of nary artery by an intravascular microscope [Abstract]. Cir- coronary balloon angioplasty in clinical studies [Abstract]. culation 199&8&I-503. Circulation 1992;86:1-785. 73. Beatt KJ, Serruys PW, Luijten HE, Rensing BJ, Suryapra- 54. Rensing BJ, Hermans WR, Vos J, Beatt KJ, Bossuyt P, nata H, de Feyter P, van den Brand M, Laarman GJ, Rutsch W, Serruys PW, on behalf of the Carport study group. Roelandt J. Restenosis after coronary angioplasty: the para- Quantitative angiographic risk factors of luminal narrowing dox of increased lumen diameter and restenosis. J Am Co11 after coronary balloon angioplasty using balloon measure- Cardiol 1992;19:258-66. ments to reflect stretch and elastic recoil at the dilatation site. 74. Kent KM, Bonow RO, Rosing DR, Ewels CJ, Lipson LC, Am J Cardiol 1992;69:584-91. McIntosh CL. Bacharach S. Green M. Eastein SE. Imnroved I / 55. Steele PM, Chesebro JH, Stanson AW, Holmes DR Jr, De- myocardial function during exercise after successful percu- wanjee MK, Badimon L, Fuster V. Balloon angioplasty: nat- taneous transluminal coronary angioplasty. N Engl J Med ural history of the pathophysiological response to injury in 1982;306:441-6. the pig model. Circ Roman Res 1985;57:105-12. 75. Califf RM, Ohman EM, Frid DJ, Fortin DF, Mark DB, 56. Uchida Y, Hasegawa K, Kawamura K, Shibuya I. Angio- Hlatky MA, Herndon JE, Bengtson JR. Restenosis: the clin- scopic observation of the coronary luminal changes induced ical issues. In: Top01 EJ, ed. Textbook of interventional car- by coronary angioplasty. AM HEART J 1989;117:769-76. diology. Philadelphia: WB Saunders, 1990:363-94. 57. Chesebro JH. Lam JYT. Badimon L. Fuster V. Restenosis 76. Popma JJ, van den Berg EK, Dehmer GJ. Long-term after arterial angioplasty: a hemorrheologic response to outcome of patients with asymptomatic restenosis after per- injury. Am J Cardiol 1987;60:10-6B. cutaneous transluminal angioplasty. Am J Cardiol 1989; 58. Clowes AW, Reidy MA, Clowes MM. Mechanisms of steno- 62:1298-g. sis after arterial iniurv. Lab Invest 1983:49:208-15. 17. Jain A, Mahmarian JJ, Borges-Neto S, Johnston DL, Cash- 59. Essed CE, van denBrand M, Becker AE: Transluminal cor- ion R, Lewis JM, Raizner AE, Verani MS. Clinical signifl- onary angioplasty and early restenosis: fibrocellular occlu- cance of perfusion defects by thallium-201 single photon sion after wall laceration. Br Heart J 1983;49:393-6. emission tomography following oral dipyridamole-early after 60. Austin GE, Ratliff NB, Hollman J, Tabei S, Phillips DF. In- coronary angioplastv.- _ J Am Co11 Cardiol 1988:11:970-6. timal proliferation of smooth muscle cells as an explanation 78. Vlay SC, Chernilas”J, Lawson WE, Dervan J’P. Restenosis for recurrent coronary artery stenosis after percutaneous after angioplasty: don’t rely on the exercise test. AM HEART transluminal coronary angioplasty. J Am Co11 Cardiol J 1989;4:980-6. 1985;6:369-75. 79. Laarman GJ, Luijten HE, van Zeyl LG, Beatt KJ, Tijssen 61. Ip JH, Fuster V, Israel D, Badimon L, Chesebro JH. The role JGP, Serruys PW. Assessment of “silent” restenosis and of platelets, thrombin and hyperplasia in restenosis after a long-term follow-up after successful angioplasty in single- coronary angioplasty procedure. J Am Co11 Cardiol1991;17: vessel coronary artery disease: the value of quantitative ex- 77-88B. ercise electrocardiography and quantitative coronary angiog- 62. Waller BF, Pinkerton CA, Orr SM, Slack JD, Van Tassel JW, raphy. J Am Co11 Cardiol 1990;16:578-85. Pinto RP. Restenosis 1 to 24 months after clinically success- 80. Klocke FJ. Measurements of coronary blood flow and degree ful coronary balloon angioplasty: a necropsy study of 20 pa- of stenosis: current clinical implications and continuing un- tient. J Am Coil Cardiol 1991;17:58-7OB. certainties. J Am Co11 C&did 1983;1:31-41. 63. Schwartz RS, Holmes DR, Top01 EJ. The restenosis para- 81. Serruys PW, Zijlstra F, Laarman GJ, Reiber JHC, Beatt K, digm revisited: an alternative proposal for cellular mecha- Roelandt J. A comparison of two methods to measure coro- nisms. J Am Co11 Cardiol 1992;20:1284-93. nary flow reserve in the setting of coronary angioplasty: in- 64. Schwartz RS, Murphy JG, Edwards WD, Camrud AR, tracoronary bloodflow velocity measurements with a Doppler Vliestra RE. Holmes DR. Restenosis after balloon anaio- catheter and digital subtraction cineangiography. Eur Heart plasty. A practical proliferative model in porcine coronary J 1989;10:725. arteries. Circulation 1991;82:2190-200. 82. Zijlstra F, den Boer A, Reiber JHC, Van Es GA, Lubsen J, 65. Schwartz RS, Huber KC, Murphy JG, Edwards WD, Camrud Serruys PW. The assessment of immediate and long-term AR, Vlietstra RE, Holmes DR. Restenosis and the propor- functional results of percutaneous transluminal coronary tional neointimal response to coronary artery injury: results angioplasty. Circulation 1988;1:15-24. in a porcine model. J Am Co11 Cardidl 1992$9:267-74. 83. Hodgson JM, Riley RS, Most AS, Williams DO. Assessment 66. Feial W. Susani M. Ulrich W. Mateika M. Losert U. Sinzinaer of coronary flow reserve using digital angiography before and H. &gamsation of experimental thrombosus by blood ceils. after successful percutaneous transluminal coronary angio- Evidence of the transformation of mononuclear cells into plasty. Am J Cardiol 1987;60:61-5. myofibroblasts and endothelial cells. Virchows Arch A Pathoi 84. Kirkeeide RL, Gould KL, Parse1 L. Assessment of coronary Anat Histopathol 1985;406:133-48. stenoses by myocardial perfusion imaging during pharmaco- 67. Leu HJ, Feigl W, Susani M. Angiogenesis from mononuclear logic coronary vasodilation. Validation of coronary flow cells in thrombi. Virchows Arch A Path01 Anat Histopathol reserve as a single integrated functional measure of stenosis 1987;411:5-14. severity reflecting all its geometric dimensions. J Am Co11 68. Karas SP, Gravanis MB, Santoian EC, Robinson KA, Ander- Cardiol 1986;7:103-13. berg KA, King SB III. Coronary intimal proliferation after 85. Honye J, Mahon DJ, Jain A, White CJ, Ramsee SR, WaIlis balloon injury and stenting in swine: an animal model of res- JB, Al-Zarka A, Tobis JM. Morphological effects of coronary tenosis. J Am Co11 Cardiol 1992;20:467-74. balloon angioplasty in vivo assessed by intravascular uitra- 69. Santonian EC, Gravanis MB, Karas SP. Sequence of the re- sound imaging. Circulation 1992;85:1012-25. parative phenomena and development of smooth muscle cell 86. Serruys PW, Wijns W, van den Brand M, ??????. Is translu-

I November 1993 1266 Serruys et al. American Heart Journal

minal coronary angioplasty mandatory after successful 101. Foley DP, Hermans WR, de Jaegere PP, Umans VA, Vas J, thrombolysis? ‘A quantitative angiographic study. Br Heart Escaned J, de Feyter PJ, Serruys PW. Is “bigger” really J 1983;50:257. “better”? A quantitative angiographic study of immediate 87. Serruys PW, Reiber JHC, Wijns W, van den Brand M, and long term outcome following balloon angioplasty, direc- Kooijman CJ, ten Katen HJ, Hugenholtz PG. Assessment of tional atherectomy and stent implantation [Abstract]. Cir- percutaneous transluminal coronary angioplasty by quanti- culation 1992;86:1-530. tative coronary angiography: diameter versus densitometric 102. Foley DP, Hermans WR, Umans VA, de Jaegere PP, Serruys area measurements. Am J Cardiol 1984;54:482. PW. The influence of vessel size on restenosis following per- 88. Reiber JHC, Serruys PW, Kooijman CJ, Wijns W, Slager CJ, cutaneous coronary interventions [Abstract]. Circulation Gerbrands JJ, Schuurbiers JCH, den Boer A, Hugenholtz 1992;86:1-255. PG. Assessment of short-, medium-, and long-term variations 103. Umans VAWM, Strauss BH, Rensing BJWM, de Jaegere P, in arterial dimensions from computer-assisted candidate of de Feyter P, Serruys PW. Comparative angiographic quan- coronary cineangiograms. Circulation 1985;71:280-8. titative analysis of the immediate efficacy of coronary atherec- 89. Serruys PW, Juilliere Y, Bertrand ME, Puel J, Rickards A, tomy with balloon angioplasty, stenting and rotational abla- Sigwart U. Additional improvement of stenosis geometry in tion. AM HEART J 1991;122:836-43. human coronary arteries by stenting after balloon dilation: a 104. de Jaegere P, Strauss BH, de Feyter P, Suryapranata H, van quantitative angiographic study. Am J Cardiol 1988;61:71- den Brand M, Serruvs PW. Stent versus balloon anaionlasty: 6G. matching based on QCA, a surrogate for randomized studies. 90. Strauss BH, Juilliere Y, Rensing BJ, Reiber JHC, Serruys AM HEARTJ 1993;125:310-8. PW. Edge detection versus densitometry for assessing coro- 105. de Jaegere P, Bertrand M, Wiegand V, Kober G, Marquis JF, nary stenting quantitatively. Am J Cardiol 1991;67:484-90. Valeix B, Uebis R, Piessens J, de Feyter P, Serruys PW. An- 91. Umans VA, Strauss BH, de Feyter PJ, Serruys PW. Edge giographic predictors of restenosis following Wiktor stent detection versus videodensitometry for quantitative angio- implantation [Abstract]. J Am Co11 Cardiol 1992;19(Suppl): graphic assessment of directional coronary atherectomy. Am 277A. J Cardiol 1991;68:534-9. 106. Umans VA, Hermans WR, Foley DP, de Feyter PJ, Serruys 92. de Feyter PJ, Serruys PW, Davies MJ, Richardson P, Lub- PW. Restenosis after directional coronary atherectomy and sen J, Oliver MF. Quantitative coronary angiography to balloon angioplasty: comparative analysis based on matched measure progression and regression of coronary atheroscle- lesions: J Am Co11 Cardiol 1993;21:1382-90. rosis. Values, limitations and implications for clinical trials. 107. Umans VA, Hermans WR, Rensing BJ, de Feyter PJ, Serruys Circulation 1991;84:412-23. PW. Directional coronary atherectomy versus balloon angi- 93. Strauss BH, Serruys PW, de Scheerder IK, Tijssen JGP, oplasty: a matched comparative quantitative angiographic Bertrand ME. Puel J. Meier 3. Kaufmann U. Strauffer JC. analysis [Abstract]. J Am Co11 Cardiol 1992;19:276A. Rickards AF, Sigwart U. A relative risk analysis of the angio- 108. Hermans WRM, Rensing BJ, Foley DP, Deckers JW, Rutsch graphic predictors of restenosis within the coronary Wall- W, Emanuelsson H. Danchin N. Wiins W. Channuis. Serruys stent. Circulation 1991;84:1636-43. PW. on behalf of the Mercatdr Study group.*Therapeutic 94. Rensing BJ, Hermans WR, Deckers JW, de Feyter PJ, Tijs- dissection after successful coronary angioplasty: no effect on sen JGP, Serruys PW. Luminal narrowing after percutaneous restenosis nor on clinical outcome nor in 693 patients. J Am transluminal coronary balloon angioplasty follows a near Co11 Cardiol 1992;20:767-80. gaussian distribution. A quantitative angiographic study in 109. Gould KL. Identifying and measuring severity of coronary 1445 successfully dilated lesions. J Am Co11 Cardiol 1992; artery stenosis. Quantitative coronary arteriography and 19:939-45. positron emission tomography. Circulation 1988;78:237-45. 95. di Mario C, Hermans WRM, Rensing BJ, Serruys PW. Cal- 110. Savage M, Fischman D, Leon M, Ellis S, Goldberg S. Rest- ibration using angiographic catheters as scaling devices - im- enosis risk of single Palmaz-Schatz stents in native coronar- portance of filming the catheters not filled with contrast me- ies: report from the core angiographic laboratory [Abstract]. dium [Letter]. Am J Cardiol 1992;69:1377. J Am Co11 Cardiol 1992:(sunnl):19:277A. 96. di Mario C, Haase J, den Boer A, Reiber JHC, Serruys PW. 111. Serruys PW, Rutsch W; ‘Heyndrickx GR, Danchin N, Mast Edge detection versus videodensitometry in the quantitative EG, Wijns W, Rensing BJ, Vos J, Stibbe J. Prevention of assessment of stenosis phantoms: an in-vivo comparison in restenosis after percutaneous transluminal coronary angio- porcine coronary arteries. AM HEART J 1992;124:iI81-9. plasty with thromboxane As receptor blockade. A random- 97. Haase J. de Mario C. Slaeer CJ. van der Giessen WJ. den Boer ized, double-blind, placebo controlled trial. Circulation 1991; A, de Feyter PJ, V&do; PD,‘Reiber JHC, Serruys PW. In- 84X68-81. vivo validation of on-line and off-line geometric coronary 112. The Mercator Study Group. Does the new angiotensin-con- measurements using insertion of stenosis phantoms in por- verting enzyme inhibitor cilazapril prevent restenosis after cine coronary arteries. Cathet Cardiovasc Diagn 1992;27:16- percutaneous transluminal coronary angioplasty? The re- 27. sults of the Mercator study: a multicenter randomized dou- 98. Serruys PW, de Jaegere P, Bertrand M, Kober G, Marquis ble-blind placebo-controlled trial. Circulation 1992;86:100- 11. JF. Piessens J. Uebis R. Valeix B. Wieeand V. Mornholoni-I cal change of coronary artery stenosis-with the Medtrogic 113. Top01 EJ, Leya F, Pinkerton CA, Whitlow PL, Hofling B, Si- Wiktor stent. Initial results from the core laboratory for monton CA, Masden RR, Serruys PW, Leon MB, Williams quantitative angiography. Cathet Cardiovasc Diagn 1991; DO, King SB III, Mark DB, Isner JM, Holmes Dr Jr, Ellis SG, 24:237-45. Lee KL, Keeler GP, Berdan LG, Hinohara T, Califf RM for 99. Hermans WRM, Rensing BJ, Paameyer J, Serruys PW. Ex- the Caveat Study group. A comparison of directional atherec- periences of a quantitative coronary angiographic core labo- tomy with coronary angioplasty in patients with coronary ratory in restenosis prevention trials. In: Reiber JHC, Ser- artery disease. N Eng J Med 1993;329:221-7. rugs PW, eds. Advances in quantitative coronary arteriogra- 114. Cohen EA, Adelman AG Kimball BP, Bonan R, Ricci DR, phy. Dordrecht: Kluwer Academic Publishers, i993:117195. Webb JG, Laramee L, Barbeau G, Traboulsi M, Corbett BN, 00. Hermans WR. Rensins BJ. Kelder JC. de Fevter PJ. Serruvs Schwartz L, Logan AG. A comparison of directional atherec- PW. Postangibplasty restenosis rate betweensegments of the tomy with balloon angioplasty for lesions of the left anterior major coronary arteries. Am J Cardiol 1992;69:194-200. descending coronary artery. N Engl J Med 1993;329:228-33. Volume 126, Number 5 American Heart Journal Serruys et al. 1267

115. Popma JJ, Califf RM, Top01 EJ. Clinical trials of restenosis term results of directional coronary atherectomy: predictors following angioplasty. Circulation 1991;84:1426-37. of restenosis. J Am Co11 Cardiol 1992;20:1101-10. 116. Ellis SG, Muller DWM. Arterial injury and the enigma of 129. Carroza JP Jr.. Kuntz RE. Levine MJ. Pomerantz RM. Fish- restenosis. J Am Co11 Cardiol 1992;19:275-7. man RF, Mansour M, Gibson CM Senerchia CC, Diver DJ, 117. SerruysPW, Hermans WRM,RensingBJ, Thijssen J, Rutsch Safian RD, Bairn DS. Angiographic and clinical outcome of W. Emanuelsson H. Danchin N. Wiins W. Channuis F on be- intracoronary stenting: immediate and long-term results half of the Mercator study group. Are clinical; angiographic from a single-center experience. J Am Co11 Cardiol 1992;20: or procedural variables predictive of luminal renarrowing af- 328-37. ter successful coronary balloon angioplasty [Abstract]? Cir- 130. Kuntz RE, Hinohara T, Robertson GC, Safian RD, Simpson culation 1992;86:1-849. JB, Bairn DS. Influence of vessel selection, on the observed 118. Serruys PW, Rensing BJ, Luijten HE, Hermans WR, Beatt restenosis rate after endoluminal stenting or directional KJ. Restenosis following coronary angioplasty. In: Meier B, atherectomy. Am J Cardiol 1992;70:1101-8. ed. Interventional cardiology. Stuttgart: Hogrefe and Huber, 131. Kuntz RE, Hinohara T, Safian RD, Selmon MR, Simpson JB, 1990;79-115. Bairn DS. Restenosis after directional coronary atherectomy 119. Beatt KJ, Luijten HE, Suryapranata H, de Feyter PJ, Ser- effects of luminal diameter and deep wall excision. Circula- ruys PW. Dilatation variables: the paradox of optimal im- tion 1992;86:1394-9. provement in stenosis by percutaneous transluminal coro- 132. Forrester JS, Eigler N, Litvack F. Interventional cardiology: nary angioplasty and restenosis [Abstract]. Br Heart J the decade ahead. Circulation 1991;84:942-4. 1989;61:439. 133. Tcheng JE, Frid DJ, Fortin DF, Nelson CL, Stack RK, Peter 120. Beatt KJ, Luijten HE, Suryapranata H, de Feyter PJ, Ser- RH, Stack RS, Califf RM. Anatomic propensity for resteno- ruys PW. Dilatation parameters: the paradox of optimal lu- sis following coronary angioplasty [Abstract]. J Am Co11 minal improvement in stenosis by PTCA and restenosis [Ab- Cardiol 1991;17(suppl):345A. stract]. Eur Heart J 1989;(supp1);10:A28. 134. Reifart N, Preusler W, Storger H, Schwarz F, Hofmann M, 121. Beatt KJ. Luiiten HE. Survaaranata H. de Fevter PJ. Ser- Klopper JW, Vandormael M. Comparison of excimer laser, ruys PW.‘Suboptimal post-*an‘gioplasty iesult. The principle rotablator and balloon angioplasty for the treatment of com- risk factor for “restenosis” [Abstract]. Circulation 1989(sup- plex coronary lesions: a randomized trial [Abstract]. Circula- p1)80;1024. tion 1992;86:1-374. 122. Kuntz RE, Schmidt DA, Levine MJ, Reis GJ, Safian RD, 135. Serruys PW, de Jaegere P, Kiemeneij F, Macaya C, Eman- Bairn DS. Importance of post-procedure Iuminal diameter on uelsson H, Heyndrickx G, de Bruyne B, Rutsch W, Legrand restenosis following new coronary interventions [Abstract]. V, Matterne P, Bolland J, Marco J, Fajadet J, van den Heu- Circulation 1990;82:111-314. Se1 P, van den Brande F, Bonnier H, Bracke F, Colombo A, 123. Kuntz RE, Safian RD, Schmidt DA, Levine MJ, Reis GJ, Morel MA for the Benestent study group. Clinical events and Bairn DS. Restenosis following new coronary devices: the in- angiographic results of the first 120 patients randomized in fluence of post-procedure luminal diameter [Abstract]. J Am the Benestent Study [Abstractl. Circulation 1992:86:1-373. Co11 Cardiol 1991;17:2A. 136. Rensing BJ, Herma& WRM, Vds J, Thijssen JGP, Rutsch W, 124. Kuntz RE, Safian RD, Levine MJ, Reis GJ, Diver DJ, Bairn Danchin N, Heyndrickx GR, Mast EG, Wijns W, Serruys DS. Novel approaches to the analysis of restenosis after the PW, on behalf of the Coronary Artery Restenosis Prevention use of three new coronary devices. J Am Co11 Cardiol 1992; on Repeated Thromboxane Antagonism (CARPORT) study 19:1493-g. group. Luminal narrowing after percutaneous transluminal 125. King SB III, Weintraub WS, Tao X, Hearn J, Douglas JS Jr. coronary angioplasty. A study of clinical, procedural and le- Bimodal distribution of diameter stenosis 4 to 12 months af- sional factors related to long term angiographic outcome. ter angioplasty: implications for definitions and interpreta- Circulation 1993 (in press). tion of restenosis [Abstractl. J Am Co11 Cardiol 1991;17 137. Barbeau GR, Fried1 SE, Santon JM, Federman M, Abela GS. (suppl):345A. Rupture of internal elastic lamina is essential for restenosis 126. Danchin N, Marie PY, Julliere Y, Karcher G, Bertrand A, following balloon angioplasty [Abstract]. Circulation 1991; Aliot E. Serruvs ” PW. , Cherrier F. What is the minimum lu- 84~11-603. minal diameter necessary to avoid exercise-induced myocar- 138. Nobuyoshi M, Kimura T, Ohishi H, Horiuchi H, Nosaka H, dial ischemia in post-PTCA patients? Long-term quantita- Hamasaki N, Yokoi H, Kim K. Restenosis after percutane- tive angiographic study. Eur Heart J 1991;12(suppl):A-889. ous trausluminal coronary angioplasty: pathologic observa- 127. Rensing BJ, Hermans WRM, Deckers JP, de Feyter PJ, Ser- tions in 20 patients. J Am Co11 Cardiol 1991;17:433-9. ruys PW. Which angiographic parameter best describes 139. Kuntz RE, Gibson CM, Nobuyoshi M, Bairn DS. Generalized functional status 6 months after successful single vessel cor- model of restenosis after conventional balloon angioplasty, onarv balloon aneioalastv. J Am Co11 Cardiol1993:21:317-24. stenting and directional atherectomy. J Am Co11 Cardiol 128. Fischman RF, Kuntz RI?, Carroza JP Jr, Miller MJ, Sener- 1993;21:15-25. chia CC, Schnitt SJ, Diver DJ, Safkn RD, Bairn DS. Long-